Method and apparatus for test and characterization of semiconductor components

Abstract

A method and apparatus for testing and characterizing circuits is provided. In one embodiment, a high-speed interface of a semiconductor component includes high-speed test circuitry. The high-speed test circuitry obviates the need for an external high-speed testing system for testing and characterization. In one embodiment, the high-speed test circuitry includes a test pattern generation circuit, and various differential comparators to compare low bandwidth reference signals with interface signals during testing and characterization. In one embodiment, an interface that includes the test circuitry can test itself or another interface. In one embodiment, a timing reference signal decouples the individual parameters of two interfaces testing each other to avoid any errors introduced by the combination of individual interface circuit parameters, such as receiver parameters and transmitter parameters. The testing can be performed at the wafer stage, at the component stage, and in a system.

Description

FIELD OF THE INVENTION

[0002] The present invention relates to testing and characterization of electronic circuits implemented in semiconductor components.

BACKGROUND OF INVENTION

[0003] Modern semiconductor circuit components are manufactured by first placing electronic circuit components on semiconductor wafers and then packaging different circuits into circuit component packages. Packages include semiconductor “chips” that are placed in systems with many other chips to create a product such as a personal computer or a network router. At various stages in the manufacturing process, it is important to verify that the components function properly and meet specifications. Testing a component includes, for example, sending a pattern of bits to a component and verifying that the component receives and interprets the pattern correctly. Testing can also include component characterization, which establishes or verifies a set of operating parameters for the component. For example, for a component that performs a binary signaling protocol, it is necessary to guarantee that the component will interpret voltages in certain ranges as ones or zeroes. Typically, components are tested with a specialized piece of equipment called a test system or tester.

[0004] Recent improvements in the ability of components to produce a greater range of signaling bandwidths and to operate at higher frequencies have increased the cost and difficulty of testing and characterizing components. In order to test and characterize components, a test system must operate at a frequency at least as high as the operating frequency of the component under test. The test system requires some tester guard band, which is a margin around the tested values required to guarantee the tested values. In the past, it was relatively easy to produce test systems that were faster than the tested components. That is not the case now, as even mass-produced metal oxide semiconductor (“MOS”) components achieve ever greater speeds. High-speed test systems that are capable of exceeding the speeds of the fastest components today are very expensive. As component designers produce faster and faster component designs, it becomes more problematic for potential manufacturers to produce the designs because of the required investment in faster test systems.

[0005] FIG. 1 is a block diagram of a prior art high-speed interface 100 of a component A. The high-speed interface 100 is coupled to an interconnect 108, which is an electrical conductor or conductors. The high-speed interface 100 includes a receiver block 104, a transmitter block 102, and a termination block 106. The receiver 104 includes a voltage discriminator element 132, a register element 130, and a receive clock aligner element 128. The voltage discriminator element 132 is a differential comparator that compares the VP (high logic value) and VN (low logic value) voltages values on interconnect 108 to determine what symbol is present (e.g., a “1” or a “0”). The register element 130 is a time discriminator that samples the output of the voltage discriminator element 132 on the rising edge of the CLKRA signal (the receive clock signal of component A), stores it, and drives it as the RA signal (receive signal of component A), to be used by other circuits on the component A. The receive clock aligner element 128 creates the CLKRA signal from the CLKRREF (receive clock reference) reference signal. The receive clock aligner element 128 is typically a phase locked loop (“PLL”) or a delay locked loop (“DLL”) circuit.

[0006] The termination block 106 includes two load devices 134a and 134b, coupled to interconnect 108 with one terminal and to the VT supply with the other terminal. A terminal of load device 134a is coupled to a first signal line, in interconnect 108, that provides a VP voltage value and a terminal of load device 134b is coupled to a second signal line, in interconnect 108, that provides a VN voltage value. The termination block 106 further includes a control element 136, which adjusts the resistive value RT of the load devices 134 to match an external reference resistance, RREF.

[0007] The transmitter block 102 includes a predriver element 118, two differential driver elements 120 and 122, two current source elements 124 and 126, a current control element 114, a register element 116, and a transmit clock aligner element 112. The register element 116 is a time drive element that samples the TA signal (transmit signal from other circuits on component A) and drives it on the rising edge of the CLKTA signal. The transmit clock aligner element 112 creates the CLKTA signal from the CLKTREF reference signal. The transmit clock aligner element 112 is typically a phase locked loop (“PLL”) or a delay locked loop (“DLL”) circuit. The output of the register element 116 connects to the predriver element 118, which connects to the differential driver elements 120 and 122. Each driver element 120 and 122 is an open drain transistor, and is connected in series with one of the current source elements 124 and 126, respectively. The current control element 114 maintains a sink current of IOL (output low current) in the current source elements using an external reference value IREF. When the transmitter 102 drives a bit, one of the two driver elements 120 and 122 is on, and the other is off.

[0008] In an alternate embodiment, the two current source elements 124 and 126 could be merged into a single current source element, with one terminal connecting to the low supply voltage (ground) and the other terminal connecting to the source terminals of the two driver elements 120 and 122. This will be equivalent to the circuit shown in transmitter block 102 since only one of the two driver elements 120 and 122 may be on at any time.

[0009] FIG. 2 shows a prior art test system 202 used to characterize and test the transmitter 102 and the receiver 104 of component A. Typically, component A will be mounted on a load board 222, which is a printed circuit board. The testing system 202 and the load board 222 approximately duplicate the environment seen by component A during normal operation. The load board 222 has an interconnect 212 that couples the interface 100 of the component A to the high-speed pin electronics 204 of test system 202. The interconnect 212 has a characteristic impedance ZO which may be the same as, or different from, that of the manufactured system interconnect 108. The interconnect 108 is terminated on one end by termination 214 in high-speed pin electronics 204, and on the other end by termination 106. The high-speed pin electronics 204 include timing circuitry 206 to adjust the drive point and the sample point of signals. The high-speed pin electronics 204 further include voltage/current force/sense circuitry 208 to force and sense voltage values and to force and sense current values. The high-speed pin electronics 204 also include pattern generation circuitry 210 for storing, generating, and comparing test patterns. Typically, the accuracy and resolution of circuitry 206, circuitry 208 and circuitry 210 are very high in order to minimize the uncertainty due to measurement error. As a result, the cost of pin electronic 204 is also very high. This is particularly true in the case in which the component A has high-speed signals, each requiring its own set of pin electronics. The test system 202 must simultaneously provide a high signaling rate, precise timing control, and precise voltage/current control for each signal to be tested.

[0010] The test system 202 further includes circuitry that generates reference signals 110 used by component A. The reference signals 110 include IREF, RREF, CLKRREF, and CLKTREF. These reference signals are typically shared across the entire component A. These reference signals 110 often do not need the simultaneous combination of high-speed, voltage/current accuracy and timing accuracy. Therefore, the circuitry necessary to produce the reference signals is relatively easier and cheaper to produce compared to the high-speed pin electronics 204. For example, the RREF signal will typically shift through a small set of discrete values during testing. The IREF signal will typically shift through some range of direct current (“DC”) values during testing. The CLKRREF and CLKTREF signals provide a frequency and phase reference for the component A. The clock frequency will typically shift through a small set of discrete values during testing. The clock phase will typically shift through some range of values during testing. Typically, a component will utilize frequency multiplication, so that the required signaling rate of the CLKRREF and CLKTREF signals is much lower than the signaling rate of the high-speed signals.

[0011] Another disadvantage of prior art test systems, such as the high-speed test system 202, is that they cannot be used at every stage of the manufacturing process. The result is more defective components being passed to later stages of the manufacturing process. In a typical manufacturing process, there are at least three possible testing stages: 1) component wafer testing; 2) component package testing; and 3) in-system testing. Component wafer testing determines which components are acceptable to be packaged. Component package testing determines which components are acceptable to be used in a system. In-system testing determines which systems work properly. Usually, components are tested twice, once at the wafer level before packaging, and again after packaging. High-speed testing, however, is only performed after packaging when packages can be placed on a load board such as load board 222. All high-speed testing must be performed with a high-speed test system, such as high-speed test system 202. Typically a high-speed test system performs the most exhaustive testing during the component package test. It is usually not possible to perform component wafer testing at full speed. The final in-system test can be performed at full speed, but must use nominal parameter values because of the difficulty of probing high-speed interconnects within an operating system.

[0012] There is a need for a method and apparatus for testing and characterizing high-speed components that does not require an expensive high-speed test system. There is also a need for a method and apparatus for providing more uniform testing at each stage of the manufacturing process.

BRIEF DESCRIPTION OF THE DRAWING

[0013] The present invention is illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings in which:

[0014] FIG. 1 is a block diagram of a prior art high-speed interface;

[0015] FIG. 2 is a block diagram of a prior art high-speed test system;

[0016] FIG. 3 is a block diagram of an embodiment of a high-speed component interface;

[0017] FIGS. 4A-B are block diagrams of portions of two components configured to test each other according to one embodiment;

[0018] FIG. 5 is a timing diagram illustrating a transmitter testing operation using the embodiment of FIGS. 4A-B;

[0019] FIGS. 6A-B are block diagrams of portions of two components configured to test each other according to one embodiment;

[0020] FIG. 7 is a timing diagram illustrating a receiver testing operation using the embodiment of FIG. 6;

[0021] FIGS. 8A-B are block diagrams of portions of two components configured to test each other with parameters of respective components being uncoupled according to one embodiment;

[0022] FIG. 9 is a timing diagram illustrating a transmitter testing operation using the embodiment of FIG. 8;

[0023] FIG. 10 is a timing diagram illustrating a receiver testing operation using the embodiment of FIG. 8;

[0024] FIG. 11 is a block diagram of an embodiment with bipolar drivers;

[0025] FIG. 12 is a block diagram of an embodiment with single-ended signals;

[0026] FIG. 13 is a block diagram of an embodiment of IOL current control;

[0027] FIG. 14 is a block diagram of an embodiment of IOL current control;

[0028] FIG. 15 is a block diagram of an embodiment of VOL voltage control;

[0029] FIG. 16 is a block diagram of an embodiment of RT termination load control;

[0030] FIG. 17 is a block diagram of an embodiment of IOL current control;

[0031] FIG. 18 is a block diagram of an embodiment of clock aligner circuit;

[0032] FIG. 19 is a block diagram of an embodiment of clock aligner circuit;

[0033] FIG. 20 is a block diagram of an embodiment of a transmit clock aligner circuit;

[0034] FIG. 21 is a block diagram of an embodiment of a high-speed interface with a slew rate control circuit;

[0035] FIG. 22 is a block diagram of an embodiment of a high-speed interface with a test current source;

[0036] FIG. 23A is a block diagram of an embodiment of a clock aligner circuit used for frequency multiplication;

[0037] FIG. 23B is a timing diagram that shows the clock signals from the embodiment of FIG. 23A for both a transmitter circuit and a receiver;

[0038] FIG. 24 is a block diagram of an embodiment of a high-speed interface with clock, current, and termination value adjustment circuitry FIG. 25 is a block diagram of an embodiment of a clock aligner circuit with a PLL and a DLL;

[0039] FIG. 26 is a block diagram of an embodiment of adjustable termination circuit;

[0040] FIG. 27 is a block diagram of an embodiment of adjustable current circuit;

[0041] FIG. 28 is a block diagram of an embodiment of a high-speed interface with a single external clock reference;

[0042] FIG. 29 is a block diagram of an embodiment with two high-speed interfaces communicating using unidirectional signaling;

[0043] FIGS. 30A-B are block diagrams of an embodiment of a high-speed interface with an analog sampling circuit in a transmitter;

[0044] FIG. 31 is a block diagram of an embodiment of a coupled transmitter and receiver illustrating timing relationships;

[0045] FIG. 32 is a block diagram of an embodiment of a transmitter and receiver uncoupled by an analog sampling circuit, illustrating timing relationships; and

[0046] FIG. 33 is a block diagram of an embodiment of a transmitter and receiver uncoupled by two analog sampling circuits, illustrating timing relationships.

[0047] FIG. 34 is a block diagram of a transmit interface with pattern storage and a receive interface with pattern storage according to one embodiment;

[0048] FIG. 35A is a block diagram of a transmit interface, a receive interface, and interconnects according to one embodiment;

[0049] FIG. 35B is a block diagram of a transmit interface, a receive interface, and interconnects according to one embodiment;

[0050] FIG. 35C is a block diagram of a transmit interface, a receive interface, and an interconnect according to one embodiment;

[0051] FIG. 36 is a flow diagram of an embodiment of a pattern generation, storage and comparison process;

[0052] FIG. 37 is a block diagram of a transmit interface, a receive interface, and shared pattern generation and storage elements, according to one embodiment;

[0053] FIG. 38 is a block diagram of a transmit interface, a receive interface, and shared pattern generation and storage elements, according to one embodiment;

[0054] FIG. 39 is a flow diagram of an embodiment of a pattern generation, storage and comparison process;

[0055] FIG. 40 is a block diagram of a transmit interface, a receive interface, and dedicated pattern generation and storage elements, according to one embodiment;

[0056] FIGS. 41A-C are flow diagrams of an embodiment of a pattern generation, storage and comparison process;

[0057] FIG. 42 is a block diagram of a transmit interface, a receive interface, and dedicated pattern generation and storage elements, according to one embodiment;

[0058] FIGS. 43A-C are flow diagrams of an embodiment of a pattern generation, storage and comparison process;

[0059] FIG. 44 is a block diagram of a transmit interface, a receive interface, and dedicated pattern generation and storage elements, according to one embodiment;

[0060] FIGS. 45A-D are flow diagrams of an embodiment of a pattern generation, storage and comparison process;

[0061] FIG. 46 is a block diagram of a transmit interface, a receive interface, and dedicated pattern generation and storage elements, according to one embodiment; and

[0062] FIGS. 47A-C are flow diagrams of an embodiment of a pattern generation, storage and comparison process.

DETAILED DESCRIPTION

[0063] A method and apparatus is described for testing and characterizing semiconductor circuit components. A substantial part of the testing and characterization is performed by one component on itself or on another component. For example, a test system is not required to generate high accuracy and high-resolution signals for the test and characterization process. In one embodiment, a high-speed component interface includes additional circuitry to perform testing of the high-speed component. For example, the additional circuitry includes pattern generation, pattern memory, and comparison circuitry. A Rambus® XIO interface cell includes pattern storage and comparison logic. Also, a Rambus® ASIC cell (“RRAC”) interface includes pattern generation logic and pattern comparison logic. Various embodiments of component interface circuitry perform all normal testing and characterization on the interface itself, or on a similar interface, reducing the performance requirements of an external test system significantly. For example, an interface capable of very high data transfer rates, such as 3 GHz or greater normally requires a testing system with pin electronics capable of the same transfer rate. With the embodiments described, the external test system data transfer rates may be reduced by a factor of four or eight to 800 MHz or 400 MHz. This significantly reduces component testing costs. It also increases the number of test systems that are available for testing, since the highest speed testers have only limited availability. The embodiments also permit the same high-speed testing and characterization to be performed at each stage in the manufacturing process: at the wafer stage, at the package stage, and at the system stage. This further reduces manufacturing costs because components with out-of-specification interface parameters will be rejected at the earliest possible stage of manufacture.

[0064] The interface embodiment described may reside on one or multiple components. A component may test itself using test circuitry to be described below. Components may also test each other using test circuitry in their respective interfaces. Respective interfaces may be coupled through an interconnect for the specific purpose of testing one another. Alternatively, interfaces that are coupled to each other within a system for normal system operation, such as a memory component and a memory controller component, can test each other within the system through their usual system interconnect.

[0065] FIG. 3 is a block diagram of an embodiment of a high-speed component interface 300. The high-speed interface 300 is part of a component A and includes circuitry that facilitates communication with the “outside world” external to the component A via high-speed digital signaling. In one embodiment, a system including the component A is built with discrete integrated circuit components. In one embodiment, the digital signal travels between the interface 300 and another interface external to the interface 300 using the interconnect 308. In one embodiment, the digital signaling is differential signaling as known in the art, and the interconnect 308 is two conductive media. In other embodiments, other known signaling schemes, such as single-ended signaling, are used.

[0066] Information in the signal is carried as voltage values on the interconnect 308. The voltage values are restricted to two voltage ranges, because there are two possible symbols (bits with logic values of “1” or “0”) in this example. A voltage value within a first voltage range on interconnect 308 represents a bit having a logical 1 value or high value; while, a different voltage value within a second voltage range represents a bit having a logical 0 value or low value. A bit is driven for a specific length of time called a bit interval. The characteristic impedance of each interconnect 308 is ZO. In order to maximize the signaling rate of the system, a termination element, such as the termination elements 334a and 334b of termination block 306, are present at each end of interconnect 308. The resistive value RT of the termination elements 334a and 334b is approximately the same as the real part of the characteristic impedance ZO of the interconnect 308. One end of each termination element 334a and 334b connects to an interconnect 308, and the other connects to a power supply that sources a termination voltage of VT volts.

[0067] The interface 300 includes a transmitter 302 and a receiver 304. The high-speed signal is bidirectional. That is, information may travel from component A to another component B or from another component B to component A. In this example, the bit intervals of the component A transmitter may not overlap the bit intervals of the component B transmitter (including the effects of the propagation delay of signal wavefronts traveling on the interconnect 308). The transmitter 302 uses an internal timing signal, CLKTA, to frame its bit interval. An event on the timing signal determines when a new bit is to be driven.

[0068] Alternate embodiments might permit the bit intervals of the component A transmitter to overlap the bit intervals of the component B transmitter. This is called simultaneous bidirectional signaling. The testing features that have been added to interface 300 could also be added to an interface that permitted simultaneous bidirectional signaling.

[0069] The transmitter 302 further includes a predriver element 318, two differential driver elements 320 and 322, two current source elements 324 and 326, a current control element 314, a register element 316, and a transmit clock aligner element 312. In one embodiment, the register element 316 is a register cell that samples the TA signal (transmit signal from other circuits on component A) and drives it on the rising edge of the CLKTA signal. The transmit clock aligner element 312 creates the CLKTA signal from the CLKTREF reference signal. In various embodiments, the transmit clock aligner element 312 is a phase lock loop (“PLL”) or a delay lock loop (“DLL”) circuit. The output of the register element 316 connects to the predriver element 318, which connects to the differential driver elements 320 and 322. Each driver element 320 and 322 is an open drain transistor, and is connected in series with one of the current source elements 324 and 326, respectively. The current control element 314 maintains a sink current Of IOL (output low current) in the current source elements using an external reference value IREF.

[0070] In an alternate embodiment, the two current source elements 324 and 326 could be merged into a single current source element, with one terminal connecting to the low supply voltage (ground) and the other terminal connecting to the source terminals of the two driver elements 320 and 322. This will be equivalent to the circuit shown in transmitter 302 since only one of the two driver elements 320 and 322 may be on at any time.

[0071] When the transmitter 302 drives a bit, one of the two driver elements 320 and 322 is on, and the other is off. The output voltage of the driver element that is off will remain high at the termination value, and the output voltage of the driver element that is on will assume a lower voltage. As a result, high output voltage VOH (high logic value) and low output voltage VOL (low logic value) on interconnect 308 are:

VOH=VT  Equation 1

VOL =VT−IOL*RT/2  Equation 2

[0072] There is a “RT/2” factor in Equation 2 reflects the fact that the interconnect 308 is terminated at both ends in a termination load of RT which is presumed to match Zo, the characteristic impedance of interconnect 308. The specification for the transmitter circuit requires that the VP and VN voltage values remain above VOL,MAX or below VOH,MIN for a time tQ,MAX after the rising edge of CLKTREF until a time tV,MIN after the next rising edge of CLKTREF. The width of the (tQ,MAX−tV,MIN) window depends upon three factors in this example. The first factor is the width of the (tQ-T,MAX-tV-T,MIN) window of the TA register relative to the CLKTA rising edge. The second factor is the range of propagation delay (tPROP-T,MAX−tPROP-T,MIN) through the predriver element 318 and driver elements 320 and 322. The third factor is the range of timing skew (tSKEW-T,MAX−tSKEW-T,MIN) between the CLKTREF and CLKTA signals from the transmit clock aligner element 312. The following equation can be written:

(tQ,MAX−tV,MIN)=(tQ-T,MAX−tV-T,MIN)+(tPROP-T,MAX−tPROP-T,MIN)+(tSKEW-T,MAX−tSKEW-T,MIN)  Equation 3

[0073] Of the three factors, the width of the (tQ-T,MAX−tV-T,MIN) window of the TA register relative to the CLKTA rising edge will usually be the smallest, and the range of timing skew (tSKEW-T,MAX−tSKEW-T,MIN) between the CLKTREF and CLKTA signals will usually be the largest. The three factors have a timing range because of variations of manufacturing process (between different components) and variations in temperature and supply voltages during system operation.

[0074] The transmitter 302 elements described so far are similar to known transmitter elements. The transmitter 302 further includes a multiplexer 327. The multiplexer 327 selects data from one of two sources to be transmitted via the TA signal. Specifically, multiplexer 327 selects data from other circuits to be transmitted in normal operation. Alternatively, the multiplexer 327 selects a test pattern set from a test circuit 340, as described in more detail below.

[0075] The receiver 304 includes a voltage discriminator element 332, a register element 330, and a receive clock aligner element 328. In one embodiment, the voltage discriminator element 332 is a differential comparator that compares the voltage values on the VP (positive voltage) and VN (negative voltage) interconnects to determine what symbol is present (e.g., a “1” or a “0”). The register element 330 is a time discriminator element that samples the output of the voltage discriminator element 332 on the rising edge of the CLKRA signal (the receive clock signal of component A), stores it, and drives it as the RA signal (receive signal of component A), to be used by other circuits on the component A. The receive clock aligner element 328 creates the CLKRA signal from the CLKRREF (receive clock reference) reference signal. In various embodiments, the receive clock aligner element 328 is a phase lock loop (“PLL”) or a delay lock loop (“DLL”) circuit.

[0076] Typical specification for the receiver 304 requires that the VP and VN voltage values remain above VIL,MAX or below VIH,MIN for a time tS,MIN (minimum setup time) before and a time tH,MIN (minimum hold time) after the rising edge of CLKRREF. The width of the (tS,MIN+tH,MIN) window depends upon three factors in this example. The first factor is the width of the (tS-R,MIN+tH-R,MIN) window of the RA signal from register element 330 relative to the CLKRA rising edge. The second factor is the range of propagation delay (tPROP-R,MAX−tPROP-R,MIN) through the voltage discriminator 332. The third factor is the range of timing skew (tSKEW-R,MAX−tSKEW-R,MIN) between the CLKRREF and CLKRA signals from the receive clock aligner element 328. The following equation can be written:

(tS,MIN+tH,MIN)=(tS-R,MIN+tH-R,MIN)+(tPROP-R,MAX−tPROP-R,MIN)+(tSKEW-R,MAX−tSKEW-R,MIN)  Equation 4

[0077] Of the three factors, the width of the (tS-R,MIN+tH-R,MIN) window of the RA register relative to the CLKRA rising edge will usually be the smallest, and the range of timing skew (tSKEW-R,MAX−tSKEW-R,MIN) between the CLKRREF and CLKRA signals will usually be the largest. The three factors have a timing range because of variations of manufacturing process (between different components) and variations in temperature and supply voltages (during system operation). The receiver 304 uses an internal timing signal, CLKRA, to sample a bit during its bit interval. An event on the timing signal determines when a new bit is to be sampled and held.

[0078] The receiver 304 further includes a test circuit 340 and a test circuit 342. In one embodiment, the test circuit 340 consists of pattern memory and pattern generation circuitry 346 that creates a stream of bits, or pattern set, that may be steered to the transmitter 302 and driven onto the interconnect 308. Alternatively, the stream of bits may be compared in the compare circuit 348 to a second stream of bits which the receiver 304 has sampled from the interconnect 308, and the compare results (pass or fail) indicated. The test circuit 340 is controlled by other circuitry (not shown), or by other signals received by component A (not shown), or both. Typically, these other signals will utilize a low bandwidth interface, which will not present the testing and characterization issues that exist in prior testing systems. This other circuitry or other signals control the operation of the test circuit 340. For example, the other circuitry or signals dictates when the pattern set starts and ends, which pattern set is to be used, whether the transmitter 302 or receiver 304 is being used, and whether some of the pattern comparisons are to be ignored, or masked.

[0079] Alternatively, the compare circuit 348 could include delay circuitry that permits the stream of bits from the pattern memory and pattern generation circuitry 346 to be aligned to the bit stream from receiver 304. This would permit a single bit stream from the pattern memory and pattern generation circuitry 346 to be steered to the transmitter 302, driven to the receiver 304 in the same interface 300, and compared to the delayed version of the original bit stream.

[0080] The receiver 304 further includes a test circuit 342. The test circuit 342 includes an analog sampling circuit 356, a series of differential comparator elements 354, a series of register elements 352, and a multiplexer 350. The analog sampling circuit 356 includes two NMOS transistors 360a and 360b, and two capacitors 358a and 358b. As will be described in more detail below, the analog sampling circuit 356 samples by utilizing a TREF 362 signal that is used in embodiments for facilitating testing and characterization of one component by another and of one component by itself by decoupling respective transmitter and receiver parameters.

[0081] The differential comparator element 354a receives the outputs of analog sampling circuit 356. The analog sampling circuit 356 includes NMOS transistors 360a and 360b and capacitors 358a and 358b to sample voltage values VP and VN. The analog sampling circuit 356 and its function will be described in more detail below. The differential comparator element 354b receives VN and VREF. The differential comparator element 354c receives VP and VREF. The register element 352a receives the output of the differential comparator element 354a. The output of the register element 352a is the signal RTT, or receiver time test signal which has the same digital value as the received RA signal but is time-sampled by the analog sampling circuit 356 instead of being time-sampled by register element 330. The register element 352b receives the output of the differential comparator element 354b. The differential comparator elements 354b and 354c compare the respective VP and VN voltage values to a reference voltage signal VREF. The results of the two comparisons are received by the respective register elements 352b and 352c on the rising edge of CLKRA to produce the RTN (receiver test negative signal) and RTP (receiver test positive signal) signals. Note that the RTN signal is inverted relative to the RTP signal because the VP and VN voltage values are complementary.

[0082] Multiplexer 350 receive the outputs of all of the register elements 352. The multiplexer 350 selects one of the outputs of the register elements 352 or the received signal RA to transmit to the compare circuit 348 of the test circuit 340 for comparison with a test pattern. Signal RA is also continually sent to other circuits as in normal operation.

[0083] The component A receives a set of low bandwidth reference signals. The set of reference signals includes VREF 364, IREF 366, RREF 368, CLKRREF 370, CLKTREF 372, and TREF 362. The IREF 366 signal controls one of the voltage levels used by the transmitter. The RREF 368 signal controls the value RT of the termination elements 334a and 334b. The CLKRREF 370 and CLKTREF 372 signals provide timing references used to generate the internal timing signals CLKRA and CLKTA. In one embodiment, VREF 364, IREF 366, RREF 368, CLKRREF 370, CLKTREF 372 are reference signals normally supplied by a system to the components in the system to permit normal system operation, but they are also used to perform component testing and characterization. The TREF 362 signal is used in embodiments for facilitating testing and characterization of one component by another and of one component by itself by decoupling respective transmitter and receiver parameters.

[0084] Component-to-Component Test Example: Transmitter Test

[0085] FIGS. 4A-B are block diagrams showing relevant portions of two interfaces 400A1 and 400A2 on one component configured to test each other using the test circuits described with reference to FIG. 3, and a low bandwidth testing system. Alternatively, the two interfaces 400A1 and 400A2 could reside on two different components. Relevant portions of the interfaces 400A1 and 400A2 are shown. For interface 400A1, a transmitter 402, a termination block 406A1, and a test circuit 440A1 are shown. Elements in FIGS. 4A-B function similarly to their similarly numbered counterparts in FIG. 3. Where there are instances of a similar element or sub-element in both component A1 and component A2 are shown in one figure, the elements are numbered with a suffix “A1” or “A2” as appropriate. In some instances, individual sub-elements are not numbered or discussed as to simplify FIGS. 4A-B. For example, sub-elements of the termination blocks 406A1 and 406A2 are not numbered. For interface 400A2 a receiver 404, a test circuit 440A2, a test circuit 442, and a termination block 406A2 are shown.

[0086] Interfaces 400A1 and 400A2 communicate with each other via the interconnect 408. Interfaces 400A1 and 400A2 further receive low bandwidth, or low speed, reference signals from low bandwidth testing system 403. The low bandwidth testing system 403 includes a low-speed digital interface 405. The low bandwidth testing system 403 drives four reference signals previously discussed: IREF 466, RREF 468, CLKRREF 470, and CLKTREF 472. The low bandwidth testing system 403 also drives the VREF 464 signal, which is used by the test circuit 442. A low-speed digital interface 405 produces signals that control the test circuits 440A1, 440A2, and 442 in the two interfaces 400A1 and 400A2. FIGS. 4A-B will be used to illustrate an embodiment in which two sets of interface logic can be configured to test each other using the additional test circuitry previously described. For example, the transmitter of one interface tests the receiver of the other interface, and vice-versa. The two sets of interface logic can reside on the same component when component testing and characterization is being performed at either the wafer level or the package level. When testing and characterization is being performed within an actual system (typically when normal operation has been suspended), the two sets of interface logic will usually reside on different components. In other words, interfaces 400A1 and 400A2 in FIGS. 4A-B may or may not reside on the same component.

[0087] For component testing it is also possible that interface logic for a single signal could be configured to test itself; that is, the transmitter for interface 400A1 could test the receiver for interface 400A1 and the receiver for interface 400A1 could test the transmitter for interface 400A1. Embodiments with this capability are described in a later section.

[0088] In the configuration of FIGS. 4A-B, the transmitter 402 of the interface 400A1 is being tested. The receiver 404 of the interface 400A2 is used to scan a sense voltage and to scan a sample time. In embodiments of the present invention, scanning is defined as varying a timing of a transition of a signal in a time range and/or varying a voltage value of a signal in a voltage range. In an embodiment of the present invention, scanning is done to determine a pass/fail transition boundary. When the transmitter 402 drives a bit, one of the two driver elements 420 and 422 is on and the other is off. The output voltage of the one that is off will remain high at the termination voltage, VT, and the output voltage of the one that is on will assume a lower voltage. The following equations apply.

V OH,T =V T   Equation 5

V OL,T =V T −I OL,T *R T /2  Equation 6

[0089] There is an “RT/2” factor reflects the fact that the interconnect 408 is terminated at both ends in a termination load of RT which is presumed to match Zo, the characteristic impedance of interconnect 408. The IREF 466 and CLKTREF 472 reference signals to interface 400A1 are adjusted to nominal values. The test circuit 440A1 produces a pattern set from the pattern memory and pattern generator circuitry. This pattern set is transmitted on the interconnect 408 to the interface 400A2. Test circuit 442 receives the pattern set. The VREF 464 reference voltage is scanned across voltage ranges, and the CLKRREF 470 reference signal is scanned across timing intervals to test and characterize the transmitter 402.

[0090] FIG. 5 is timing diagram showing the voltage value for the signals on a number of interconnects as a function of time during the scanning operations to evaluate the transmitter 402. Signals from the transmitter 402 of interface 400A1 are shown on approximately the top half of the diagram and signals from the receiver 404 of component A2 are shown on approximately the bottom half of the diagram. Consider that the specification for the transmitter 402 requires that the timing and voltage parameters for the VP/NN signal fall within the specification-output-eye (“Output Eye (From Component Transmitter)”) region 502. That is, the signal must be valid no later than tQ,MAX after a rising edge of CLKTREF 472, and must remain valid until at least a time tV,MIN after the next rising edge of CLKTREF 472. During this valid interval, the voltage level must be above VOH,MIN or must be below VOL,MAX. The actual transmitter 402 has the following parameter values: tQ,T, tV,T, VOH,T and VOL,T. tQ,T is the time from the active (rising, in this case) clock edge to valid data. tV,T is the time from the next rising clock edge to non-valid data. VOH,T is the high output voltage (high logic value), and VOL,T is the low output voltage (low logic value). These parameter values define an output-eye-from-transmitter (“Output Eye (from Component Transmitter)”) region 504, which lies outside the specification-output-eye region 502. These parameter values are a function of several factors, including the manufacturing process that produced the component under test, the reference values applied, the operating temperature and supply voltages of the component under test, and the test patterns used when characterizing the component under test. The test circuits 440 and 442 on the components measure the actual output-eye-from-transmitter region 504 of the transmitter 402 to confirm that it is larger than the specification-output-eye region 502. This is done by scanning the sample time and the sense voltage used by the receiver 404.

[0091] The actual receiver 404 has the following characteristics: tS,R, tH,T, VIH,R and VIL,R. tS,R is the receiver setup time, tH,T is the receiver hold time, VIH,R is the receiver high input voltage, and VIL,R is the receiver low input voltage. These parameter values define an input-eye-to-component-receiver (“Input Eye (To Component Receiver)” region 514. The receiver 404 parameters and the input-eye-to-component-receiver region 514 will be discussed in more detail below with reference to receiver testing.

[0092] To characterize the margin in the VOH parameter, the VREF 464 voltage supplied by the low bandwidth testing system 403 scans from the VOH,MIN value to the VOH,T value. The sampled patterns will match the expected patterns (pass) at each parameter value until a fail point 506 just past (above) VOH,T. If a simple pass/fail test is desired, the voltage VREF 464 is set to the VOH,MIN parameter value. When the VP node is checked, the RTP multiplexer input is selected in test circuit 442. When the VN node is checked, the RTN multiplexer input is selected in test circuit 442.

[0093] To characterize the margin in the VOL parameter, the VREF 464 voltage supplied by the low bandwidth testing system 403 scans from the VOL,MAX value to the VOL,T value. The sampled patterns will match the expected patterns (pass) at each parameter value until a fail point 508 just past (below) VOL,T. If a simple pass/fail test is desired, the voltage VREF 464 is set to the VOL,MAX parameter value. Note that when the VP node is checked, the RTP multiplexer input must be selected in test circuit 442. When the VN node is checked, the RTN multiplexer input must be selected in test circuit 442.

[0094] To characterize the margin in the tQ parameter, the position of the CLKRREF 470 edge in the receiver 404 relative to the CLKTREF 472 edge in the transmitter 402 (tCLKTR-SQ) is scanned. The following equation will be true during the scanning process:

t CLKTRS −t s =t Q   Equation 7

or

t CLKTR-SQ =t s +t Q   Equation 8

[0095] Thus, the position of the CLKRREF 470 edge in the receiver 404 relative to the CLKTREF 472 edge (tCLKTR-SQ) will begin scanning at a passing value of (tS,MIN+tQ,MAX), and will decrease in value until it fails somewhere around point 510 as it reaches the value (tS,R+tQ,T). This process will not yield the minimum value of (tQ,T), but instead yields the minimum value of (tS,R+tQ,T). In other words, the transmitter parameter tQ,T is coupled to the receiver parameter tS,R. The consequences of this will be discussed shortly. To characterize the margin in the tV parameter, the position of the CLKRREF 470 edge in the receiver 404 relative to the CLKTREF 472 edge in the transmitter 402 (tCLKTR-Hv) is scanned. The following equation will be true during the scanning process:

tCLKTR-HV+tH=tV  Equation 9

or

t CLKTR-HV =t V −t H   Equation 10

[0096] Thus, the position of the CLKRREF 470 edge in the receiver 404 relative to the CLKTREF 472 edge (tCLKTR-Hv) will begin scanning at a passing value of (tV,MIN−tH,MIN), and will increase in value until it fails somewhere around point 512 as it reaches the value (tV,T−tH,T). This process will not yield the maximum value of (tV,T), but instead yields the maximum value of (tV,T−tH,T). In other words, the transmitter parameter tVT is coupled to the receiver parameter tH,T. The consequences of this will be discussed shortly.

[0097] In one embodiment, the scanning process that is performed for each parameter includes the following steps.

[0098] 1. Set CLKRREF 470 time or VREF 464 voltage to an initial passing value;

[0099] 2. Transmitter 402 drives pattern set;

[0100] 3. Receiver 404 compares sampled pattern set to expected pattern set;

[0101] 4a. If pass (pattern sets match), change CLKRREF 470 time or VREF 464 voltage to the next (more difficult) value, and go to step [2]; or

[0102] 4b. If fail, exit test with last passing value.

[0103] Alternative processes are possible.

[0104] As each parameter is characterized or tested, the other parameters are usually left at nominal values to ensure that they are not affecting the test and characterization results. In some cases one or more other parameters are given a more constrained value.

[0105] Component-to-Component Test Example: Receiver Test

[0106] In a single component having two interfaces, each interface is able to test the other interface in an embodiment of the present invention. FIGS. 6A-B are block diagrams showing relevant portions of two interfaces 600A1 and 600A2 on one component configured to test each other using the test circuits described with reference to FIG. 3, and a low bandwidth testing system. Alternatively, the two interfaces 600A1 and 600A2 could reside on two different components. Relevant portions of the interfaces are shown. For interface 600A1, a receiver 604, a termination block 606A1, a test circuit 640A1, and a test circuit 642 are shown. Elements in FIGS. 6A-B function similarly to their similarly numbered counterparts in FIG. 3. Where there are instances of a similar element or sub-element in both interfaces 600A1 and 600A2 are shown in one figure, the elements are numbered with a suffix “A1” or “A2” as appropriate. In some instances, individual sub-elements are not numbered as to simplify FIGS. 6A-B. For interface 600A2, a transmitter 602, a test circuit 640A2, and a termination block 606A2 are shown.

[0107] In the configuration of FIGS. 6A-B, the receiver of interface 600A1 can be tested by the transmitter of interface 600A2. Interfaces 600A1 and 600A2 will usually reside on the same component when component testing and characterization is being performed (at both the wafer level and the component level). However, when testing and characterization is being performed within an actual system (typically when normal operation has been suspended), interfaces 600A1 and 600A2 will usually reside on different components. For component testing, a single interface can be configured to test itself; that is, the transmitter could test the receiver and vice-versa. Such an embodiment will be described in a later section.

[0108] In FIGS. 6A-B, a receiver 604 is being tested. The interface 600A2 includes a transmitter 602, which is used to scan a force current and to scan a drive time. There is also a low bandwidth testing system 403, which has no high-speed pin electronics. An interconnect 608 is used to couple interface 600A1 to interface 600A2, and is not coupled to the low bandwidth testing system 403. The low bandwidth testing system 403 does drive the four reference signals previously discussed: IREF466, RREF 468, CLKRREF 470, and CLKTREF 472. It also drives the VREF 464 signal, which is used by test circuit 642. Also shown is the low-speed digital interface 405, which produces signals that control the test circuits 640 and 642 in the interfaces 600A1 and 600A2.

[0109] When the transmitter 602 drives a bit, one of the two driver elements 620A2 and 622A2 is on (sinking current ITEST,L) and the other is off. The output voltage of the driver element that is off will remain high at the termination value, and the output voltage of the driver element that is on will assume a lower voltage. In addition, both driver elements 620A1 and 622A1 in interface 600A1 are on, sinking current ITEST,H. As a result, the high voltage VIH,R (high logic value) and low voltage VIL,R (low logic value) produced for the receiver in interface 600A1 are:

V IH,R =V T −I TEST,H *R T /2  Equation 11

V IL,R =V T −I TEST,H *R T /2−ITEST,L*RT/2  Equation 12

[0110] The IREF and IREF2 signals are part of the 661 and 662 reference signals supplied by the test system, and they control the values of ITEST,H and ITEST,L, respectively. The ITEST,H and ITEST,L signals control the values of the VIH,R and VIL,R voltage levels as indicated by Equations 11 and 12. The “RT/2” factor reflects the fact that the interconnect 608 is terminated at both ends in a termination load of RT. This allows the VIH,R and VIL,R voltage levels to be independently adjusted by changing the values of the IREF1 and IREF2 signals.

[0111] Alternatively, the VIH,R and VIL,R voltage levels could be independently adjusted by changing the value of the IREF2 signals and the value of the VT termination voltage. In this case, it would not be necessary to turn on the two driver elements 620A1 and 620A1 so that the ITEST,H current is pulled from the signal lines providing the VP and VN voltage values. This could be beneficial if the transmitter and the receiver were in the same interface, rather than belonging to two different interfaces.

[0112] The CLKRREF 470 reference signal to interface 600A1 is adjusted to a nominal value. The IREF1 signal in 661 reference signals to interface 600A1 is adjusted to the value ITEST,H and both driver elements 620A1 and 620A2 of interface 600A1 are enabled. The IREF2 signal in 662 reference signals to interface 600A2 is adjusted to the value ITEST,L.

[0113] Test circuit 640A2 in the interface 600A2 produces a pattern set. This pattern set is transmitted on the interconnect 608 from the interface 600A2 to the interface 600A1. The pattern set is received by the test circuit 642A1 in the interface 600A1. The IREF1 signal in 661 reference signals and IREF2 signal in 662 reference signals are scanned across current ranges, and the CLKTREF signal in 662 reference signals is scanned across timing intervals to test and characterize the receiver 604 in the interface 600A1.

[0114] FIG. 7 is a timing diagram showing the voltage values for the signals on a number of interconnects as a function of time for the scanning operations that evaluate the receiver 604 of interface 600A1. The specification for the receiver requires that the timing and voltage parameters for the VP/NN signal fall within the specification-input-eye (“Input Eye (Specification)”) region 704 in order for receiver 604 to pass a test based on the specification requirements. Receiver 604 passes the test if it meets or slightly exceeds specification requirements. However, it should be noted that receiver 604 might still operate properly if one or more timing or voltage parameters do not meet specification requirements. For example, the receiver 604 must be able to sample a signal that becomes valid at time tS,MIN (or later) before a rising edge of CLKRREF 470 in order to pass the test. The receiver 604 must also be able to sample a signal that remains valid until a time tH,MIN (or earlier) after a rising edge of CLKRREF 470 in order to pass the test. Likewise, during this valid interval, the receiver 604 must be able to sense a high voltage level that is equal to or less than VIH,MIN. Also, the receiver must be able to sense a low voltage level that is equal to or more than VIL,MAX. Consider that the actual receiver 604 has the following characteristics: tS,R, tH,T, VIH,R and VIL,R. These parameter values define the input-eye-to-component-receiver region 514, which lies inside a specification-input-eye (“Input Eye (Specification)”) region 704. These parameter values are a function of several factors, including the manufacturing process that produced the component under test, the reference values applied, the operating temperature and supply voltages of the component under test, and the test patterns used when characterizing the component under test.

[0115] The test circuits 640A1 and 642A1 on the component measures the input-eye-to-component-receiver region 514 of the receiver 604 and confirms that it is smaller than the specification-input-eye region 704. It does this by scanning the drive time and the force voltage (or current) used by the transmitter 602. To characterize the margin in the VIH parameter, the ITEST,H value that is driven on the IREF1 reference current supplied by the testing system is scanned. It scans from a value that generates the VIH,MIN parameter value, and is increased until the VIH,T value is reached (recall that VIH=VT−ITEST,H*RT/2). The sampled patterns will match the expected patterns (pass) at each parameter value until it fails at a point 708 just past (below) VIH,T. If a simple pass/fail test is desired, the sense voltage of the tester is set to the VIH,MIN parameter value. The ITEST,H value can be scanned from a passing value until it just fails. It can be placed back to the last passing value, and then the VREF reference signal can be scanned (and the RTP or RTN node monitored) to determine the exact value of VIH,R. In this way, any inaccuracy in the values of ITEST,H, VT and RT can be eliminated and VIH,R determined directly.

[0116] To characterize the margin in the VIL parameter, the ITEST,L value that is driven on the IREF2 reference current supplied by the testing system is scanned. It scans from a value that generates the VIL,MAX parameter value, and is increased until the VIL,R value is reached (recall that VIL=VT−ITEST,H*RT/2−ITEST,L*RT/2). The sampled patterns will match the expected patterns (pass) at each parameter value until it fails at a point 710 just past (above) VIL,R. If a simple pass/fail test is desired, the sense voltage of the tester is set to the VIL,MAX parameter value. The ITEST,L value can be scanned from a passing value until it just fails. It can be placed back to the last passing value, and then the VREF reference signal can be scanned (and the RTP or RTN node monitored) to determine the exact value of VIL,R. In this way, any inaccuracy in the values of ITEST,H, ITEST,L, VT and RT can be eliminated and VIL,R determined directly.

[0117] To characterize the margin in the tS parameter, the position of the CLKTREF 472 edge in the transmitter 602 relative to the CLKRREF 470 edge in the receiver 604 (tCLKRT-SQ) is scanned. The following equation will be true during the scanning process:

t CLKRT-SQ +t Q =t CYCLE −t s   Equation 13

or

t CLKRT-SQ =t CYCLE −t S −t Q   Equation 14

[0118] Thus, the position of the CLKTREF 472 edge in the transmitter 602 relative to the CLKRREF 470 edge (tCLKRT-SQ) will begin scanning at a passing value of (tCYCLE−tS,MIN−tQ,MAX), and will increase in value until it fails at approximately point 712 as it reaches the value (tCYCLE−tSR−tQ,T). This process will not yield the minimum value of (tS,R), but instead yields the minimum value of (tS,R+tQ,T). In other words, the transmitter 602 parameter tQ,T is coupled to the receiver parameter tS,R. The consequences of this will be discussed shortly.

[0119] To characterize the margin in the tH parameter, the position of the CLKTREF 472 edge in the transmitter 602 relative to the CLKRREF 470 edge in the receiver 604 (tCLKTR-Hv) is scanned. The following equation will be true during the scanning process:

t CLKRT-HV +t V =t H   Equation 15

or

t CLKRT-HV =t H −t V   Equation 16

[0120] The position of the CLKTREF 472 edge in the transmitter 602 relative to the CLKRREF 470 edge (tCLKRT-HV) will begin scanning at a passing value of (tH,MIN−tV,MIN), and will decrease in value until it fails at approximately point 714 as it reaches the value (tH,T−tV,T). This process will not yield the minimum value of (tH,T), but instead yields the minimum value of (tH,T−tV,T). In other words, the transmitter parameter tV,T is coupled to the receiver parameter tH,T. The consequences of this will be discussed shortly.

[0121] In one embodiment, the scanning process that is performed for each parameter includes the following steps. Alternative processes are possible.

[0122] 1. Set CLKTREF time or IREF current to an initial passing value

[0123] 2. Transmitter A2 drives pattern set

[0124] 3. Receiver A1 compares driven pattern set to expected pattern set

[0125] 4a. If pass (pattern sets match), change CLKTREF time or IREF current to the next (more difficult) value, and go to step [2].

[0126] 4b. If fail, go to [5] with last passing value

[0127] 5. For the VIHNIL tests, measure the final passing VIH,R/VIL,R voltage values on the signal lines providing VP/NN voltage values using the VREF comparators; i.e. scan the VREF value until the RTP or RTN digital value changes

[0128] Note that as each parameter is characterized or tested, the other parameters will usually be left at nominal values to ensure that they are not affecting the test and characterization results. In some cases one or more other parameters will be given a more constrained value.

[0129] Coupled Transmitter and Receiver Parameters

[0130] The scanning processes for the timing parameters that are illustrated in FIG. 5 and FIG. 7, and explained in the accompanying text, yield the following four equations:

t CLKTR-SQ =t S,R +t Q,T   Equation 17

t CLKTR-HV =t V,T −t H,T   Equation 18

tCLKRT-SQ=tCYCLE−tS,R−tQ,T  Equation 19

t CLKRT-HV =t H,T −t V,T   Equation 20

[0131] The first two equations (Equations 17 and 18) and the last two equations (Equations 19 and 20) will reduce to essentially the same values for (tS,R+tQ,T) and (tH,T−tV,T). A distinction between the processes illustrated in FIGS. 5 and 7 is that CLKTREF 470 signal is scanned in one, and CLKTREF 472 is scanned in the other. This means that the testing procedure can only guarantee that:

(tS,R+tQ,T)≦(tS,MIN+tQ,MAX)  Equation 21

(tH,T−tV,T)≦(tH,MIN−tV,MIN)  Equation 22

[0132] In some testing situations, this might be acceptable. For example, for an in-system test the transmitter and receiver are located on different components. Also, this same transmitter and receiver pair must actually communicate during normal system operation. The constraints that must be satisfied during normal system operation are the above two equations. If the interconnect couples more than two components together, then each transmitter/receiver pair that must communicate may be tested to the above two constraints. When a single component is tested, however, the transmitter and receiver pair that are coupled together by interconnect on the load board of the test system are not the same transmitter and receiver pair that are used for normal system operation. This means that even if a transmitter and receiver pair can communicate on the same component, it's possible that a transmitter and receiver pair on two different components might not be able to communicate. For example, assume that component A and component B have the following tS and tQ parameter values:

t S,A =t S,MIN +tΔ   Equation 23

t S,B =t S,MIN −tΔ   Equation 24

t Q,A =t Q,MAX −tΔ   Equation 25

t Q,B =t Q,MAX +tΔ   Equation 26

[0133] When component A and component B are tested separately, and the transmitter and receiver are on the same component, the coupled parameter constraint is satisfied even though each component has an individual parameter that is out of specification:

(tS,A+tQ,A)=(tS,MIN+tΔ+tQ,MAX−tΔ)≦(tS,MIN+tQ,MAX) Pass

(tS,B+tQ,B)=(tS,MIN−tΔ+tQ,MAX+tΔ)≦(tS,MIN+tQ,MAX) Pass

[0134] When component A and component B are tested together, but the transmitter and receiver are on different components, the coupled parameter constraint is not satisfied when B transmits and A receives:

(tS,A+tQ,B)=(tS,MIN+tΔ+tQ,MAX+tΔ)>(tS,MIN+tQ,MAX) FAIL

(tS,B+tQ,A)=(tS,MIN−tΔ+tQ,MAX−tΔ)≦(tS,MIN+tQ,MAX) Pass

[0135] For some components, there may be correlation between the tS,R and tQ,T values and between the tH,T and tV,T values such that if the coupled parameters meet the specification, then the individual parameters do, as well. Further, the in-system testing guarantees that the actual components in each system can communicate, and serves as a final screening test to insure that all systems have adequate parameter margin to communicate reliably. If there is correlation, then the testing process described with reference to FIGS. 5 and 7 will be acceptable for component and in-system testing and characterization.

[0136] If the proper parameter correlation between the tS,R and tQ,T values and between the tH,T and tV,T values cannot be established, some systems would fail in-system testing due to poorly matched components. This would require repair of the failing systems, adding cost to the manufacturing process. It is possible that this cost could be offset by improved component screening earlier in the manufacturing process. In a typical manufacturing process, components are tested twice, once at the wafer level before packaging, and again after packaging. In other words, there are at least three testing stages:

[0137] 1. Component wafer test—determine which components are packaged

[0138] 2. Component package test—determine which components are used in system

[0139] 3. In-system test—determine which systems work properly

[0140] Usually high-speed testing is only performed during the second test. Using the testing process described in FIGS. 5 and 7, it is possible to perform high-speed testing during both tests (testing to the coupled parameter constraints). This reduces the number of components that pass the component wafer test and fail the component package test. This reduces manufacturing cost, because fewer out-of-specification components would pass the component wafer test. This might serve to offset the increased cost that is incurred because of components passing at the component package level and failing at the system level.

[0141] Uncoupled Transmitter Parameter Test—Test System

[0142] Returning to the topic of coupled parameter values for the transmitter and receiver, one solution is to determine the values of the individual timing parameters. Determining individual timing parameters by scanning a low bandwidth TREF signal, will now be discussed. In addition, as will be further explained, the TREF signal is used to sample VP and VN with an analog sampler to produce a delay or offset between a data transmission by a transmitter and the sampling of the data by a receiver. The sampling by the receiver should be distinguished from the sampling by the analog sampling circuit. The sampling of the receiver is sampling of a symbol, or evaluation of a voltage to determine whether it represents a “1” or a “0”.This offset effectively decouples respective parameters of the transmitter and receiver. In one embodiment, the values of the individual timing parameters are determined using circuitry such as test circuit 342 of FIG. 3. Referring to FIG. 3, the test circuit 342 includes analog sampling circuit 356, differential comparator elements 354, and register 352, which are clocked by CLKRA. The register output RTT is selectable by the multiplexer to be transmitted to the compare logic in the test circuit 340. The signal lines providing the VP and VN voltage values can be coupled to the inputs of the differential comparator element 354a by raising the gates of two NMOS transistors high with the TREF 362 signal. When the TREF 362 signal goes low, the voltage values on the inputs are retained by the two capacitors 358a and 358b.

[0143] The VP and VN voltage values on the signal lines must satisfy a setup time ts′ and hold time tH′ constraint relative to the falling edge of the TREF 362 signal. This setup and hold window is relatively narrow and remains relatively constant across process, voltage, and temperature variations. This is because the TREF 362 signal is routed directly from the component interface to the test circuit 342 with only a conductive interconnect; there is no alignment and buffering circuitry as there is with the CLKRREF 370 signal, for example. The slew rate and voltage swing of the TREF 362 signal affects the characteristics of the analog sampling circuit 356 most directly, and this is controlled by a low bandwidth external test system (such as the low bandwidth testing system 403). The RC delay of the conductive interconnect used for the TREF 362 signal on the component under test is another factor, but it also will not be strongly affected by process, voltage, and temperature variations.

[0144] The analog sampling circuit 356 may be operated as a timing gate in front of the normal sampler (the RTT register element 352a clocked by CLKRREF 370). As mentioned earlier, the normal sampler (the RTT register element 352a) also has a relatively narrow setup and hold window relative to the CLKRA signal, but the receive clock aligner element 328 adds a great deal of uncertainty to the position of the CLKRA signal relative to the CLKRREF 370 signal because of process, voltage, and temperature variations. Placing the analog sampling circuit 356 in series with the normal sampler avoids much of this variation.

[0145] The voltage discriminator element, the differential comparator elements 354, also adds to the uncertainty to the position of the input signals to register elements 352 relative to the CLKRA signal because of process, voltage, and temperature variations. Placing the analog sampling circuit 356 before the differential comparator elements 354 avoids widening the effective setup and hold window of the analog sampler. This, of course, requires that the analog VP and VN voltage values on the signal lines are retained in the time-discrimination (sampling) process, hence the name “analog sampler”. In contrast, the time-discrimination process (the register elements 352) are “digital samplers”, since they accept a digital input from the differential comparator elements 354. The effective setup and hold window of the register elements 352 are widened by the presence of the differential comparator elements 354 in the input path because of process, voltage, and temperature variations.

[0146] The analog sampling circuit 356 is more limited than the normal sampler, and may not be suitable for use as the normal sampler in the component. One reason for this is that in some components the frequency of CLKRA will be an integer multiple of the frequency of CLKRREF 370. This permits more than one bit to be transferred in a clock cycle. This allows a high signaling rate for the transmitter and receiver circuitry, while permitting the signaling rate of the clock reference to be relatively low. A clock signal distributed like the TREF signal is probably not a good choice for such an application.

[0147] This embodiment takes advantage of the fact that it is relatively easy for an external tester to provide highly accurate voltage (VREF 364, for example) and timing values (TREF 362, for example) to a component under test by using directly coupled reference signals (signals traveling directly on interconnect from the component interface to internal test circuits). It is not necessary for these precise reference signals to be simultaneously capable of high signaling rates. Instead, these precise reference signals complement the high signaling rate testing that can be performed by the components circuits on themselves. The circuitry shown in test circuit 342 is an example of an implementation that solves the problem of separating the coupled parameter values. Other implementations that solve the same problem are possible. The various pieces of circuitry in test circuit 342 may also be combined in other ways to reduce the cost to the component.

[0148] FIGS. 8A-B are block diagrams of an embodiment in which an analog sampling circuit is used in a transmitter timing test. FIGS. 8A-B are similar to the previous FIGS. 4A-B, except that the scanning for the output voltage levels has been suppressed, and the time scanning is now being performed on the TREF 862 signal instead of the CLKRREF 470 signal. The CLKRREF 470 signal is placed at a nominal value so that it does not interfere with the determination of the transmitter timing parameters.

[0149] An interface 800A1 and an interface 800A2 are shown coupled by an interconnect 808. The interconnect 808 is not coupled to the low bandwidth testing system 403. The interfaces 800A1 and 800A2, however, receive signals from the low bandwidth testing system 403. The low bandwidth testing system 403 produces the reference signals VREF 464, IREF 466, RREF 468, CLKRREF 470, and CLKTREF 472, as previously described. The low bandwidth test system also produces a TREF 862 signal that is received by the test circuit 842 of the interface 800A2 on the component A2. In the embodiment shown, the transmitter 802 of the interface 800A1 is under test. Only portions of the interfaces 800A1 and 800A2 that are pertinent to the following test example are shown. The interface 800A1 includes the transmitter 802, a termination block 806A1, and a test circuit 840A1. The interface 800A2 includes a receiver 804, a termination block 806A2, a test circuit 840A2, and a test circuit 842. The test circuit 842 is similar to the test circuit 342, but here only pertinent portions of the test circuit 842 are shown for the example, including an analog sampling circuit 856, a differential comparator 854a, a register 852a, and a multiplexer 850.

[0150] The analog sampling circuit 856 is operated as a timing gate in front of the normal testing sampler (the RTT register 852a clocked by CLKRREF 470). As mentioned earlier, the normal testing sampler (the RTT register 852a) also has a relatively narrow setup and hold window relative to the CLKRA signal, but the receive clock aligner 828 adds a great deal of uncertainty to the position of the CLKRA signal relative to the CLKRREF 470 signal because of process, voltage, and temperature variations. Also, the differential comparator 854a adds a great deal of uncertainty to the position of the input signal to register 852a signal relative to the CLKRA signal because of process, voltage, and temperature variations

[0151] Placing the analog sampling circuit 856 before the differential comparator 854a and register 852a avoids much of this variation.

[0152] In one embodiment (not shown), the receiver 804 is duplicated, with one receiver being used for normal operation, and the other being used for evaluation as described further below.

[0153] Uncoupled Transmitter Parameter Test—Time Scanning

[0154] The operation of the embodiment of FIGS. 8A-B will be further described with reference to FIG. 9, which is a diagram showing the voltage values for the signals on a number of interconnects as a function of time during the test of the transmitter 802. FIG. 9 shows an actual output-eye-from-component-transmitter (“Output Eye (From Component Transmitter)”) region 902, and a specification-output-eye (“Output Eye (Specification)”) region 901. The output-eye-from-component-transmitter region 902 is the “actual” output eye of the transmitter under test. A successful test of the transmitter 802 should verify that the specification-output-eye 901 is smaller than the output-eye-from-component-transmitter 902. The voltage signals listed on the left from the top of the diagram are as follows: signals CLKTREF 472 and VP/NN associated with the transmitter 802, the TREF 862, VP′/VN′, and CLKRREF 470 signals associated with finding the tQ,T value for the transmitter 802, and the TREF 862, VP′/VN′, and CLKRREF 470 signals associated with finding the tV,T values for the transmitter 802. The tQ,T and tV,T values define the output-eye-from-component-transmitter region 902.

[0155] The CLKTREF 472 signal is used as a timing reference. The CLKTREF 472 signal generates the output-eye-from-component-transmitter region 902 whose left boundary is at time tQ,T after the rising edge of CLKTREF 472, and whose right boundary is at time tV,T after the next rising edge of CLKTREF 472. These two boundaries are scanned by the TREF 862 signal.

[0156] The falling edge of the TREF 862 signal is adjusted to occur at a time tTREF-Q after the rising edge of the CLKRREF signal. Then the following equation will be true:

t TREF-Q =t Q +t S,R′   Equation 27

[0157] The tTREF-Q value is then scanned from the passing time (tQ,MAX+tS,R′) to a point 904 just past time (tQ,T+tS,R′) where it fails. Here the parameter tS,R′is the time needed to setup the analog sampling circuit. This parameter is relatively small, and will not vary much as a function of process, temperature and voltage. The tQ,T parameter will be equal to:

t Q,T =t TREF-Q −t S,R′   Equation 28

[0158] Where tTREF-Q is the offset of the TREF 862 signal at the last passing scan point. The range of tS,R′is known from characterization, and its minimum value can be subtracted from tTREF-Q. This determines tQ,T. Because the sum of (tS,R+tQ,T) is known from the earlier testing, the value of tSR can be determined.

[0159] During this scanning process, the CLKRREF 470 signal has been set to a value that guarantees that the output of the differential comparator has settled ahead of the tS,R setup time of the RTT register 852a. The capacitors of the analog sampling circuit 856 hold the sampled VP/NN voltage values during this interval, as shown at 1011. This allows the TREF 862 signal to determine the sampling point rather than the CLKRREF 470 signal (with its greater variability).

[0160] If the component under test allows the internal CLKRA and CLKTA clock frequencies to be an integer multiple of the CLKRREF 470 and CLKTREF 472 clock frequencies (not shown in this example), then more than one bit can be transferred between transmitter and receiver during each tCYCLE interval. In this case, it is likely that the TREF 862 signaling rate will be limited to the same rate as the CLKRREF 470 and CLKTREF 472 clock signals; a rising edge and a falling edge in each tCYCLE interval. In this case, scanning the TREF 862 signal can check only one of the output-eye regions in each tCYCLE interval. In this situation, masking logic can be added to the compare logic of the test circuit 840A2.

[0161] The masking logic will allow only one bit in each tCYCLE interval to be checked, and the rest ignored. However, the other bits of the interval can be checked (one at a time) by adjusting the masking logic and the offset of the TREF 862 signal. For example, if four bits are transferred in each tCYCLE interval, then the scanning process described with reference to FIG. 9 is repeated four times. During each of the four iterations, one out of every four of the bits RTT (from the register 852a) would be compared in the test circuit 840A2. Between each iteration, the tTREF-Q value is shifted by tCYCLE/4. In this way, every bit of every pattern is checked to ensure that the worst case tQ,T value is found. This can be done without requiring a tester that generates high-speed signals; the tester is required to only generate signals with timing events that are accurately positioned. This will be described in more detail with FIG. 23B.

[0162] Referring now to the lower portion of FIG. 9, finding the tV,T value using a similar scanning process is illustrated. The falling edge of the TREF 862 signal is adjusted to occur at a time tTREF-V after the rising edge of the CLKRREF 470 signal.

[0163] The following equation will be true:

t TREF-V =t v −t H,R′   Equation 29

[0164] The tTREF-V value is then scanned from the passing time (tV,MIN−tH,R′) to just past the time (tV,T−tH,R′) where it fails. Here the parameter tH,R′is the time needed to hold an input to the analog sampling circuit. This parameter is small, and will not vary much as a function of process, temperature and voltage. The tV,T parameter is defined as:

t V,T =t TREF-v +t H,R′   Equation 30

[0165] Where tTREF-V is the offset of the TREF 826 signal at the last passing scan point before a fail point 906. The range of tH,R′is known from characterization, and its minimum value can be subtracted from tTREF-V. This determines tV,T. Because the difference of (tH,T−tV,T) is known from the earlier testing, the value of tH,T can be determined. During this scanning process, the CLKRREF 470 signal is set to a value that guarantees that the output of the differential comparator 854a has settled ahead of the tS,R setup time of the RTT register 852a. The capacitors of the analog sampling circuit 856 hold the sampled VP/NN voltage values during this interval. This allows the TREF 862 signal to determine the sampling point rather than the CLKRREF 470 signal (with its greater variability).

[0166] If the component allows the internal CLKRA and CLKTA clock frequencies to be an integer multiple of the CLKRREF 470 and CLKTREF 472 clock frequencies (not shown in this example), then more than one bit can be transferred between transmitter and receiver during each tCYCLE interval. In this case, it is likely that the TREF 862 signaling rate will be limited to the same rate as the CLKRREF 470 and CLKTREF 472 clock signals; a rising edge and a falling edge in each tCYCLE interval. In this case, scanning the TREF 862 signal can check only one of the output-eye regions in each tCYCLE interval. In this situation, masking logic can be added to the compare logic of the test circuit 840A2.

[0167] The masking logic will allow only one bit in each tCYCLE interval to be checked, and the rest ignored. However, the other bits of the interval can be checked (one at a time) by adjusting the masking logic and the offset of the TREF 862 signal.

[0168] For example, if four bits are transferred in each tCYCLE interval, then the scanning process described for FIG. 9 is repeated four times. During each of the four iterations, a different one out of every four of the bits RTT (from the register 852a) would be compared in test circuit 840A2. Between each iteration, the tTREF-Q value would be shifted by tCYCLE/4. In this way, every bit of every pattern is checked to ensure that the worst case tV,T value is found. And this can be done without high-speed signals generated by a high-speed tester; the tester is required to only generate signals with timing events that are accurately positioned. This will be described in more detail with FIG. 23B.

[0169] Uncoupled Receiver Parameter Test—Time Scanning

[0170] FIG. 10 shows the timing waveforms used for determining the receiver timing parameters using the embodiment of FIGS. 8A-B. It is similar to FIG. 5, except that both the TREF signal and CLKTREF signal are placed at nominal values so that they do not interfere with the determination of the receiver timing parameters. The CLKRREF 470 signal is scanned in order to determine the receiver parameters. The CLKTREF 472 signal generates an output-eye-from-component-transmitter region 1002 whose left boundary is at time tQ,T after the rising edge of CLKTREF 472, and whose right boundary is at time tV,T after the next rising edge of CLKTREF 472. These two boundaries are scanned by the TREF 862 signal to determine the actual receiver input eye 1004.

[0171] First, the rising edge of the TREF 862 signal is adjusted to occur at a time tTREF-SH after the rising edge of the CLKTREF 472 signal. This is a point within the specification-output-eye 1001. The falling edge of the TREF 862 signal is adjusted to occur within the specification-output-eye 1001. The CLKRREF 470 signal is adjusted to occur at a time tCLKTR-S after the rising edge of the CLKTREF 472 signal. Then the following equation will be true:

t TREF-SH +t Q,R′ =t CLKTR-S −t S,R   Equation 31

or

t CLKTR-S =t TREF-SH +t Q,R′ +t S,R   Equation 32

[0172] Where tQ,R′ is the output delay of the analog sampling circuit 856 (the maximum time from the rising edge of TREF to VP′/VN′ valid). This parameter is relatively small, and will not vary much as a function of process, temperature and voltage. Then the tCLKTR-S value is scanned (decreasing values) from the passing time (tTREF-SH tQR′+tS,MIN) to a point 1006 just past time (tTREF-SHtQR′ +tS,R) where it fails. At the last passing value of tCLKTR-S the tS,R parameter will be equal to:

t S,R =t CLKTR-S −t TREF-SH −t Q,R′   Equation 33

[0173] The range of tQ,R′ is known from characterization, and its minimum value can be subtracted from tCLKTR-S-tTREF-SH. This determines tS,R. Because the sum of (tS,R+tQ,T) is known from the earlier testing, the value of tQ,T can be determined.

[0174] During this scanning process, the rising edge of the TREF 862 signal is set to a value that guarantees that the output (VP′/VN′) of the analog sampling circuit 856 has settled after the input (VP/NN) is known to be valid. The capacitors of the analog sampling circuit hold the VP′/VN′ voltage at their previous values, prior to the rising edge. This allows the TREF 862 signal to determine the driving point rather than the CLKTREF 472 signal (with its greater variability). If the component allows the internal CLKRA and CLKTA clock frequencies to be an integer multiple of the CLKRREF 470 and CLKTREF 472 clock frequencies (not shown in this example), then more than one bit can be transferred between transmitter and receiver during each tCYCLE interval. In this case, it is likely that the TREF 862 signaling rate will be limited to the same rate as the CLKRREF 470 and CLKTREF 472 clock signals; a rising edge and a falling edge in each tCYCLE interval. In this case, scanning the TREF 862 signal can check only one of the output eye regions in each tCYCLE interval. In this situation, masking logic can be added to the compare logic of the test circuit 840. The masking logic will allow only one bit in each tCYCLE interval to be checked, and the rest ignored. However, the other bits of the interval can be checked (one at a time) by adjusting the masking logic and the offset of the TREF 862 signal.

[0175] For example, if four bits were transferred in each tCYCLE interval, then the scanning process described for FIG. 10 would be repeated four times. During each of the four iterations, one out of every four of the bits RTT register 852a would be compared in the test circuit 840A2. Between each iteration, the tTREF-SH value would be shifted by tCYCLE/4. In this way, every bit of every pattern could be checked to ensure that the worst case tS,R value was found. This can be done without high-speed signals generated by a high-speed tester; the tester is required to only generate signals with timing events that are accurately positioned. This will be described in more detail with FIG. 23B.

[0176] Referring now to the lower portion of FIG. 10, finding the tH,T value using a similar scanning process is illustrated. The rising and falling edges of the TREF 862 signal remain in their same places within the specification-output-eye 1001. However, since the hold time tH,T of the receiver RT is to be evaluated, the rising edge of the CLKRREF 470 signal is adjusted to occur at approximately a time tH-MIN before the next rising edge of the TREF 862 signal. The right boundary of the input-eye 1004 of receiver RT is to hold the output (VP/NN) of the analog sampling circuit 856 before it is replaced by the new data after the next rising edge of TREF 862. Then the following equation will be true:

t CLKTR-H +t H,R =t TREF-SH +t CYCLE +t V,R′   Equation 34

or

t CLKTR-H =t TREF-SH +t CYCLE +t V,R′ −t H,R   Equation 35

[0177] Then the tCLKTR-H value is scanned (increasing values) from the passing time (tTREF-SH+tCYCLEtV,R′−tH,MIN) to point 1008 just past the time (tTREF-SH+tCYCLE+tV,R′−tH,T) where it fails. Here the parameter tV,R′is the time that the output (VP′/NN′) of the analog sampling circuit 856 remains valid after the rising edge of TREF 862. This parameter is relatively small, and will not vary much as a function of process, temperature and voltage. The tH,T parameter will be equal to:

tH,T=tTREF-SH+tCYCLE+tV,R′−tCLKTR-H  Equation 36

[0178] Where tCLKTR-H is the offset of the CLKRREF 470 signal at the last passing scan point. The range of tV,R′ is known from characterization, and its maximum value can be added to tTREF-SH+tCYCLE-tCLKTR-H. This determines tH,T. Because the difference of (tH,T−tV,T) is known from the earlier testing, the value of tV,T can be determined.

[0179] During this scanning process, the CLKTREF 472 signal has been set to a value that ensures that the output VP′/VN′ of the analog sampling circuit is valid during the input-eye 1004 of the RT register. The capacitors of the analog sampling circuit 856 hold the VP′/VN′ voltage values during much of this interval. This allows the TREF 862 signal to determine the driving point rather than the CLKTREF 472 signal (with its greater variability).

[0180] If the component allows the internal CLKRA and CLKTA clock frequencies to be an integer multiple of the CLKRREF 470 and CLKTREF 472 clock frequencies (not shown in this example), then more than one bit can be transferred between transmitter and receiver during each tCYCLE interval. In this case, it is likely that the TREF 862 signaling rate will be limited to the same rate as the CLKRREF 470 and CLKTREF 472 clock signals; a rising edge and a falling edge in each tCYCLE interval. In this case, scanning the TREF 862 signal can check only one of the output-eye regions in each tCYCLE interval. In this situation, masking logic can be added to the compare logic of test circuit 840A2. The masking logic will allow only one bit in each tCYCLE interval to be checked, and the rest ignored. However, the other bits of the interval can be checked (one at a time) by adjusting the masking logic and the offset of the TREF 862 signal.

[0181] For example, if four bits were transferred in each tCYCLE interval, then the scanning process described for FIG. 10 would be repeated four times. During each of the four iterations, a different one out of every four of the bits RT would be compared in the test circuit 840. Between each iteration, the tTREF-SH value would be shifted by tCYCLE /4. In this way, every bit of every pattern is checked to ensure that the worst case tV,T value is found. This can be done without high-speed signals generated by a high-speed tester; the tester is required to only generate signals with timing events that are accurately positioned. This will be described in more detail with FIG. 23B.

[0182] The previous figures and text have described embodiments that permit component interface circuitry to perform all of the normal high-speed testing and characterization on itself, reducing the performance requirements of the external test system significantly. The embodiments were described with reference to specific interface examples that illustrate some possible circuit enhancements that provide these benefits. Specifically, the example interface used the following circuit and signaling techniques:

[0183] Open-drain, pull-down (unipolar) drivers in the transmitter circuit

[0184] Differential signals (two interconnects per signal)

[0185] Current calibrated signaling with calibration to external reference value

[0186] No slew rate control in the transmitter circuit

[0187] Two transmitters needed for full receiver circuit test

[0188] One bit transferred per clock cycle

[0189] External reference values used for calibration of internal circuit elements

[0190] PLL or DLL used for transmit and receive clock alignment

[0191] Separate external reference signals for transmit and receive clocks

[0192] Internal termination elements with calibration to external reference value

[0193] Bidirectional signals

[0194] Point-to-point interconnect topology

[0195] Two element symbol set (bits)

[0196] Each of these circuit and signaling techniques could be modified in other embodiments to use alternate techniques. In some cases, the variation could have an effect upon the previously described circuit and procedural enhancements that improve the ability to evaluate a component. The following sections will refer to previously described embodiments and describe alternative embodiments. Any of these alternatives can be combined with any of the others to realize the advantages of the resulting embodiments.

[0197] Interface Variation—Push-Pull (Bipolar) Drivers

[0198] FIG. 11 shows one way in which the pull-down drivers of the previous examples could be converted to push-pull drivers. Note that the terms “unipolar” and “bipolar” are sometimes used to refer to these driver classifications (these terms are also used to refer to transistor classifications), but they will not be so used here. An interface 1100 is shown. The interface 1100 includes a transmitter 1102, a receiver 1104, a termination block 1106, a test circuit 1140, and a test circuit 1142. The test circuit 1142 includes an analog sampler 1156. The interface 1100 further includes additional circuitry 1101 that includes driver elements 1120 and 1122 and current source elements 1124 and 1126. The resulting push-pull driver arrangement creates a high voltage by pushing current through the resistive load of the termination block 1106.

[0199] In an alternate embodiment, the two current source elements 1124 and 1126 could be merged into a single current source element, with one terminal connecting to the high supply voltage (VDD) and the other terminal connecting to the source terminals of the two driver elements 1120 and 1122. This will be equivalent to the circuit shown in transmitter block 1101 since only one of the two driver elements 1120 and 1122 may be on at any time.

[0200] Interface Variation—Single-Ended Signals

[0201] FIG. 12 shows a single-ended signaling embodiment of a high-speed component interface 1200. The high-speed component interface 1200 includes a transmitter 1202, a receiver 1204, a termination block 1206, a test circuit 1240, and a test circuit 1242. For the single-ended signaling embodiment, circuit elements present in previous embodiments are eliminated from the transmitter 1202, the test circuit 1242, and the termination block 1206.

[0202] The transmitter 1202 includes one output driver 1220 and one current source 1224, rather than having two of each. The test circuit 1242 includes only one differential comparator, differential comparator 1254, which receives a VREF signal as well as an output of an analog sampling circuit 1256. The VREF signal is not just used for testing in this embodiment, but is also used during normal operation as a voltage reference for the receiver 1204 (which drives the RA signal) and the register 1252. The register 1252 drives the RT signal and is gated by the analog sampling circuit 1256. The analog sampling circuit 1256 is halved with respect to previously described analog sampling circuits, and includes only one transistor 1260 connected to TREF, and one capacitor 1258. The VREF signal is typically set to a value midway between VIH and VIL for normal operation, but can be adjusted by an external tester during test and characterization. The termination block 1206 includes only one termination load resistor 1234. The receiver 1204 and the test circuit 1240 are similar to those corresponding elements in previously described embodiments.

[0203] Interface Variation—Calibrated Driver Alternatives

[0204] FIG. 13 is a block diagram showing one way to implement an IOL current control, such as the IOL current control element 314 of FIG. 3. FIG. 13 is a block diagram of an embodiment of an IOL current control 1314 used to control a current source IOL 1302 like the current source elements 324 and 326 of FIG. 3. A reference current IREF is provided externally, and connects to one terminal of a load RT,NOM 1334a, which has the other terminal connected to VT. The current control 1314 includes six transistors, 1326a-1326f, with binary-scaled widths. They are connected in parallel between ground and one terminal of a second load RT,NOM 1334b, which also has the other terminal connected to VT. The voltages on one terminal of each of the RT,NOM loads are compared by a differential voltage comparator 1316. The output of the voltage comparator 1316 controls an up/down counter 1315 whose six bit value I[5:0] controls the gates of the six scaled transistors 1326a-1326b. The counter value I[5:0] is adjusted until the IREF and IOL values are approximately the same. The counter value I[5:0] is also used by other copies of the IOL current source, such as the current source that includes the six transistors 1320a-1320f. The value of the load RT,NOM is not critical, but should fall within a nominal RT range. It is not necessary to provide a calibrated termination. In fact convergence problems might result if RT control circuitry and IOL control circuitry interacted.

[0205] The value of the up/down counter 1315 is normally updated when the system is powered on. With calibration of the I[5:0] value at system initialization, the difference between IREF and IOL due to manufacturing process variations between components can be made very small. The value of the up/down counter 1315 can also be updated between intervals of normal system operation. With periodic calibration of the 1[5:0] value, the difference between IREF and IOL due to temperature and supply voltage variations can also be made very small. Other alternative circuits are possible. For example, the current source IOL 1302 element could be a single transistor whose gate voltage is varied to adjust its current value. Or the IOL elements could be built from PMOS transistors instead of NMOS transistors, as shown. The reference value could be a resistance value or a voltage value instead of a current value. These alternative circuits would all share the following characteristics, however. An external reference value is provided, control circuitry adjusts a generated value to match the reference value, and the generated value can be replicated throughout the component.

[0206] In another embodiment, adjusting the current through the output driver structure involves changing the voltage on the gate of the calibrated device rather than turning on a subset of device segments (scaled transistors) with a digital control value, as shown in FIG. 13. An example of such an embodiment is shown in FIG. 14. FIG. 14 shows a current source 1424 controlled by a control circuit 1402. The IOL control circuit 1402 includes an up/down charge pump 1404, load devices 1407a and 1407b, a comparator 1406, transistors 14081410, and 1412, and a capacitor 1414. A control value Vc is maintained in the form of an analog voltage, which changes the amount of current the calibrated device 1420 can sink. This is an alternative to a six bit digital value I[5:0] that indicates which of several binary weighted transistors are on. This control voltage approach could be used with either an external current reference or an external voltage reference.

[0207] The use of a calibrated output driver with an external reference value is one of the features that permit receiver and transmitter circuits of a component to test one another. Changing the external reference value changes the output voltage levels (and output current levels) of the driver, and permits the various voltage parameters of the transmitter and receiver to be tested. In previously described embodiments, a reference current IREF is used to specify the IOL sink current of an output driver. FIG. 15 shows an alternate embodiment in which a VOL output voltage value is used as a reference value (VOL-REF). This may be an easier reference signal to generate in a test environment, particularly since it must be swept across a range of values for receiver testing. FIG. 15 is a block diagram of pertinent circuit elements of such an embodiment. FIG. 15 shows an output driver structure 1502 that includes a voltage source and an output driver. The output driver structure 1502 is coupled to a VOL control circuit 1504.

[0208] The control circuit 1504 includes six transistors, 1522a-1522f, with scaled widths. They are connected in parallel between ground and one terminal of a resistive load RT,NOM 1534, by way of transistor 1536, which has the other terminal connected to VT. The voltage on one terminal of each of the resistive load RT,NOM 1534 load and an externally supplied voltage VOL-REF are compared by a voltage comparator 1532. The output of the voltage comparator 1532 controls an up/down counter 1530 whose six bit value V[5:0] controls the gates of the six scaled transistors 1522a-1522f. The counter value V[5:0] is adjusted until the VOL-REF and VOL values are approximately the same. The counter value V[5:0] is also used by other copies of the VOL current source, such as the current source that includes the six transistors 1540a-1540f. The value of the resistive load RT,NOM 1534 is not critical, but should fall within a nominal RT range. It is not necessary to provide a calibrated termination. In fact convergence problems might result if RT control circuitry and VOL control circuitry interacted.

[0209] The value of the up/down counter 1530 is normally updated when the system is powered on. With calibration of the V[5:0] value at system initialization, the difference between VOL-REF and VOL due to manufacturing process variations between components can be made small. The value of the up/down counter 1530 can also be updated between intervals of normal system operation. With periodic calibration of the V[5:0] value, the difference between VOL-REF and VOL due to temperature and supply voltage variations can also be made small. Other alternative circuits are possible. For example, the VOL elements (transistors 1540a-f and 1522a-f) could be a single transistor whose gate voltage is varied to adjust its current value. Or the VOL elements (transistors 1540a-f and 1522a-f) could be built from PMOS transistors instead of NMOS transistors, as shown. The reference value could be a resistance value or a current value instead of a voltage value.

[0210] FIG. 16 is a block diagram of an embodiment of an RT control, such as the RT control element 336 of FIG. 3. The RT control 1636 is coupled to an adjustable resistive load arrangement 1635 consisting of a termination load RT and a termination voltage VT. The RT control 1636 receives an externally supplied reference load RREF, which has one terminal connected to VT and the other connected to a current source 1637 of value IOL,NOM. The termination load RT consists of six transistors, 1644a-1644f, with scaled widths. They are connected in parallel between VT and a second current source 1639 of value IOL,NOM. A voltage comparator 1640 compares the voltages on one terminal of RREF and RT. The output of the voltage comparator 1640 controls an up/down counter 1642 whose six bit value R[5:0] controls the gates of the six scaled transistors 1644a-1644f. The counter value R[5:0] is adjusted until the RREF and RT values are approximately the same. The counter value R[5:0] is also used by other copies of the RT termination load, such as the transistors 1645a-1645f of the load arrangement 1635.

[0211] The value of the current source IOL,NOM is not critical. The current source need not be a calibrated current source. In fact, a calibrated current source may cause convergence problems if the RT control circuitry and IOL control circuitry interacted. It is only necessary to generate a current value that falls within the nominal IOL range. The value of the up/down counter 1642 is normally updated when the system is powered on. With calibration of the R[5:0] value at system initialization, the difference between RREF and RT due to manufacturing process variations between components is made very small. The value of the up/down counter 1642 can also be updated between intervals of normal system operation. With periodic calibration of the R[5:0] value, the difference between RREF and RT due to temperature and supply voltage variations is also made very small.

[0212] Other alternative circuits are possible. For example, the RT element could be a single transistor whose gate voltage is varied to adjust its resistive value. The RT element could also be built from PMOS transistors instead of NMOS transistors. The reference value could be a current value or a voltage value instead of a resistance value. The reference value could consist of a length of interconnect with termination loads. The RT element could be turned on and off during normal operation to reduce power, or to eliminate termination at the transmitter end of the interconnect. These alternative circuits would all share the following characteristics, however. The external reference value is provided, control circuitry adjusts a generated value to match the reference value, and the generated value can be replicated throughout the component.

[0213] Referring briefly to FIG. 13, a calibrated output driver consists of an upper switch device (the transistor 1324, which is designed to be as similar as possible to the driver elements 320 and 322 of FIG. 3), and a set of six weighted devices (the transistors 1326a-1326f) forming a lower calibration device. In an alternative embodiment, AND gating of the switch device and the lower calibration device is performed in a predriver instead of the driver. FIG. 17 is a block diagram of such an alternative embodiment of a calibrated output driver 1702 and an IOL control circuit 1720. The output driver 1702 includes a predriver 1704 coupled to a driver 1706. The driver 1706 consists of a set of six weighted transistors, 1706a-1706f, forming a calibrated device. Transistors 1706a-1706f are enabled by six AND gates, 1716a-1716f, in the predriver 1704. A transmit bit TA′ is one of the inputs to each of the AND gates 1716a-1716f. The other input to each of the AND gates 1716a-1716f is one of the six bits I[5:0] forming an IOL control value. The IOL control value is generated as previously described, with a set of six weighted devices 1714a-1714f, termination values 1712a and 1712b, an external reference value IREF, a voltage comparator 1710, and an up/down counter 1708.

[0214] Another way in which the output driver can be varied in other embodiments is in the selection of gate and drain voltages used. The gate voltages of the driver devices are usually the supply voltage VDD used for all the digital circuits on the component (an exception to this is when an analog control voltage VC is used for driver calibration). The drain voltage of the upper device is usually at a value of VT (the termination voltage) as the driver begins conducting. The values of VDD and VT will determine the part of the transistor characteristic (the graph of IDS versus VDS for different VGS values) used by the devices. When VDD is larger than VT, the transistor will operate in the linear region with a lower source impedance. When VDD is smaller than VT, the transistor will operate in the saturated region with a higher source impedance. A zero source impedance is called voltage-mode signaling, and an infinite source impedance is called current-mode signaling. In the embodiments previously described, with point-to-point interconnect topology and termination present at both ends of the termination; this distinction is unimportant to first order. However, if the termination value and the interconnect impedance value are not well matched, the source impedance can become important when assessing the accumulation of reflections. Also, if multi-drop interconnect topology is used, or if termination is not present at each end of an interconnect, the source impedance can also become an important factor. The variations described for implementing a calibrated output driver could also be applied to a calibrated termination load.

[0215] Clock Aligner Circuit Variations

[0216] The clock aligner elements 312 and 328 of FIG. 3 may be implemented in various ways in different embodiments. FIG. 18 is a block diagram of one embodiment of a clock aligner circuit 1812 implemented using a delay-locked-loop (“DLL”). The clock aligner circuit 1812 includes a phase detector 1802, a filter 1804, a voltage controlled delay line 1806 with delay tD, and a buffer 1808. The clock aligner circuit 1812 compares the reference clock CLKXREF with the generated clock CLKXA and delays the generated clock until they have the same phase. In this way, the delay of the buffer tB can be effectively eliminated, although there will be some range of timing skew (tSKEW-X,MAX−tSKEW-X,MIN) between the reference clock CLKRXREF and the generated clock CLKXA because of the intervening circuitry.

[0217] The timing equation that is satisfied will be:

t D +t B =N*t CYCLE N ={1,2, . . . }  Equation 37

[0218] where N is usually chosen to be one, and tCYCLE is the time interval between successive timing events (a rising edge, for example) on the reference clock CLKRXREF.

[0219] FIG. 19 is a block diagram of one embodiment of another clock aligner circuit 1912 implemented using a phased-locked-loop (“PLL”). The clock aligner circuit 1912 includes a phase detector 1902, a filter 1904, a voltage controlled oscillator 1906, and a buffer 1908. The clock aligner circuit 1912 compares the reference clock CLKXREF with the generated clock CLKXA, and adjusts the generated clock until they have the same frequency and phase. In this way, the delay tB of the buffer 1908 can be effectively eliminated, although there will be some range of timing skew (tSKEW-X,MAX−tSKEW-X,MIN) between the reference clock CLKRXREF and the generated clock CLKXA because of the intervening circuitry.

[0220] Either the DLL clock aligner circuit 1812 or the PLL clock aligner circuit 1912 may be used to implement the receive clock aligner and the transmit clock aligner circuit blocks in the embodiments described. The PLL clock aligner circuit 1912 might be chosen if the reference clock is not the same frequency as the generated clock, because the PLL can perform frequency multiplication at the same time that it is aligning the generated clock to the reference clock. For example, the generated clock CLKXA could have four times the frequency of the reference clock CLKRXREF, permitting four bits to be transferred in each tcYCLE interval.

[0221] The DLL clock aligner circuit 1812 might be chosen if the generated clock needed a fixed offset relative to the reference clock. This can be done by tapping the delay line at intermediate points. For example, the CLKRTA clock can be adjusted so its rising edge is +180 degrees relative to the rising edge of CLKRTREF in the transmitter.

[0222] The PLL and DLL circuit features could also be merged into one structure, permitting both frequency multiplication and phase adjustment. Such an embodiment will be discussed further below.

[0223] Interface Variation—Output Register/Predriver/Driver Loop Compensation

[0224] FIGS. 18 and 19 illustrated embodiments of clock aligner circuits. A transmit clock aligner circuit 1812 or 1912 includes a clock buffer delay in the feedback loop. In another embodiment, the delay of the register, the predriver, and driver circuits in the transmitter (316, 318, and 320/322 in FIG. 3, for example) can also be included in the feedback loop so that process, supply voltage, and temperature variations can be tracked out. FIG. 20 is a block diagram of such an embodiment of a transmit clock aligner circuit 2012. The transmit clock aligner circuit 2012 includes a phase detector 2002, a filter 2004, a voltage controlled oscillator (“VCO”) 2006, buffers 2008a and 2008b, a register 2010, a predriver 2014, and a driver 2016. The output of the VCO 2006 is connected to the two buffers 2008a and 2008b. The buffer 2008a supplies CLKTA to the transmitter, and the buffer 2008b provides an equivalent delay to a feedback signal. The register 2010 provides a delay that is equivalent to the register 316 clock-to-output delay. The predriver 2014 provides a delay that is equivalent to the predriver element 318 delay. The driver 2016 provides a delay that is equivalent to the driver elements 320/322 delay. The first order delay of the entire path from the CLKTREF signal to the VP/NN output signal is compensated in the feedback path, reducing the variation in the tV-T and tQ-T timing parameters.

[0225] A similar modification could be made to the clock aligner for the receiver circuit, including the delays that are equivalent to the delays of the voltage comparator and input register in the feedback loop. This would reduce the variation in the tS-R and tH-R timing parameters.

[0226] Interface Variation—Slew Rate Control

[0227] FIG. 21 is a block diagram of an embodiment of a high-speed interface 2100. The high-speed interface 2100 is a variation of the high-speed interface 300 of FIG. 3. Where circuit elements and signals of the interface 2100 are similar to corresponding circuit elements and signals of the interface 300, the circuit elements and signals are not numbered or discussed. The interface 2100 includes an additional calibration element, slew rate control circuit 2111, in the output path. The slew rate control circuit 2111 adds the ability to control the predriver strength with an external reference signal SREF 2161, and thus control the slew rate of the driver. The slew rate is a measure of the rate of change (in volts per second) of the output voltage as the transmitter changes the value of the bit it is driving (high voltage to low voltage or the reverse). Controlling the slew rate is an important aspect of high-speed signaling. By controlling the slew rate, the higher frequency spectral components of the output signal are limited, allowing the signaling rate to approach the limit of the interconnect media more closely.

[0228] The slew rate control 2111 can be implemented using segmented devices, as previously described, for example with reference to the termination calibration and FIG. 16. An external reference value (a resistance, for example) could be used with a voltage comparator and an up/down counter that generates a control value, which changes the number of predriver segments that are enabled.

[0229] Interface Variation—Test Current Source

[0230] In the previous receiver test examples, when the VIH and VIL receiver parameters are checked, it is assumed that the two interface circuits that are coupled by the interconnect are bidirectional with pull-down drivers. This allows the two transmitter circuits to be set the to VIH and VIL parameters independently. In some cases with this arrangement, it is not possible to vary the high output level for checking the VIH parameter. For example, this is true in the case of a single bidirectional interface circuit testing itself. This is also true in the case of two unidirectional interface circuits (one a transmitter and one a receiver) coupled by an interconnect for testing each other.

[0231] In another embodiment, shown in FIG. 22, circuitry is added to the interface 2200 to allow a single bidirectional interface circuit to test itself, and to allow two unidirectional interface circuits (one a transmitter and one a receiver) coupled by an interconnect to test each other. The high-speed interface 2200 is a variation of the high-speed interface 300 of FIG. 3. Where circuit elements and signals of the interface 2200 are similar to corresponding circuit elements and signals of the interface 300, the circuit elements and signals are not numbered or discussed. The interface 2200 includes a test circuit 2202 and a test current reference signal ITL-REF 2261 input to ITL test current control 2204. The test circuit 2202 is similar to the calibrated output driver structure previously described with reference to other embodiments. The test circuit 2202 includes drivers 2206 and 2208, and current sources 2210 and 2212. The ITL-REF 2261 is externally supplied, and sets the sink current ITL in both legs of the test circuit 2202. In effect, this duplicates the action of the interface 2200 for the VIH test (for example see the description of FIGS. 6 and 7 for a description of a VIH test).

[0232] An alternative embodiment uses a low bandwidth test system to vary the VT termination voltage for the VIH test. The VOH level produced by the transmitter is determined by the termination voltage value, which is made adjustable in some test systems.

[0233] In the case in which push-pull drivers are used in the transmitter, a receiver can be tested with a single transmitter without the test circuit 2202 or an adjustable VT. The push-pull driver calibration circuitry (with two external reference signals) permits both the VOH and VOL levels to be adjusted, allowing both VIH and VIL receiver parameters to be tested.

[0234] Interface Variation—Multiple Bits per Clock Cycle

[0235] In the embodiments described so far, one bit was transferred in each clock cycle. In many applications more than one bit is transferred in each clock cycle. In these applications, external clock signals provide a frequency reference and a static phase reference for the clock circuitry on the component. FIGS. 23A and 23B illustrate an embodiment in which more than one bit is transferred in one clock cycle. FIG. 23A is a block diagram of a clock aligner circuit 2300 that is used with a receiver and a transmitter. The clock aligner circuit 2300 is a PLL that includes a phase detector 2302, a filter 2304, and a VCO 2306. A signal from the VCO 2306 is coupled to two frequency divider circuits, 2308 and 2310. In this embodiment, the frequency divider circuit 2308 reduces the frequency of the VCO output by a factor of two and the frequency divider circuit 2310 reduces the frequency of the VCO output by a factor of eight. The outputs of the frequency divider circuits 2308 and 2310 are coupled to respective buffers 2312 and 2314. The buffered ½ frequency signal becomes the CLKXA clock signal used by the transmitter or receiver. The buffered {fraction (1/8)} frequency signal becomes the feedback that is compared to the external CLKXREF clock signal. This forces the VCO 2306 output to run at eight times the frequency of the CLKXREF clock signal, and the CLKXA clock signal to run at four times the frequency of the CLKXREF clock signal.

[0236] FIG. 23B is a timing diagram that shows the clock signals from the clock aligner circuit 2300 plotted together with the VP/NN signals 2301 and 2301R for both a transmitter circuit and a receiver circuit. In the case of the CLKTA 2322 and CLKTREF 2320 clock signals, the rising edges are closely aligned (within the specified error of the clock circuitry). Each of four successive output eyes 2330-2333 are defined around timing points at tCYCLE/4 increments across one tCYCLE 2334 interval. One tCYCLE 2334 interval is one cycle of CLKTREF 2320. The tV,T and tQ,T timing parameters are defined with respect to these timing points.

[0237] Similarly, in the case of the CLKRA 2422R and CLKRREF 2420R clock signals, the rising edges are closely aligned (within the specified error of the clock circuitry). Each of four successive input eyes 2430-2433 are defined around timing points at tCYCLE/4 increments across one tCYCLE 2434 interval. One tCYCLE 2434 interval is one cycle of CLKTREF 2320. The tSR and tH,T timing parameters are defined with respect to these timing points.

[0238] Interface Variation—Adjustable Clocking, Termination, and Current Elements

[0239] The clocking calibration elements in the previously described embodiments used a value on an external reference signal to set their operating points. For example, with reference to FIG. 3, an IREF 366 signal, an RREF 368 signal, a CLKRREF 370 signal, and a CLKTREF 372 signal are used to set respective operating points. In another embodiment, an alternative approach is used. FIG. 24 is a block diagram of an embodiment of an interface 2400 in which an adjustment value is provided in addition to a reference value to set an operating point. In the embodiment of FIG. 24, circuitry is added to the interface 2400 to provide adjustment values. The high-speed interface 2400 is a variation of the high-speed interface 300 of FIG. 3. Where circuit elements and signals of the interface 2400 are similar to corresponding circuit elements and signals of the interface 300, the circuit elements and signals are not numbered or discussed. The interface 2400 includes a receiver 2404, a transmitter 2402, a termination block 2406, a test circuit 2440, and a test circuit 2442. The interface 2400 further includes adjustment circuitry added to four calibration elements. The four calibration elements are a receive clock aligner 2428, a transmit clock aligner 2412, an IOL current control 2414, and an RT control 2436. The adjustment value received by each of the calibration elements is a digital value that is held in a respective register and passed to the respective calibration element. The registers include a CLKRA adjust register 2409, a CLKTA adjust register 2405, an IOL adjust register 2403, and an RT adjust register 2407.

[0240] FIG. 25 is a block diagram that illustrates how an adjustment value from a CLKXA adjust register 2505 is used by a transmitter or receiver clock aligner 2529. The clock aligner 2529 includes merged PLL and DLL circuits, including a phase detector 2502, a filter 2504, and a VCO 2506. The clock aligner 2529 further includes a voltage-controlled delay line 2512, a filter 2508, a phase detector 2510, and buffers 2514. The clock aligner 2529 further includes a phase selector 2516, and a phase interpolator 2518, each of which is coupled to the CLKXA adjust register 2505. The PLL circuit receives an external clock signal CLKXREF. The output of the VCO 2506 is fed into the voltage controlled delay line 2512 with 2N delay elements. The delay of each of the delay elements is “1*Δt” delay. The VCO 2506 signal and the 2N delay taps drive 2N+1 buffers 2514. The buffer 2514 with “0*Δt” delay is fed back to the phase detector 2502 of the PLL, causing this signal to lock to the external clock signal CLKXREF. The buffer 2514 with “0*Δt” delay and the buffer 2514 with “2N*Δt” delay are fed to the phase detector 2510 of the DLL. This causes the total delay of the delay line (2N*Δt) to adjust to one clock cycle. The 2N delay taps now provide 2N clock signals whose phase is offset by increments of 360°/2N. The CLKXA adjust register 2505 selects one of these 2N clock signals with an N bit field with multiplexing logic in a phase selector 2516. An M bit field from the CLKXA adjust register 2505 specifies an interpolation value between this clock signal and the next higher clock signal.

[0241] This improves the phase resolution of the output clock CLKXA from the phase interpolator 2518.

[0242] In other embodiments, the delay of the phase selector 2516 and the phase interpolator 2518 can be modeled in a delay circuit that is inserted in the feedback loop for the PLL to reduce the error in the CLKXA signal due to process, supply voltage, and temperature variations.

[0243] When the phases of the transmitter and receiver clocks are adjustable, the phase of the external clock reference signal CLKXREF becomes unimportant as long as it remains stable. The phase of the internal CLKXA clock signal can be set to the proper value by an initialization process prior to normal system operation. The phase can also be modified by an update process between intervals of normal system operation to improve the accuracy of the internal clock phase.

[0244] When the adjustable phase feature is added to the component, test and characterization of the component is improved, as is normal system operation. In embodiments that allow the phase of the transmitter and receiver clocks to be adjusted by simply writing to a register, all the time scanning procedures described with reference to previous embodiments can be performed via this mechanism. Therefore, it is not necessary for a test system to adjust the phase of the external clock reference signals. The external clock reference signals serve simply as a frequency source and a static phase reference. This reduces component testing costs, and consequently, component manufacturing costs.

[0245] The benefit just described is also available to a component in which either one of the two clock aligner circuits can also be adjusted. All of the time scanning tests require that one of the external clock signals be held at a fixed value while the other is scanned through a range of values. The use of the clock adjustment circuitry with just one of the clock aligners permits all the time scanning procedures to be performed. For example, an internal receiver clock CLKRA can be phase and frequency locked to an external signal CLKRREF using a PLL or DLL clock aligner with no phase adjustment. An internal transmitter clock CLKTA can be frequency locked to the external signal CLKTREF, but can have an adjustable phase by using a clock aligner/adjuster block such as the one shown in FIG. 25. This permits the phase difference between the internal receiver clock and internal transmitter clock to be set to any arbitrary value (while the external clock signals are at static phase values), but only requires the phase selector and phase interpolator for the transmitter clock.

[0246] This benefit can also be extended to use of the TREF signal for uncoupling the transmitter and receiver parameters as detailed in FIGS. 9 and 10. In both cases, a TREF edge must be moved relative to CLKRA or CLKTA. This can be accomplished by keeping the TREF edge fixed, and moving CLKRA or CLKTA using the adjustment mechanism just described. This is particularly useful because the TREF must be carried directly from the test system to the analog sampler in the interface in order to minimize timing variation; the TREF cannot use alignment circuitry or adjustment circuitry like CLKR or CLKTA. With the adjustment feature present in either the transmit or receive circuit (or both) it is possible to perform all the time-scanning procedures needed for uncoupling the transmitter and receiver parameters.

[0247] FIG. 26 is a block diagram showing how an RT adjust register 2607 is used with an RT control element 2636 in one embodiment. RT control element 2636 is similar to the RT control element 336 of FIG. 3. However, in the embodiment of FIG. 26, the R[5:0] termination control value that is passed to the RT termination load 2634 is now passed through a digital multiply block 2638 in the calibrated resistive load RT (such as 1635 of FIG. 16 or 334a and 334b of FIG. 3). The R[5:0] value is interpreted as an unsigned binary number of the form “xxxxxx.”, and is multiplied by the value in the RT adjust register 2607, which is interpreted as an unsigned binary number of the form “xxx.xxx”. The upper 3 bits and the lower 3 bits of the product are discarded, the middle six bits are interpreted as an unsigned binary number of the form. “xxxxxx.”, and are passed to the RT termination load 2634. This permits the R[5:0] termination control value to be scaled by the adjustment value. For example, assume that an external RREF value produces a R[5:0] termination control value of “000110.” If the value in the RT adjust register 2607 is “010.000”, then the control value used by the RT termination load 2634 is “001100.” This causes the real RT termination load 2634 to have about half the resistance of the dummy RT termination load 2635 in the RT control element 2636, and about half the resistance of the external RREF load, assuming that the calibration mapping is linear.

[0248] If the calibration mapping is not linear, then a calibration process is performed. In one embodiment, this consists of measuring and comparing internal signals. For example, the voltage dropped across the RT termination load 2634 is compared to the voltage dropped across the dummy RT termination load 2635. Measuring each voltage with a voltage comparator against a scanned external voltage reference signal VREF performs the measurement process. Alternatively, if the calibration mapping is not linear, it may not be necessary to actually determine the value of the voltage dropped across the RT termination load 2634. It may be sufficient to use a calibration process to measure the effect of adjusting the RT termination load 2634 upon a related parameter of the receiver or transmitter (like VOL). This allows the RT termination load 2634 to be optimally adjusted without determining what its actual value is. It would not be necessary to precisely match the RREF external reference value to the impedance of the interconnect in either the actual system environment or in the testing environment.

[0249] The flexibility of adjusting a calibrated element like the RT termination load 2634 prior to normal operation or prior to test and characterization has beneficial effects. The matching of external reference values is not critical, and the external reference values do not need to be adjusted. Writing to internal registers of the component can perform matching and adjustment. This reduces component testing costs, and consequently, component manufacturing costs. It also means that writing to an internal register of the component rather than changing a reference value supplied by the test system needed for test and characterization can carry out the scanning process. This can simplify the test system requirements and reduce cost.

[0250] FIG. 27 is a block diagram showing how an IOL adjust register 2703 is used with an IOL control element 2714 in one embodiment. The IOL control element 2714 is similar to the IOL current control element 314 of FIG. 3. However, the I[5:0] current control value that is passed to the IOL current sink is now passed through a digital multiply block 2738 in the calibrated current source IOL (such as 1302 of FIG. 13 or current source elements 324 and 326 of FIG. 3). The I[5:0] value is interpreted as an unsigned binary number of the form “xxxxxx.”, and is multiplied by the value in the IOL adjust register 2703 which is interpreted as an unsigned binary number of the form “xxx.xxx”. The upper 3 bits and the lower 3 bits of the product are discarded, and the middle six bits are interpreted as an unsigned binary number of the form “xxxxxx.” and passed to the IOL current sink 2734. This permits the I[5:0] current control value to be scaled by the adjustment value.

[0251] For example, assume that the external IREF value produces a I[5:0] current control value of “000101.”. If the value in the IOL adjust register 2703 is “010.000”, then the control value used by the IOL current sink 2734 would be “001010”. This would cause the real IOL current sink to become about twice as great as the dummy IOL current sink 2715 in the IOL control block and to become about twice as great as the external IREF load, assuming that the calibration mapping is linear.

[0252] If the calibration mapping is not linear, then a calibration process is performed. The calibration process consists of measuring and comparing internal signals. For example, the voltage dropped across the IOL current sink 2734 is compared to the voltage dropped across the dummy IOL current sink 2715. Measuring each voltage with a voltage comparator against the scanned external voltage reference signal VREF performs the measurement process.

[0253] Alternatively, if the calibration mapping is not linear, it may not be necessary to actually determine the value of the voltage dropped across the IOL current sink. It may be enough to use a calibration process to measure the effect of adjusting the IOL current sink 2734 upon a related parameter of the receiver or transmitter (like VOL). This would allow the IOL current sink 2734 to be optimally adjusted without determining what its actual value is.

[0254] The flexibility of adjusting a calibrated element like the IOL current sink 2734 prior to normal operation or prior to test and characterization has beneficial effects. The matching of external reference values is not critical, and the external reference values do not need to be adjusted. Writing to internal registers of the component performs matching and adjustment. This reduces component testing costs, and consequently, component manufacturing costs. It also means that writing to an internal register of the component rather than changing a reference value supplied by the test system needed for test and characterization can carry out the scanning process. This can simplify the test system requirements and reduce cost.

[0255] Interface Variation—Adjustable Voltage Reference

[0256] Embodiments previously described use an external voltage reference signal VREF to measure voltages on internal nodes of a component. For example, this is done by connecting an internal node and VREF to the inputs of a voltage comparator, scanning VREF across some range, and monitoring the digital output of the comparator for when it changes from high to low or low to high. The VREF signal is assumed to directly couple from the external test system to the voltage comparator(s) to which it is coupled. This permits a voltage forcing circuit in the pin electronics of the test system to make accurate voltage measurements of internal signals. An alternative to this is to use an external voltage reference signal that remains at a fixed value. This eliminates the requirement that the test system be able to scan the voltage reference value across some range, potentially reducing the test system cost. The component includes a voltage calibration and adjustment circuit (not shown) that is similar to the calibration and adjustment arrangements of FIGS. 26 and 27, but with the difference that the voltage calibration and adjustment circuit operates on a fixed external voltage reference to create an adjustable internal voltage reference. For this embodiment, the voltage calibration and adjustment circuit should be linear, since it is assumed that the test system no longer can generate a range of accurate voltage reference values, but instead is limited to generating a single accurate voltage reference value. The adjustable internal voltage reference connects to voltage comparators to allow the voltage on other internal signals to be measured against it. The measurement process consists of writing different values to a voltage adjustment register to scan the internal voltage reference across a range of values.

[0257] In yet another alternative embodiment, an adjustable internal voltage reference is generated as described in the preceding paragraph, but instead of using a single, fixed external voltage reference, an internal voltage reference is generated using a supply voltage or a voltage reference circuit, such as a band gap reference.

[0258] All of these alternative embodiments allow a test system to trade off accuracy and flexibility with cost. The method that is used depends upon the test accuracy that is required by a component specification.

[0259] Interface Variation—Single Internal Clock Reference

[0260] Embodiments previously described use two clock reference signals (CLKTREF and CLKRREF) to indicate when a transmitter is to drive a bit and when a receiver is to sample a bit for both normal operation, and for test and characterization. This provides flexibility in setting the position of the transmitter and receiver eye regions, and also provides an easy way to scan either the transmitter or receiver in order to evaluate a timing parameter. However, some components may only have a single external clock reference CLKREF. FIG. 28 is a block diagram of an interface 2800 of such a component. The interface 2800 is similar to the interface 300 of FIG. 3 with the exception of the replacement of CLKRREF and CLKTREF signals with a single CLKREF 2871 signal.

[0261] In the case of internal clock signals for the transmitter and receiver with a fixed phase relationship a component will be unable to evaluate the timing parameters for its transmitter and receiver in the manner described with reference to FIGS. 4, 5, 6, and 7. This is true because there is no phase adjustment circuitry in such an arrangement. The test system is unable to scan a transmitter drive point to evaluate a receiver sample point, and is unable to scan a receiver sample point to evaluate a transmitter drive point.

[0262] One solution to permit testing and characterization with a single clock reference is to couple two identical components together on the load board during testing. This allows a transmitter on one component to evaluate a receiver on the other component by adjusting the external clock reference of one component relative to the external clock reference of the other component. This solution still has a transmitter timing parameter of one component coupled to a receiver timing parameter of the other component, but this may be acceptable in some situations such as when the component parameters are correlated as previously described.

[0263] Another solution to allow testing and characterization with a single clock reference is to add adjustment circuitry to at least one of the clock aligner circuits (either the transmitter or the receiver). This would permit one of the internal clock signals to be scanned relative to the other, and allow the timing parameters to be evaluated. This solution still has a transmitter timing parameter of one component coupled to a receiver timing parameter of the other component, but this may be acceptable in some situations such as when the component parameters are correlated as previously described.

[0264] Yet another solution is to provide an analog sampling circuit, such as the analog sampling circuit 2856 of FIG. 28. In such an embodiment a TREF external timing signal is scanned across a timing range, permitting the receiver and transmitter timing parameters to be evaluated without changing the position of the transmitter drive point or the receiver sample point. This is potentially the best solution since it allows scanning to be performed, and also keeps the transmitter and receiver parameters uncoupled.

[0265] Interface Variation—External Termination

[0266] The embodiments previously described show termination loads for the interconnect located within a component interface. In alternative embodiments (not shown), the termination loads are located external to the component. The test and characterization procedures described with reference to previously embodiments apply equally to embodiments with external termination loads.

[0267] Interface Variation—Unidirectional Signals

[0268] The embodiments previously described show an interface for each signal including a transmitter and a receiver, a bidirectional interface, and signals that travel in either direction. When one such bidirectional interface is coupled to another, the transmitter of one can test the receiver of the other. Alternatively, such a bidirectional interface may test itself, without being coupled to another interface. However, some components contain a unidirectional signal, in which case the interface for the signal contains either a receiver or a transmitter but not both. In a normal system environment, such an interface is usually coupled to another unidirectional interface of the opposite type. In an alternative embodiment, a single component contains unidirectional interfaces, and the test and characterization procedures previously described with reference to bidirectional interfaces can be performed by pairing each receiver interface with a transmitter interface.

[0269] FIG. 29 is a block diagram of one embodiment in which a first component 2902 is coupled to a second component 2904 through an interconnect 2912. The components 2902 and 2904 communicate using unidirectional signaling. In the embodiment shown, the component 2904 is transmitting with a transmitter 2906, and the component 2902 is receiving with a receiver 2908. For unidirectional signaling, termination loads are only required at the end of the interconnect 2912 that is opposite the transmitter 2906. The single termination load 2910 in the component 2902 is the only termination load required.

[0270] In other embodiments, both the components 2902 and 2904 include transmitters, receivers, and termination loads, but only the termination load on the receiving component is active. In this case (not shown), the two components are coupled to each other with point-to-point interconnect and are capable of bidirectional signaling. Activating only one termination load at a time reduces power consumption because the transmitting component only needs to supply half the current for the same voltage swing. An enable signal is added for the termination loads. The enable signal in a component is deasserted when the component is to transmit or, alternatively, asserted when the component is to receive.

[0271] In yet another embodiment, bidirectional signaling is used on a multi-drop interconnect to which more than two components are coupled (typically the interconnect is non-branching, and the components are placed along the length of the interconnect). If a transmitter is located in a component at either end of the interconnect, the termination load at the transmitter end can be disabled, as in previously described in the point-to-point case. If the transmitter is located in a component not at either end of the interconnect, then both termination loads (at either end of the interconnect) are be enabled to limit reflections, and the transmitter must drive twice as much current as a transmitter located at either end of the interconnect. The enable signal in the components at the ends of the interconnect are deasserted when a respective component is to transmit, otherwise it is asserted. The enable signal for a component not at the end of the interconnect is always deasserted.

[0272] Alternatively, if signaling power is not an issue, the enable signals may be set to static values in a multi-drop interconnect system. The enable signal for a termination element in a component at the ends of the interconnect is always asserted. The enable signal for a component not at the end of the interconnect is always deasserted.

[0273] Interface Variation—Analog Sampling Circuit in Transmitter

[0274] FIGS. 8A-B illustrate an embodiment with an analog sampling circuit added to the existing receiver circuit so that the individual timing parameters of the transmitter and receiver can be evaluated separately. FIGS. 9 and 10 show the results of the “uncoupled” evaluations for a transmitter and a receiver. These separate uncoupled evaluations are combined with the coupled evaluations. Coupled component evaluations are described with reference to FIGS. 4 and 5 (for coupled transmitter evaluation) and FIGS. 6 and 7 (for coupled receiver evaluation).

[0275] In an alternative to the previous embodiments that include an analog sampling circuit in a receiver, an embodiment shown in FIGS. 30A-B include an analog sampling circuit added to an existing transmitter circuit. This embodiment provides a benefit similar to the previously described embodiments, namely the ability to evaluate individual timing parameters of the transmitter and receiver. The interface 3000A1 is similar to the interface 600A1 of FIGS. 6A-B, but does not include an analog sampling circuit. Elements in FIGS. 30A-B with numbering analogous to elements in FIGS. 6A-B have similar functions.

[0276] The interface 3000A2 of FIGS. 30A-B includes a transmitter 3002 and an analog sampler 3056 between an output register 3016 and a predriver 3018 of the transmitter. The interface 3000A2 further includes drivers 3020 and 3022, current sources 3024 and 3026 and current control 3014. The interface 3000A2 further includes a termination block 3006A2 and a test circuit 3040A2. The embodiment of FIGS. 30A-B performs receiver testing as illustrated in FIG. 7.

[0277] As explained with reference to FIGS. 8A-B, in an embodiment in which the analog sampler is present with the receiver, the receiver circuit is duplicated. One receiver is active during evaluation, and the other receiver functions for normal operation. There is a cost associated with duplicating circuits on a component. In most cases, the embodiment of FIGS. 8A-B is more economical than the embodiment of FIGS. 30A-B because the transmitter components (the predriver 3018 and the drivers 3020 and 3022) are typically relatively large in area than the duplicated components in the receiver case.

[0278] Interface Variation—Additional Analog Sampler

[0279] The scanning procedures described with reference to FIGS. 5 and 7 (coupled transmitter evaluation and coupled receiver evaluation, respectively) yield the following timing parameter relationships. Here, the parameter name formats “tS,R ”, “tH,T ”, “tQ,T ” and “tV,T” have been changed to the format “tS”, “tH”, “tQ” and “tV”):

t s +t Q =t CLKTR-SQ =k 1  Equation 38

t V −t H =t CLKTR-HV =k 2  Equation 39

t s +t Q =t CYCLE-SQ −t CLKRT-SQ =k 1  Equation 40

t V −t H =t CYCLE −t CLKRT-SQ =k 1  Equation 41

[0280] The values tCLKTR-SQ, tCLKTR-HV, tCYCLE, tCLKRT-SQ and tCLKRT-HV are timing delays from the test system (the offsets between the CLKRREF and CLKTREF edges for the different cases). The four equations above are not independent. When the timing delay values are inserted equations 38 and 40 become equivalent, and equations 39 and 41 become equivalent. This equivalence is indicated by assigning the same constants (k1 and k2) for each pair of equations. FIG. 31 summarizes this relationship for a transmitter 3104 coupled to a receiver 3102. This is the case of coupled transmitter and receiver parameters.

[0281] FIGS. 9 and 10 illustrate evaluation of transmitter and receiver when the transmitter and receiver parameters are uncoupled as previously described. The scanning procedure as described with reference to FIG. 9 yields the following timing parameter relationship. The parameter name formats “tS,R” and “tH,R ” have been changed to the format “tS1” and “tH1”:

t s +t Q =t CLKTR-SQ =k 1  Equation 42

t s +t Q =t CLKTR-SQ =k 1  Equation 43

[0282] The scanning procedure as described with reference to FIG. 10 yields the following timing parameter relationship. The parameter name formats “tQ,R′” and “tV,R” have been changed to the format “tQ1” and “tv1”:

t S =t CLKTR-S −t TREF-SH −t Q1 =k 7−tQ1  Equation 44

t H =t TREF-SH +t CYCLE +t V1 −t CLKTR-H =k 8+tV  Equation 45

[0283] The values tTREF-Q, tTREF-V, tCLKTR-S, tTREF-SH, tcLKTR-H, and tCYCLE are timing delays from the test system. The values tS1, tH1, tQ1, and tV1 are the timing parameters of the analog sampling circuit. FIG. 32 illustrates this situation in which one analog sampling circuit 3256 is coupled between a transmitter 3204 and a receiver 3202. Six equations are shown, with four unknowns for the transmitter 3204 and receiver 3202 (tS, tH, tQ, and tV) and four unknowns for the analog sampler (tS1, tH1, tQ, and tV1). The equations may be rewritten:

t s +t Q =k 1  Equation 46

t V −t H =k 2  Equation 47

t s1 +t Q =k 3  Equation 48

t V −t H1 =k 4  Equation 49

t S +t Q1 =k 7  Equation 50

t V1 −tH=k8  Equation 51

[0284] There are not enough equations to solve for the unknowns. However, by adding a second sampler circuit, as shown in FIG. 33, the equations may be solved. FIG. 33 shows a transmitter 3304, a receiver 3302. Between the transmitter 3304 and receiver 3302 are coupled two analog sampling circuits 3356a and 3356b. The analog sampling circuits 3356a and 3356b have approximately the same parameters. For the embodiment illustrated by FIG. 33, this implies that the pass transistor and storage capacitors required for the analog sampling circuit 3356a larger than the corresponding transistor and capacitors for the analog sampling circuit 3356b. The transistor and capacitor values chosen take into account the capacitance presented by the receiver 3302, and the current that can be supplied by the transmitter 3304. In other embodiments, other circuit designs are possible. The design shown in FIG. 33 represents a relatively simple approach, since it only performs the time discrimination function of a receiver, and doesn't attempt to perform voltage discrimination. It records the analog voltages and passes them along, letting the receiver 3302 perform the voltage discrimination task.

[0285] If the timing parameters of the two analog sampling circuits 3356a and 3356b are made to be essentially the same, and they scale with manufacturing process, supply voltage, and temperature in the same way, then the following eight equations provide enough information to solve for the eight unknowns:

t S +t Q =k 1  Equation 52

t V −t H =k 2  Equation 53

t S1 +t Q =k 1  Equation 54

t V −t H1 =k 4  Equation 55

t S2 +t Q1 =k 5  Equation 56

t V1 −t H2 =k 6  Equation 57

t S +t Q2 =k 7  Equation 54

t V2 −t H =k 8  Equation 58

[0286] Pairs of equations are generated by choosing one transmitter/receiver pair, tightening the transfer of information for that pair, and simultaneously loosening the transfer between the other transmitter/receiver pairs so that they do not affect the determination for the pair under test. For example, when the transfer between analog sampling circuit 3356a and analog sampling circuit 3356b is tested, a TREF1 timing event and a TREF2 timing event (rising and falling edges, respectively) are adjusted so that they end at separations of (tS2+tQ1) or (tV1−tH2) for the two scanning procedures. During the scanning procedures, the separation of CLKT and TREF1 timing events (rising edges) are kept larger than the required interval of (tS1+tQ) and (tV1−tH1) so that the transfer from the transmitter 3304 to the analog sampling circuit 3356a does not affect the timing measurement. Likewise, the separation of the TREF2 and CLKR timing events (falling and rising edges, respectively) are kept larger than the required interval of (ts+tQ2) and (tV2−tH) so that the transfer from the transmitter 3304 to the analog sampling circuit 3356a does not affect the timing measurement. In this manner, the values of the timing parameters of the transmitter 3304 and the receiver 3302 may be determined. Costs associated with the embodiment include the costs of an extra (dummy) receiver circuit, two analog sampling circuits, and two interconnects carrying the TREF, and TREF2 timing reference signals.

[0287] In some applications, a single analog sampling circuit and a single timing reference signal may be adequate. In other applications, no analog sampling circuit may be needed. This depends upon the design approach used for the transmitter and receiver circuits, and upon the correlation of parameters between respective transmitter and receiver circuits.

[0288] Test Pattern Generation

[0289] A number of embodiments for testing and characterizing high-speed signaling circuits are described in the foregoing text and figures. The following sections describe hardware and methods for generating test patterns according to various embodiments.

[0290] FIG. 34 is a block diagram of a transmit interface 3400A2 and a receive interface 3400A1 according to one embodiment. The transmit interface 3400A2 includes a test circuit 3440A2 and a transmitter 3452. The receive interface 3400A1 includes a test circuit 3440A1, which will be described in more detail below, and a receiver 3450. The interfaces 3400A1 and 3400A2 may have one of several relationships to one another, including the relationships illustrated in FIGS. 35A, 35B, and 35C.

[0291] FIG. 35A illustrates one possible relationship between a receive interface 3500A1 and a transmit interface 3500A2, which are similar to the interfaces 3400A1 and 3400A2, respectively. Two components, a component 3501 and a component 3502, communicate with each other via an external interconnect. The receive interface 3500A1 and the transmit interface 3500A2 both reside on the component 3501, communicate with each other via an internal interconnect. The receive interface 3500A1 and the transmit interface 3500A2 are also coupled to the external interconnect. The external interconnect shown is not required for test and/or characterization. The two interfaces 3500A1 and 3500A2 may undergo wafer test, package test, or system test in the configuration shown. In the case of system test, there will usually be an external interconnect present, coupling to one or more additional component(s) which do not participate in the test and/or characterization.

[0292] FIG. 35B illustrates another possible relationship between a receive interface 3500A1 and a transmit interface 3500A2, which are similar to the interfaces 3400A1 and 3400A2, respectively. Two components, a component 3501 and a component 3502, communicate with each other via an external interconnect. The receive interface 3500A1 and the transmit interface 3500A2 both reside on the component 3501, and communicate with each other via an external interconnect. In this embodiment, the external interconnect is required for test and/or characterization. The two interfaces 3500A1 and 3500A2 may undergo wafer test, package test, or system test in the configuration shown. In the case of system test, the external interconnect may couple to one or more additional component(s) which do not participate in the test and/or characterization.

[0293] FIG. 35C illustrates another possible relationship between the receive interface 3500A1 and the transmit interface 3500A2. The receive interface 3500A1 resides on the component 3501 and the transmit interface 3500A2 resides on the component 3501. In this case the external interconnect is required for test and/or characterization. The two interfaces 3500A1 and 3500A2 may undergo wafer test, package test, or system test in the configuration shown. In the case of system test, the external interconnect may couple to one or more additional component(s) which do not participate in the test and/or characterization.

[0294] Variations of Pattern Source—Transmit Pattern Source (“PST”) Element

[0295] Referring again to FIG. 34, the transmit interface 3400A2 contains a transmit pattern source—(“PST”) element 3446A2. The PST element 3446A2 generates a set of patterns for test and characterization. This pattern set is a sequence of digital symbols, typically bits, which are driven by the transmit interface 3400A2 in a continuous, uninterrupted stream. There are a number of methods by which the PST element 3446A2 generates pattern sets in various embodiments.

[0296] In one embodiment, a pattern set is driven into the transmit interface 3400A2 on one or more sideband signals. These sideband signals may couple to circuitry elsewhere on the same component as the transmit interface 3400A2. Alternatively, the sideband signals may couple to circuitry that is external to the component. In the alternative case, these sideband signals will usually have a lower signaling bandwidth than the high-speed signal driven by the transmit interface 3400A2. This permits the pattern set to be provided by a low-bandwidth test system.

[0297] In another embodiment, a pattern set is driven into the transmit interface 3400A2 on one or more sideband signals, as in the previous embodiment. However, in this embodiment, some form of information compression is employed. For example, to create a pattern burst of 1024 bits with alternating one and zero bits, it may only be necessary to provide a code to indicate that the “10” pattern is repeated 512 times. Thus, a 1024 bit pattern set can be specified with only a small number of bits (16-32, for example). Using this method, a small amount of information on the sideband signals can produce a large amount of information to drive from the transmit interface 3400A2. The compression factor may be large enough to enable the slower sideband signals to keep up with the high-speed signal from the transmit interface 3400A2, permitting each pattern set to butted seamlessly with the next, and allowing an arbitrarily long pattern stream to be created.

[0298] In yet another embodiment, the pattern set is generated by circuitry in the PST element 3446A2. For example, a linear feedback shift register (LFSR) is used to produce a long pattern sequence from a small seed value. A shift register with ten shift stages can produce a pattern set that is up to 1024 bits in length. The seed value required is ten bits in size. This provides an alternative form of compression to the one already described, and has the same benefits. Namely, that if the compression factor is large enough to enable the slower sideband signal to keep up with the high-speed signal from the transmit interface, each pattern set may be butted seamlessly with the next, allowing an arbitrarily long pattern stream to be created.

[0299] In yet another embodiment, the PST element 3446A2 contains some form of non-volatile storage (such as mask ROM, Flash RAM, etc) that is always available. This storage provides a pattern set directly, and replaces the sideband signals previously described. Alternatively, this storage holds compressed pattern information, and replaces the sideband signals previously described.

[0300] Variations of Pattern Source Storage—Transmit Pattern Source Storage (“PSST”) Element

[0301] The PST element 3446A2 includes some internal read/write memory or register circuitry for pattern storage, referred to here as transmit pattern source storage (“PSST”) 3460A2. This storage permits a pattern set to be accumulated according to one of the methods described above. This accumulation may be necessary if the pattern creation process is slower than the pattern transmit process; the storage provides buffering for the two mismatched processes. The amount of storage is typically dictated by the size of the pattern set desired. This permits the pattern set to be transmitted in an uninterrupted burst.

[0302] In one embodiment, the PSST 3460A2 is dedicated to the task of buffering the pattern set. The PSST 3460A2 could alternatively be shared with other circuitry for some function during normal operation, where the other circuitry is not used during test and characterization. For example, in one embodiment, the PSST 3460A2 also functions as a write buffer in the component. A write buffer can be present in components, which are part of a system containing memory components. A write buffer typically consists of storage circuitry in the path of information being transferred to a memory component.

[0303] In another embodiment, the PSST 3460A2 also functions as a read buffer in the component. A read buffer can be present in components, which are part of a system containing memory components. A read buffer typically consists of storage circuitry in the path of information being transferred from a memory component.

[0304] In another embodiment, the PSST 3460A2 also functions as a cache in the component. A cache is storage circuitry that is used to keep a copy of information held elsewhere in the same component or in another component.

[0305] In another embodiment, the PSST 3460A2 also functions as a storage element in the transmit interface 3400A2. The normal transmit process can require some storage circuitry in some implementations. For example, the transmit data may need to go through a parallel-to-serial conversion process. This process requires storage circuitry. This storage circuitry could be adopted to also accept and store the pattern set from the PST element 3446A2 during test and characterization.

[0306] Variations of Pattern Source—Receive Pattern Source (“PSR”) Element

[0307] A receive pattern source (“PSR”) element 3446A1 in the receive interface 3400A1 generates a set of patterns for test and characterization, and is essentially identical to the PST element 3446A2 for the transmit interface 3400A2, except where explicitly noted in the following discussion. The pattern set is a sequence of digital symbols, typically bits, that are provided to a compare element 3448 in the receive interface 3400A1 in a continuous, uninterrupted stream. There are a number of methods by which the pattern set is generated by the PSR element 3446A1 in various embodiments, as described below.

[0308] In one embodiment, the pattern set is driven into the receive interface 3400A1 on one or more sideband signals. These sideband signals may be the same signals received by the transmit interface 3400A2, or they may be different. These sideband signals may couple to circuitry elsewhere on the same component. Alternatively, they may couple to circuitry, which is external to the component. These sideband signals will usually have a lower signaling bandwidth than the high-speed signal driven by the transmit interface in the alternative case. This permits the pattern set to be provided by a low-bandwidth test system.

[0309] In another embodiment, the pattern set is driven into the receive interface on one or more sideband signals, as just described. However, some form of information compression is employed. For example, to create a pattern burst of 1024 bits with alternating one and zero bits, it may only be necessary to provide a code to indicate that the “10” pattern is repeated 512 times. Thus, it is possible to specify a 1024 bit pattern set with only a small number of bits (16-32, for example). Using this method, a small amount of information on the sideband signals can produce a large amount of information to drive from the transmit interface. The compression factor may be large enough to enable the slower sideband signals to keep up with the high-speed signal from the receive interface 3400A1, permitting each pattern set to butted seamlessly with the next, and allowing an arbitrarily long pattern stream to be created.

[0310] In yet another embodiment, the pattern set is generated by circuitry in the PSR element 3446A1. For example, a linear feedback shift register (LFSR) is used to produce a long pattern sequence from a small seed value. A shift register with ten shift stages can produce a pattern set that is up to 1024 bits in length. The seed value required is ten bits in size. This provides an alternative form of compression to the one already described, and has the same benefits. Namely, that if the compression factor is large enough to enable the slower sideband signal to keep up with the high-speed signal from the transmit interface, each pattern set may be butted seamlessly with the next, allowing an arbitrarily long pattern stream to be created.

[0311] In yet another embodiment, the PSR element 3446A1 contains some form of non-volatile storage (such as mask ROM, Flash RAM, etc) that is always available. This storage provides a pattern set directly, thus replacing the sideband signals previously described. Alternatively, this storage holds compressed pattern information, and replaces the sideband signals previously described.

[0312] Variations of Pattern Source Storage—Receive Pattern Source Storage (“PSSR”) Element

[0313] The PSR element 3446A1 includes some internal read/write memory or register circuitry for pattern storage, referred to here as transmit pattern source storage (“PSSR”) 3460A1. This storage permits a pattern set to be accumulated according to one of the methods described above. This accumulation may be necessary if the pattern creation process is slower than the pattern transmit process; the storage provides buffering for the two mismatched processes. The amount of storage is typically dictated by the size of the pattern set desired. This permits the pattern set to be transmitted in an uninterrupted burst.

[0314] In one embodiment, the PSSR 3460A1 is dedicated to the task of buffering the pattern set. The PSSR 3460A1 could alternatively be shared with other circuitry for some function during normal operation, where the other circuitry is not used during test and characterization. For example, in one embodiment, the PSSR 3460A1 also functions as a write buffer in the component. A write buffer can be present in components, which are part of a system containing memory components. A write buffer typically consists of storage circuitry in the path of information being transferred to a memory component.

[0315] In another embodiment, the PSSR 3460A1 also functions as a read buffer in the component. A read buffer can be present in components, which are part of a system containing memory components. A read buffer typically consists of storage circuitry in the path of information being transferred from a memory component.

[0316] In another embodiment, the PSSR 3460A1 also functions as a cache in the component. A cache is storage circuitry that is used to keep a copy of information held elsewhere in the same component or in another component.

[0317] In another embodiment, the PSSR 3460A1 also functions as a storage element in the receive interface 3400A1. The normal transmit process can require some storage circuitry in some implementations. For example, the transmit data may need to go through a parallel-to-serial conversion process. This process requires storage circuitry. This storage circuitry could be adopted to also accept and store the pattern set from the PSR element 3446A1 during test and characterization.

[0318] Pattern Set—Transmit and Receive Process

[0319] In alternate embodiments of the present invention, steps illustrated in FIGS. 39, 41A-C, 43A-C, 45A-D, and 47A-C are carried out by hardware, software or a combination thereof. In alternate embodiments, the steps illustrated in FIGS. 39, 41A-C, 43A-C, 45A-D, and 47A-C are carried out by the components described herein.

[0320] One process for transmitting and receiving a pattern set in the embodiment of a FIG. 34 is illustrated in the flow diagram of FIG. 36. At logic block 3602, the PST element 3446A2 creates a pattern set. The pattern set is stored in the PSST 3460A2 at logic block 3604. The PSR element 3446A1 creates the same pattern set as the one stored in the PSST 3460A2 at logic block 3606, and stores it in the PSSR 3460A1 at logic block 3608. The pattern set is driven onto the interconnect at 3610, and is received in the receive interface 3400A1 at logic block 3612. The compare element 3448 compares the received pattern set and the pattern set stored in the PSSR 3460A1 at logic block 3614. At logic block 3616, the compare element 3448 indicates whether the compared pattern sets matched or not.

[0321] Variations for Compare Element

[0322] There are several variations in the way in which the compare element 3448 may be used in different embodiments. In one embodiment, the compare element couples to one or more of the sideband signals. The sideband signals may be the same signals received by the transmit interface 3400A2 or the receive interface 3400A1, or they may be different. The sideband signals may couple to circuitry elsewhere on the same component. Alternatively, the sideband signals may couple to circuitry, which is external to the component. In the alternative case, the sideband signals usually have a lower signaling bandwidth than the high-speed signal driven by the transmit interface 3400A2. The compare element 3448 drives a signal indicating a match/mismatch status onto the sideband signal(s) as each pattern set is compared. In another embodiment, the match/mismatch signal is accumulated after each pattern set, and driven onto the sideband signal(s) after two or more pattern sets have been compared.

[0323] Pattern Source (“PS”) Sharing Between Transmit and Receive Interfaces

[0324] FIG. 37 is a block diagram of an embodiment in which elements are shared between the transmit and receive interfaces. A pattern source (“PS”) element 3746 and a pattern source storage (“PSS”) element 3760 are shared between a transmit interface 3700A2 and a receive interface 3700A1. Transmit interface 3700A2 is coupled to receive interface 3700A1 by internal interconnect 3780 and an external interconnect. The receive interface includes a compare element 3748 and a receiver 3750. The transmit interface 3700A2 includes a transmitter 3752. Typically, this arrangement is possible for cases in which the transmit interface 3700A2 and the receive interface 3700A1 reside on the same component (this is the case in FIGS. 35A and 35B). When the PS element 3746 and the PSS element 3760 are shared, the area required for the circuitry is reduced.

[0325] Certain consequences of this arrangement are taken into account when the arrangement is used. For example, the pattern set should be supplied to the transmit interface 3700A2 earlier than to the receive interface 3700A1. This is a result of the time it takes for the pattern set to be driven through the circuitry of the transmit interface 3700A2, through the interconnect 3780, and through the circuitry of the receive interface 3700A1. In one embodiment, delay element 3790 is placed in the path between the PS element 3746 and the compare element 3748. The delay produced by the delay element 3790 matches the delay seen for the path that passes through the interconnect 3780. In another embodiment (not shown), the PSS element 3760 is implemented with circuitry that permits two simultaneous accesses, one for a pattern set for the transmitter 3752, and one for a pattern set for the compare element 3748.

[0326] Pattern Generation—Component with Memory Array

[0327] The discussion of the pattern generation apparatus and methods to this point is applicable to components classified as “logic” components, components classified as “memory” component, and components classified as some combination of logic and memory. For purposes of discussion herein, a memory array includes as set of storage cells in an array, surrounded by logic circuitry that permits access operations to the storage cells. Components that contain one or more such memory arrays will be called memory components. Components with no such memory array will be called logic components. Components that have one or more such memory arrays as well as significant amounts of other circuitry that does not directly support accesses to the memory array(s) will be called logic components with embedded memory.

[0328] The presence of one or more memory arrays in a memory component or a logic component with embedded memory introduces two issues that are relevant to pattern generation. First, the memory array itself needs to be tested. This memory testing is different from the testing needed for the transmit and receive interfaces of the component. It is also different from the testing needed for the logic portions of the component, if logic circuitry is present. Second, the memory array provides a storage resource that is not available in a logic-only component. This resource means that a pattern source storage element (for example PSST 3460A1 or PSST 3460A2 in FIG. 34) can reduced in size. For example, the size of the pattern source element may be reduced to only the size of a buffer between the pattern source element and the memory array(s).

[0329] Transmit and Receive Interfaces with Shared Memory Array

[0330] FIG. 38 is a block diagram of an embodiment with a memory array element incorporated into a pattern generation process similar to those already described with reference to FIGS. 34-37. In the embodiment of FIG. 38, a PS element 3846 and a memory array (“MA”) element 3871 are included in a pattern/memory (“PM”) element 3870 that is shared between a transmit interface 3800A2 and a receive interface 3800A1. This transmit interface 3800A2 includes a transmitter 3852. The receive interface 3800A1 includes a receiver 3850, a compare element 3848, and a delay element 3890. The transmit interface 3800A2 and the receive interface 3800A1 reside on the same component, as in FIGS. 35A and 35B.

[0331] In a similar fashion to the PS elements of FIGS. 34 and 37, the PS element 3846 generates a set of patterns for test and characterization. The pattern set is a sequence of digital symbols, typically bits, which are driven by the transmit interface in a continuous, uninterrupted stream. In various embodiments, the pattern set is generated by various methods by the PS element 3846. The various methods are similar to the methods previously described with reference to FIGS. 35A and 35B, and are controlled by a set of sideband signals. The sideband signals may couple to circuitry elsewhere on the same component. Alternatively, the sideband signals may couple to circuitry external to the component. The sideband signals typically have a lower signaling bandwidth than the high-speed signal driven by the transmit interface in the alternative case. This permits the pattern set to be provided by a low-bandwidth test system.

[0332] The PS element 3846 includes PSS element 3860, which includes internal read/write memory/register circuitry for pattern storage. The PSS element 3860 permits a pattern set to be accumulated from any source. Accumulation of the pattern set may be necessary if the pattern creation process is slower than the pattern transmit and pattern receive processes. The amount of storage is typically dictated by the size of the pattern set desired. Storing the pattern set permits the pattern set to be transmitted in an uninterrupted burst.

[0333] Various embodiment of the PS element 3846 are similar to those described with reference to the PS elements in FIGS. 34 and 37. Because the MA element 3871 is available for pattern storage, the PSS element may be smaller than the PSS in components with no memory array element. The size of the PSS element 3860 is dictated by the amount of information that is written (or read) by the MA element 3871 in a single access cycle. In one embodiment, the MA element 3871 is a single port structure; that is, it may perform one access at a time. Each access requires one access cycle, and may either read or write a fixed amount of information in that access cycle.

[0334] This is illustrated in FIG. 38, in which the MA element 3871 receives either a read address (AR) or write address (AW), but not both simultaneously. In a similar fashion, the data port (D) of the MA element 3871 may either supply read data (DR) or accepts write data (DW), but not both simultaneously. Other embodiments include a memory array with more than a single port. The methods described herein are equally applicable to such a structure. For example, a two-port memory array capable of performing a read operation and a write operation simultaneously can replace any of the one port memories described herein. In some cases, a two port memory can replace two of the one port memories. However, memory arrays with more than a single port are more costly, and are typically limited to specialized applications.

[0335] Various implementations of the compare element 3848 are similar to those previously described with reference to the compare elements in FIGS. 34 and 37.

[0336] FIG. 39 is a flow diagram of a pattern generation process according to the embodiment of FIG. 38. In one embodiment, a pattern set is the same size as the amount of information read or written by the MA element 3871 in a single access cycle. At logic block 3902, the PS element 3846 creates a pattern set. The PS element 3846 stores the pattern set in the PSS element 3860 at logic block 3904. At logic block 3906, the PS element 3846 drives the pattern set on DW and writes into the MA element 3871 through the D port. If the final pattern set is written to the MA element 3871, as shown at logic block 3910, the PS element generates a read address AR and places it on the A port of the MA element 3871 at logic block 3914. If the final pattern set has not yet been written to the MA element 3871, as shown at logic block 3910, then one or more logic blocks 3902-3910 are repeated.

[0337] At logic block 3916, the MA element 3871 reads a pattern set on DR and drives the pattern set to the transmit interface 3800A2. The transmitter 3852 drives the pattern set onto the interconnect at logic block 3918. The receiver 3850 receives the pattern set from the interconnect at logic block 3920, and the delay element 3890 receives the pattern set from DR at logic block 3922. At logic block 3924, the compare element 3848 compares the pattern sets from the delay element 3890 and the receiver 3850, and at logic block 3926 stores match or mismatch information. It is determined at logic block 3928 whether the final pattern set has been read from memory. If the final pattern set has been read from memory, the process stops at logic block 3932. If the final pattern set has not been read from memory, the one or more logic blocks 3914-3928 are repeated.

[0338] The match/mismatch indication is used during test and characterization to indicate whether the parameter settings of the transmit and receive interfaces 3800A2 and 3800A1 are acceptable. The match/mismatch indication in the compare element may be accumulated from one pattern set to the next. In this embodiment, the delay element 3890 is present because the pattern sets are being read from a single port memory array. This delay element 3890 matches the delay seen for the path that passes from the transmitter 3852 through the interconnect and into the receiver 3850.

[0339] Transmit and Receive Interfaces with Dedicated Memory Arrays

[0340] FIG. 40 is a block diagram of an embodiment of transmit and receive interfaces with dedicated memory arrays. A transmit interface 4000A2 includes a transmitter 4052, and is coupled to a dedicated transmit pattern/memory (“PMT”) element 4070A2. The PMT 4070A2 includes a PST element 4046A2, a PSST 4060A2, and a transmit memory array (“MAT”) 4071A2. The transmit interface 4000A2 communicates with a receive interface 4000A1 through an interconnect. The receive interface 4000A1 includes a receiver 4050, a compare element 4048, and an optional delay element 4060. The receive interface is coupled to a dedicated receive pattern/memory element (“PMR”) 4070A1. The PMR 4070A1 includes a PSR element 4046A1, a PSSR 4060A1, and a receive memory array (“MAR”) 4071A1. The dedicated pattern source and memory array elements allow the transmit interface 4000A2 and the receive interface 4000A1 to reside on the same component or on different components. For example, the transmit interface 4000A2 and the receive interface 4000A1 may be arranged as shown in any of the FIGS. 35A-35B or 35C.

[0341] FIGS. 41A-C are flow diagrams of a pattern generation process according to the embodiment of FIG. 40. At logic block 4102, the PST element 4046A2 creates a pattern set as seen in FIG. 41A. At logic block 4104, a pattern set is stored in the PSST element 4060A2. At logic block 4106, the PST element 4046A2 generates a write address AW and places on A port of the MAT 4071A2. The PST element 4046A2 then drives the pattern set on DW and writes it into the MAT 4071A2 through the D port at logic block 4108. At logic block 4110, it is determined whether the final pattern set has been written to the MAT element 4071A2. If the final pattern set has not been written, the one or more logic blocks 41024108 are repeated. If the final pattern set has been written, the transmit side pattern generation and storage are done at logic block 4112.

[0342] The PSR element 4046A1 creates a pattern set at logic block 4114, and stored the pattern set in the PSSR element 4060A1 at logic block 4116 as shown in FIG. 41B. At logic block 4118, the PSR element 4046A1 generates a write address AW and places it on the A port of the MAR element 4071A1. At logic block 4120, the PSR element 4046A1 drives a pattern set on DW and writes it into the MAR element 4071A1 through the D port. At logic block 4122, it is determined whether the final pattern set has been written to the MAR element 4071A1. If the final pattern set has not been written to the MAR element 4071A1, one or more logic blocks 4114-4120 are repeated. If the final pattern set has been written to the MAR element 4071A1, the receive side pattern generation and storage are done at logic block 4124.

[0343] Pattern generation and storage for the transmit side, as described at logic block 4102-4110, and pattern generation and storage for the receive side, as described at logic block 4114-4120 may be in the order as described. Alternatively, however, pattern generation and storage for the transmit side and the receive side are done in any order. For example, the receive side process may be done before the transmit process and vice versa, the receive side process and the transmit side process can be performed simultaneously, actions in each of the respective processes may be interleaved, etc.

[0344] At logic block 4126 shown in FIG. 41C, the PST element 4046A2 generates a read address AR and places it on the A port of the MAT element 4071A2. The MAT element 4071A2 reads the pattern set on DR and drives it to the transmit interface 4000A2 at logic block 4128. At logic block 4130, the pattern set is driven onto interconnect by the transmitter 4052. The receiver 4050 accepts the pattern set from the interconnect at logic block 4132. The PSR element 4046A1 generates a read address AR and places it on the A port of the MAR element 4071A1 at logic block 4134. The MAR element 4071A1 reads the pattern set on DR and drives it to the delay element 4060 at logic block 4136. The compare element 4048 compared pattern sets from the delay element 4060 and the receiver 4050 at 4138. At logic block 4140, the compare element 4048 stores the match or mismatch indication from the compare. At logic block 4142, it is determined whether the final pattern set has been read form the MAT element 4071A2 and from the MAR element 4071A1. If the final pattern set has not been read from the MAT element 4071A2 and from the MAR element 4071A1, one or more logic blocks 4126-4142 are repeated. If the final pattern set has been read from the MAT element 4071A2 and from the MAR element 4071A1, the pattern generation and comparison process stops at logic block 4144.

[0345] The match/mismatch indication is used during test and characterization to indicate whether the parameter settings of the transmit and receive interfaces are acceptable. The match/mismatch indication in the compare element 4048 may be accumulated from one pattern set to the next. In this embodiment, the delay element 4060 is present. This delay element 4060 matches the delay seen for the path that passes from the transmitter 4052 through the interconnect and into the receiver 4050. The delay element 4060 is optional, however. If the delay element 4060 is not present, the timing of the PSR element 4046A1 and the MAR element 4071A1 is adjusted to create an offset relative to the PST element 4046A2 and the MAT element 4071A2. This offset matches the delay seen for the path that passes from the transmitter 4052 through the interconnect and into the receiver 4050, and has the same effect as the delay element.

[0346] FIG. 42 is a block diagram of another embodiment of transmit and receive interfaces with dedicated memory arrays. Individual elements of FIG. 42 are analogous to similarly named and numbered elements of FIG. 40, and will not be described in detail. As will be further described, the embodiment of FIG. 42 is suitable for the processes of transmitting and receiving pattern sets that are similar to the processes previously described. In this embodiment, however, the MAR element is both read and written during the process, as contrast to the processes described with reference to FIGS. 38 and 40, in which the MAR and MAT elements are only written to when the patterns are initially loaded.

[0347] FIGS. 43A-C are flow diagrams of a pattern generation, storage, and comparison process, described with further reference to FIG. 42. FIGS. 43A-C illustrate an embodiment of a process that allows additional testing of memory arrays, as well as testing of transmit interfaces and receive interfaces. The PST element 4246A2 creates a pattern set at logic block 4302 and stores it in the PSST element 4260A2 at logic block 4304 as seen in FIG. 43A. The PST element 4246A2 generates a write address AW and places on A port of the MAT element 4271A2 at logic block 4306. At logic block 4308, the PST element drives the pattern set on DW and writes it into the MAT element 4271A2 through the D port. At logic block 4310, it is determined whether the final pattern set was written to the MAT element 4271A2. If the final pattern set was not written, one or more logic blocks 4302-4310 are repeated. If the final pattern set was written, writing to the MAT element 4071A2 is done at logic block 4312.

[0348] At logic block 4314 shown in FIG. 43B, the PST element 4246A2 generates a read address AR and places it on the A port of the MAT element 4271A2. The MAT element 4271A2 reads the pattern set on DR and drives it to the transmit interface 4200A2 at logic block 4316. At logic block 4318, the transmitter 4252 drives the pattern set onto the interconnect. The receiver 4250 accepts the pattern set from the interconnect at logic block 4320. At logic block 4322, the PSR element 4246A1 generates a write address AW and places it on the A port of the MAR element 4271A1. The receiver 4250 drives the pattern set onto DW and writes it to the MAR element 4271A1 at logic block 4324. At logic block 4326, it is determined whether the final pattern set was read from the MAT element 4271A2. If the final pattern set was not read, one or more logic blocks 43144326 are repeated. If the final pattern set was read, reading from the MAT element 4271A2 is done at logic block 4328.

[0349] At logic block 4330, the PST element 4246A2 generates a read address AR and places it on the A port of the MAT element 4271A2. The MAT element 4271A2 reads the pattern set on DR and drives it to the delay element 4260 at logic block 4332. At logic block 4334, the delay element 4260 drives the pattern set to the compare element 4248. The PSR element 4246A1 generates a read address AR and places it on the A port of the MAR element 4271A1 at logic block 4336. The MAR element 4271A1 reads the pattern set on DR and drives it to the compare element 4248 at logic block 4338. At logic block 4340, the compare element 4248 compares the patterns sets from the delay element 4260 (which came from the MAT element 4271A2) and the MAR element 4271A1. The compare element 4248 stores the match or mismatch indication at logic block 4342. At logic block 4344, it is determined whether the final pattern sets have been read from the MAT element 4271A2 and the MAR element 4271A1. If the final pattern sets have not been read, one or more logic blocks 4330-4344 are repeated. If the final pattern sets have been read, then the reading from the MAT element 4271A2 and the MAR element 4271A1, and comparing are done at logic block 4346.

[0350] The match/mismatch indication is used during test and characterization to indicate whether the parameter settings of the transmit and receive interfaces are acceptable. The match/mismatch indication in the compare element may be accumulated from one pattern set to the next.

[0351] The delay element 4260 is used in this embodiment. The delay element 4260 creates a delay that matches the delay seen for the path that passes from the transmitter 4252 through the interconnect and into the receiver 4250. The delay element is optional, however. In other embodiments that don't include a delay element, the timing of the PSR element 4246A1 and the MAR element 4271A1 are adjusted to be offset relative to the timing of the PST element 4246A2 and the MAT element 4271A2. The offset matches the delay seen for the path that passes from the transmitter 4252 through the interconnect and into the receiver 4250, and has the same effect as the delay element.

[0352] FIG. 44 is a block diagram of another embodiment of transmit and receive interfaces with dedicated memory arrays. Individual elements of FIG. 44 are analogous to similarly named and numbered elements of FIGS. 40 and 42, and will not be described in detail. The embodiment of FIG. 44 uses a similar process for transmitting and receiving pattern sets to those previously described. In the embodiment of FIG. 44, however, includes additional PS and MA elements not present in previously described embodiments. In the embodiment of FIG. 44, an additional MAR element is both read and written during the process. In contrast, the embodiments of FIGS. 40 and 42 are described as reading and writing MAR and MAT elements only when pattern sets are initially loaded.

[0353] FIGS. 45A-D are flow diagrams of a pattern generation, storage, and comparison process, described with further reference to FIG. 44. The process that will be described with reference to FIGS. 44 and 45A-D performs more testing of the memory array, as well as testing of the interfaces. At logic block 4502, the PST element 4446A2 generates a pattern set, and at logic block 4504, stores the pattern set in the PSST element 4460A2. The PST element 4446A2 generates a write address AW and places on the A port of the MAT element 4471A2 at logic block 4506. At logic block 4508, the PST element 4446A2 drives the pattern set on DW and writes it into the MAT element 4471A2 through D port. At logic block 4510, it is determined whether the final pattern set was written to the MAT element 4471A2. If the final pattern set was not written to the MAT element 4471A2, one or more logic blocks 45024510 are repeated. If the final pattern set was written, then writing to the MAT element 4471A2 is done at logic block 4514.

[0354] At logic block 4516 shown in FIG. 45B, the PSR1 element 4446A1 generates a pattern set, and at logic block 4518, stores it in the PSSR1 element 4460A1. The PSR1 element 4446A1 generates a write address AW and places on the A port of the MAR1 element 4471A1 at logic block 4520. At logic block 4522, the PSR1 element 4446A1 drives the pattern set on DW and writes it into D port of the MAR1 element 4471A1. At logic block 4524, it is determined whether the final pattern set was written to the MAR1 element 4471A1. If the final pattern set was not written to the MAR1 element 4471A1, one or more logic blocks 4514-4524 are repeated. If the final pattern set was written, then writing to the MAR1 element 4471A1 is done at logic block 4528.

[0355] Pattern generation and storage for the transmit side, as described at logic blocks 4502-4510, and pattern generation and storage for the receive side, as described at logic blocks 4514-4524 may be in the order as described. Alternatively, however, the actions specified are done in any order. For example, the receive side process may be done before the transmit process and vice versa, the receive side process and the transmit side process can be performed simultaneously, actions in each of the respective processes may be interleaved, etc.

[0356] At logic block 4530 shown in FIG. 45C, the PST element 4446A2 generates a read address AR and places it on the A port of the MAT element 4471A2. The MAT element 4471A2 reads the pattern set on DR and drives it to transmit interface 4400A2 at logic block 4532. The transmitter 4452 drives the pattern set onto the interconnect at logic block 4534, and at logic block 4536, the receiver 4450 accepts the pattern set from the interconnect. The PSR2 element 4447A1 generates a write address AW and places on A port of MAR2 element 4472A1 at logic block 4538. At logic block 4540, the receiver 4450 drives the pattern set onto DW and writes it to the MAR24472A1. At logic block 4542, it is determined whether the final pattern set was read from MAT element 4471A2. If the final pattern set was not read from MAT element 4471A2, one or more logic blocks 4530-4542 are repeated. If the final pattern set was read, then reading from the MAT element 4471A2 is done at logic block 4546.

[0357] At logic block 4548 shown in FIG. 45D, the PSR1 element 4446A1 generates a read address AR and places it on the A port of the MAR14471A1. The MAR1 element 4471A1 reads the pattern set on DR and drives it to the delay element 4460 at logic block 4550. At logic block 4552, the delay element 4460 drives the pattern set to the compare element 4448. At logic block 4554, the PSR2 element 4447A1 generates a read address AR and places it on the A port of the MAR2 element 4472A1. The MAR2 element 4472A1 reads the pattern set on the DR and drives to the compare element 4448 at logic block 4556. The compare element 4448 compares pattern sets from the delay element 4460 (which come from the MAR1 element 4471A1) and MAR2 element 4472A1 at logic block 4558. At logic block 4560, the compare element stores a match or mismatch indication. At logic block 4562, it is determined whether the final pattern set was read from the MAR1 element 4471A1 and the MAR2 element 4472A1. If the final pattern set was not read, one or more logic blocks 4548-4562 are repeated. If the final pattern set was read, then reading from the MAR1 element 4471A1 and the MAR2 element 4472A1 is done at logic block 4566.

[0358] The match/mismatch indication is used during test and characterization to indicate whether the parameter settings of the transmit and receive interfaces are acceptable. The match/mismatch indication in the compare element may be accumulated from one pattern set to the next. The delay element 4460 is used in the embodiment described. The delay element 4460 creates a delay the matches the delay seen for the path that passes from the transmitter 4452 through the interconnect and into the receiver 4450. The delay element 4460 is optional, however. If the delay element 4460 were left out, the timing of the PSR1 element 4446A1 and the MAR1 element 4471A1 is adjusted to be offset relative to the PSR2 element 4447A1 and the MAR2 element 4472A1. The offset matches the delay seen for the path that passes from the transmitter 4452 through the interconnect and into the receiver 4450, and has the same effect as the delay element 4460.

[0359] FIG. 46 is a block diagram of another embodiment of transmit and receive interfaces with dedicated memory arrays. Individual elements of FIG. 46 are analogous to similarly named and numbered elements of FIGS. 40, 42, and 44 and will not be described in detail. The embodiment of FIG. 46 uses a similar process for transmitting and receiving pattern sets to those previously described. FIGS. 47A-C are flow diagrams of a pattern generation, storage, and comparison process, described with further reference to FIG. 46. In the embodiment of FIGS. 46 and 47A-C, the transmit and receive interfaces are not tested, in contrast to the previous embodiments described. This embodiment demonstrates how the pattern generation circuitry that is used for testing the transmit and receive interfaces is used to test the memory arrays. This is useful, for example, when more patterns are needed for testing the memory arrays than are needed for testing the transmit and receive interfaces. By making the tests more specialized, the overall test time may be reduced.

[0360] Referring to FIGS. 46 and 47A-C, at logic block 4702, PST element 4646A2 creates a pattern set, and at logic block 4704, stores it in the PSST element 4660A2. At logic block 4706, the PST element 4646A2 generates a write address AW and places it on the A port of the MAT element 4671A2. At logic block 4708, the PST element 4646A2 drives a pattern set on DW and writes it to the D port of the MAT element 4671A2. At logic block 4710, it is determined whether the final pattern set was written to the MAT element 4671A2. If the final pattern set was not written, one or more logic blocks 4702-4710 are repeated. If the final pattern set was written, then writing to the MAT element 4671A2 is done at logic block 4714.

[0361] At logic block 4716 seen in FIG. 47B, the PSSR element 4646A1 creates a pattern set, and at logic block 4718, stores it in the PSSR element 4660A1. The PSR element 4660A1 generates a write address AW and places it on A the port of the MAR element 4671A1 at logic block 4720. At logic block 4722, the PSR element 4646A1 drives a pattern set on Dw and writes it into the D port of the MAR element 4671A1. At logic block 4724, it is determined whether the final pattern set was written to the MAR element 4671A1. If the final pattern set was not written, one or more logic blocks 4714-4724 are repeated. If the final pattern set was written, then writing to the MAR element 4671A1 is done at logic block 4728.

[0362] Pattern generation and storage for the transmit side, as described at logic blocks 4702-4710, and pattern generation and storage for the receive side, as described at logic blocks 4714-4724, may be in the order as described. Alternatively, however, the actions specified are done in any order. For example, the receive side process may be done before the transmit process and vice versa, the receive side process and the transmit side process can be performed simultaneously, actions in each of the respective processes may be interleaved, etc.

[0363] At logic block 4730 shown in FIG. 47C, the PST element 4646A2 generates a read address AR and places it on the A port of the MAT element 4671A2. The MAT element 4671A2 reads the pattern set on DR and drives it to the compare element 4648 at logic block 4732. At logic block 4734, the PSR element 4646A1 generates a read address AR and places it on the A port of the MAR element 4671A1. The MAR element 4671A1 reads the pattern set on DR and drives it to the compare element 4648 at logic block 4736. At logic block 4738, the compare element 4648 compares pattern sets from the MAT element 4671A2 and the MAR element 4671A1. The compare element 4648 stores a match or mismatch indication at logic block 4740. At logic block 4742, it is determined whether the final pattern sets were read from the MAT element 4671A2 and the MAR element 4671A2. If the final pattern sets were not read, one or more logic blocks 4728-4742 are repeated. If the final pattern sets were read, then reading from the MAT element 4671A2 and the MAR element 4671A1 is done at logic block 4746.

[0364] The match/mismatch indication is used during test and characterization to indicate whether the patterns have been written to and read from the memory arrays correctly. The match/mismatch indication in the compare element may be accumulated from one pattern set to the next. Various embodiments of a component interface that eliminates the requirement of a high-speed test system have been described. The interface embodiments described may reside on one or multiple components. In various embodiments, a component may test itself. Components may also test each other using test circuitry in their respective interfaces. Respective interfaces may be coupled through an interconnect for the specific purpose of testing one another. Alternatively, interfaces that are couple to each other within a system for normal system operation, such as a memory component and a memory controller component, can test each other within the system through their usual interconnect. The embodiments described allow uniform, high-speed testing at the wafer stage, the component stage, and the system stage without an expensive high-speed testing system. The embodiments described are consistent with the use of various signaling methods and protocols, for example, single-ended signaling, differential signaling, voltage-mode operation, current-mode operation, and more. Other variations of the embodiments described are also possible, and are within the scope of the invention.

Claims

1. A semiconductor device, comprising: a test circuit; a first element; and a second element comprising at least one of a transmitter and a receiver; wherein when the first element is coupled to the test circuit and the second element is coupled to the first element, at least one of the first and second elements is capable of testing another one of the first and second elements using the test circuit; and the first element comprises a transmitter when the second element comprises a receiver and the first element comprises a receiver when the second element comprises a transmitter.

2. The semiconductor device of claim 1, wherein the semiconductor device is configured to operate at high frequencies.

3. The semiconductor device of claim 1, wherein the semiconductor device is configured to operate at a frequency of at least 1 Gigahertz.

4. The semiconductor device of claim 1, wherein the semiconductor device is configured to operate at a frequency of at least 3 Gigahertz.

5. The semiconductor device of claim 1, wherein the test circuit includes a pattern generator and a pattern compare circuitry.

6. The semiconductor device of claim 1, wherein the semiconductor device is configured to be coupled to at least one reference signal that includes a voltage reference signal, VREF, and wherein the test circuit includes at least one comparator circuit for comparing at least one voltage signal representing received data with the VREF signal.

7. The semiconductor device of claim 6, wherein the at least one reference signal further includes a time reference signal, TREF, and the high bandwidth test circuit includes at least one analog sampling circuit controlled by the TREF signal, wherein the analog sampling circuit receives at least one voltage signal representing received data, such that TREF determines a time offset between receipt of the at least one voltage signal by the receiver and evaluation of the voltage signal by the receiver.

8. The semiconductor device of claim 7, wherein the TREF signal is scannable so as to characterize an actual output eye of the transmitter.

9. The semiconductor device of claim 7, wherein the TREF signal is scannable so as to characterize an actual input eye of the receiver.

10. The semiconductor device of claim 1, wherein the semiconductor device is coupled to an interconnect to receive and transmit a complementary voltage value VN and a true voltage value VP, wherein the test circuit further comprises: a test pattern generation circuit that generates test patterns for transmission to a component under test; a test pattern comparison circuit that compares a generated test pattern with a pattern received from a component under test; a first differential comparator that compares VN with a voltage reference value, VREF to produce a first result; a second differential comparator that compares VP with a voltage reference value, VREF to produce a second result; and a multiplexer that selects one of the first result and the second result to transmit to the comparator for comparison with the test pattern.

11. The semiconductor device of claim 10, wherein the test circuit further comprises an analog sampling circuit that delays the VP and VN signals under control of a TREF signal so as to decouple parameters of the transmitter and the receiver.

12. The semiconductor device of claim 11, wherein the delayed VP and VN signals produce a third result, and wherein the multiplexer further selects one of the first result, the second result, and the third result to transmit to the comparator for comparison with the test pattern.

13. The semiconductor device of claim 12, wherein the transmitter is coupled to the receiver via an interconnect, and wherein the transmitter and the receiver are on a semiconductor wafer.

14. The semiconductor device of claim 12, wherein the transmitter is coupled to the receiver via an interconnect, wherein the interconnect is in a system and is used for system communication between the transmitter and the receiver.

15. A semiconductor circuit component coupleable to an external receiver and an external transmitter, the semiconductor component comprising: a first transmitter coupled to a first signal path and configurable to drive a first signal according to predetermined transmitter specification parameters; and a first receiver coupled to a second signal path and configurable to sample a second signal according to predetermined receiver specification parameters, wherein, when the component is configured to operate in a first mode, the first signal path is coupled to the external receiver, and the second signal path is coupled to the external transmitter, and when the component is further configurable to operate in a second mode in which, the first signal path is coupled to the second signal path; the first receiver is operable to evaluate the first signal for compliance with the predetermined transmitter specification parameters; and the first transmitter is operable to evaluate the second signal for compliance with the predetermined receiver specification parameters.

16. The component of claim 15, wherein the component is configured to operate at high frequencies.

17. The component of claim 16, wherein the semiconductor device is configured to operate at a frequency of at least 1 Gigahertz.

18. The component of claim 16, wherein the semiconductor device is configured to operate at a frequency of at least 3 Gigahertz.

19. The component of claim 15, wherein the predetermined transmitter specifications of the first transmitter and the predetermined receiver specification parameters of the first receiver must be met for the first transmitter and the first receiver to be suitable for operation in the first mode, and wherein evaluation comprises determining that the first transmitter and the first receiver operate properly when the respective predetermined specification parameters are met.

20. The component of claim 15, wherein the predetermined transmitter specification parameters of the first transmitter and the predetermined receiver specification parameters of the first receiver must be met for the first transmitter and the first receiver to be suitable for operation in the first mode, and wherein evaluation comprises: determining an amount of margin present between operating characteristics of the first transmitter and the predetermined transmitter specification parameters being evaluated; and determining an amount of margin present between operating characteristics of the first receiver and the predetermined receiver specification parameters being evaluated.

21. The component of claim 15, wherein, in the second mode, the first transmitter and the first receiver are in different interface blocks in the component and are each configurable to be coupled via an external interconnect.

22. The component of claim 15, wherein, in the second mode, the first transmitter and the first receiver are in a same interface block in the component and are coupled via an internal interconnect.

23. The component of claim 15, further comprising a first timing circuit that receives a first external timing signal and produces a first internal timing signal.

24. The component of claim 23, wherein the first internal timing signal is coupled to the first transmitter, and an internal timing event on the first internal timing signal causes the first transmitter to drive a first symbol on the first signal.

25. The component of claim 23, wherein the first internal timing signal is coupled to the first receiver, and an internal timing event on the first internal timing signal causes the first receiver to sample a first symbol on the second signal.

26. The component of claim 23, wherein the first timing circuit aligns an internal timing event on the internal timing signal with an external timing event on the external timing signal.

27. The component of claim 23, wherein the first timing circuit creates at least two internal timing events on the first internal timing signal, that are each associated with an external timing event on the external timing signal.

28. The component of claim 23, wherein the first timing circuit creates an internal timing event on the internal timing signal, wherein the internal timing event has a fixed time offset relative to an external timing event on an associated external timing signal.

29. The component of claim 23, wherein the first timing circuit creates an internal timing event on the internal timing signal, wherein the internal timing event has an adjustable time offset relative to an external timing event on the associated external timing signal.

30. The component of claim 29, wherein the adjustable time offset is specified by a value held in a register in the component.

31. The component of claim 29, wherein evaluation of the respective predetermined specification parameters of the receiver and the transmitter includes accessing a value that determines the adjustable time offset.

32. The component of claim 23, wherein evaluation of the respective predetermined specification parameters of the receiver and the transmitter includes measuring a position of an external timing event on the first external timing signal.

33. The component of claim 15, further comprising: a first timing circuit that receives a first external timing signal and produces a first internal timing signal; and a second timing circuit that receives the second external timing signal and produces a second internal timing signal, such that the first internal timing signal is coupled to the first transmitter, and the second internal timing signal is coupled to the first receiver, so that a first timing event on the first internal timing signal causes the first transmitter to drive a first symbol on the first signal and a second timing event on the second internal timing signal causes the first receiver to sample a second symbol on the second signal.

34. The component of claim 33, wherein evaluation of the specification parameters of the receiver and the transmitter comprises: measuring a position of a first external timing event on the first external timing signal; and measuring a position of a second external timing event on the second external timing signal, wherein the first external timing event is associated with the first internal timing event and the second external timing event is associated with the second internal timing event.

35. The component of claim 34, wherein the first external timing event is aligned with the first internal timing event and the second external timing event is aligned with the second internal timing event.

36. The component of claim 15, further comprising: a first timing circuit that receives a first external timing signal and produces a first internal timing signal; and a second timing circuit that receives the first external timing signal and produces a second internal timing signal, such that the first internal timing signal is coupled to the first transmitter, and the second internal timing signal is coupled to the first receiver, so that a first timing event on the first internal timing signal causes the first transmitter to drive a first symbol on the first signal and a second timing event on the second internal timing signal causes the first receiver to sample a second symbol on the second signal.

37. The component of claim 36, wherein: the first timing circuit creates a first internal timing event on the first internal timing signal; the first internal timing event has a first adjustable time offset relative to a first external timing event on the associated external timing signal; wherein the second timing circuit creates a second internal timing event on the second internal timing signal; and wherein the second internal timing event has a second adjustable time offset relative to a second external timing event on the associated external timing signal.

38. The component of claim 37, wherein the first adjustable time offset is specified by a first value held in a first register in the component and the second adjustable time offset is specified by a second value held in a second register in the component.

39. The component of claim 37, wherein evaluation of the respective predetermined specification parameters of the receiver and the transmitter includes accessing a first value that determines the first adjustable time offset and includes accessing a second value that determines the second adjustable time offset.

40. The component of claim 15, wherein a voltage discriminator circuit is coupled to one of the first signal and the second signal, and measures a voltage level of one of the first signal and the second signal in the course of evaluating the respective predetermined transmitter and receiver specification parameters.

41. The component of claim 40, wherein a voltage reference signal is coupled to the voltage discriminator circuit to provide a voltage comparison standard for measuring a voltage level on one of the first signal and the second signal.

42. The component of claim 41, wherein measuring includes at least one comparison of a voltage level of one of the first signal and the second signal and a voltage reference signal

43. The component of claim 41, wherein the voltage discriminator circuit is a differential comparator with one input coupled to one of the first signal and the second signal and another input coupled to the voltage reference signal.

44. The component of claim 41, wherein the voltage reference signal is configurable to be generated outside the component.

45. The component of claim 41, wherein the voltage reference signal is generated inside the component by a voltage reference generator circuit, and wherein the voltage reference generator circuit is coupled to a register containing a value that adjusts a voltage level of the voltage reference signal.

46. The component of claim 15, wherein the receiver includes a voltage discriminator circuit and measures a voltage level of the second signal in the course of evaluating the respective predetermined transmitter and receiver specification parameters.

47. The component of claim 46, wherein a voltage reference signal is coupled to the voltage discriminator circuit to provide a voltage comparison standard for measuring a voltage level on the second signal.

48. The component of claim 47, wherein measuring includes at least one comparison of a voltage level of the second signal and the voltage reference signal.

49. The component of claim 47, wherein the voltage discriminator circuit is a differential comparator with one input coupled to the second signal and another input coupled to the voltage reference signal.

50. The component of claim 47, wherein the voltage reference signal is configurable to be generated outside the component.

51. The component of claim 47, wherein the voltage reference signal is generated inside the component by a voltage reference generator circuit, wherein the voltage reference generator circuit is coupled to a register containing a value that adjusts a voltage level of the voltage reference signal.

52. The component of claim 15, wherein a timing discriminator circuit couples a third signal to one of the first signal and the second signal, wherein the third signal is further coupled to a third receiver on the component, wherein the third receiver being similar to the first receiver, and wherein the time discriminator circuit is used to evaluate the respective predetermined specification parameters of the first transmitter and of the first receiver.

53. The component of claim 52, wherein a timing reference signal is coupled to the timing discriminator circuit, and wherein the timing reference signal is configurable to be generated outside the component.

54. The component of claim 53, wherein a timing event on the timing reference signal causes the third signal to assume a same value as the one of the first signal and second signal.

55. The component of claim 53, wherein a timing event on the timing reference signal causes the third signal to become independent of the one of the first signal and second signal, and to retain a value as of the timing event.

56. The component of claim 53, wherein the timing discriminator circuit includes an MOS pass transistor coupling the one of the first signal and second signal and the third signal, wherein the timing reference signal is coupled to the gate of the MOS pass transistor, and wherein a capacitor is coupled between the third signal and a fixed voltage reference.

57. The component of claim 53, wherein the timing reference signal is coupled directly from an external source to the timing discriminator circuit with an internal interconnect.

58. The component of claim 52, wherein a timing reference signal is coupled to the timing discriminator circuit and is generated inside the component by a timing reference generator circuit, to which is coupled a register containing a value which adjusts the timing offset of a timing event on the timing reference signal.

59. The component of claim 15, wherein a first timing discriminator circuit couples a third signal to the second signal, and a second timing discriminator circuit couples the third signal to a fourth signal, wherein the fourth signal is coupled to a third receiver on the component, the third receiver being similar to the first receiver, wherein the first and second time discriminator circuits are used to evaluate the respective predetermined specification parameters of the first transmitter and of the first receiver.

60. The component of claim 15, wherein the first receiver consists of a first timing discriminator circuit coupled to one of a voltage discriminator circuit and a second timing discriminator circuit, such that the in the first mode of operation the first timing discriminator couples the second signal to the one of a voltage discriminator circuit and a second timing discriminator circuit, and in the second mode of operation, the first time discriminator circuit is used to evaluate the respective predetermined specification parameters of the first transmitter and of the first receiver.

61. The component of claim 60, wherein a timing reference signal is coupled to the first timing discriminator circuit and is configurable to be generated outside the component, and wherein the timing reference signal is coupled directly from an external source to the timing discriminator circuit with an internal interconnect.

62. The component of claim 15, further comprising a current control circuit coupled to the transmitter, wherein the current control circuit determines a current passing through the first transmitter to determine a voltage level driven on the first signal in the first mode and in the second mode.

63. The component of claim 62, further comprising a current reference signal coupled to the current control circuit to provide a current comparison standard for determining a current passing through the first transmitter.

64. The component of claim 63, wherein the current control circuit and the first transmitter each include a plurality of transistor elements whose conductive strengths form a binary sequence.

65. The component of claim 63, wherein the current reference signal is configurable to be generated outside the component.

66. The component of claim 63, further comprising: a current reference generator circuit that generates the current reference signal inside the component; and a register coupled to the reference generator circuit, wherein the register stores a value that adjusts a value of the current reference signal.

67. The component of claim 63, wherein the current reference signal is varied between at least two values during evaluation of the transmitter and receiver specification parameters in the second mode.

68. The component of claim 15, further comprising: a termination control circuit; a termination element coupled to the termination control circuit and to one of the first signal and the second signal, wherein the termination control circuit determines a value of a termination load present on one of the first signal and the second signal in the second mode, and in the first mode.

69. The component of claim 68, further comprising a termination reference signal coupled to the termination control circuit to provide a termination comparison standard for determining a value of a termination load present on one of the first signal and the second signal.

70. The component of claim 69, wherein the termination control circuit and the termination element each include a plurality of transistor elements whose conductive strengths form a binary sequence.

71. The component of claim 69, wherein the termination reference signal is configurable to be generated outside the component.

72. The component of claim 69, further comprising: a termination reference generator circuit that generates the termination reference signal inside the component; and a register coupled to the termination generator circuit, wherein the register stores a value that adjusts a value of the termination reference signal.

73. The component of claim 69, wherein the termination reference signal is varied between at least two values in the second mode.

74. The component of claim 73, wherein the termination voltage signal is varied between at least two values in the second mode.

75. The component of claim 15, further comprising a termination element coupled to a termination voltage signal, and further coupled to one of the first signal and the second signal, wherein the termination element determines voltage levels present on one of the first signal and the second signal in the second mode and in the first mode.

76. The component of claim 75, wherein the termination voltage signal is configurable to be generated outside the component.

77. The component of claim 75, further comprising: a termination voltage generator circuit that generates the termination voltage signal inside the component; and a register coupled to the termination voltage generator circuit, wherein the register stores a value that adjusts a value of the termination voltage signal.

78. The component of claim 15, further comprising at least one pattern generator circuit, wherein pattern data from the at least one pattern generator circuit is coupled to the first transmitter to be driven as transmitter pattern data on the first signal, and to be sampled by the receiver as received pattern data on the second signal.

79. The component of claim 15, further comprising at least one pattern comparison circuit that compares the received pattern data to the transmitted pattern data.

80. The component of claim 15, wherein at least one of the first signal and the second signal are encoded using two differential voltage values.

81. The component of claim 15, wherein at least one of the first signal and the second signal are encoded using single-ended voltage values.

82. The component of claim 15, wherein at least one of the first signal and the second signal are encoded using single-ended voltage values with reference to an external reference voltage value.

83. The component of claim 15, wherein at least one of the first signal and the second signal are each bidirectional.

84. The component of claim 15, wherein at least one of the first signal and the second signal are each unidirectional.

85. The component of claim 15, wherein the first transmitter includes at least one unipolar driver.

86. The component of claim 15, wherein the first transmitter includes at least one bipolar driver.

87. The component of claim 15, wherein the first transmitter uses an external reference value to generate signal values.

88. The component of claim 15, further comprising at least one calibration circuit, wherein the first transmitter uses an external reference value and the at least one calibration circuit to generate signal values.

89. The component of claim 88, wherein the calibration circuitry is adjustable.

90. The component of claim 89, wherein the calibration circuitry is adjustable using a value in a register of the component.

91. The component of claim 15, wherein at least one of the first signal and the second signal is coupleable to an interconnect that couples the component to one other component.

92. The component of claim 15, wherein at least one of the first signal and the second signal is coupleable to an interconnect that couples the component to at least two other components.

93. The component of claim 15, wherein at least one of the first signal and the second signal use current-mode operation.

94. The component of claim 15, wherein at least one of the first signal and the second signal use voltage-mode operation.

95. The component of claim 15, further comprising at least one slew rate control circuit coupled to control respective slew rates of at least one of the first signal and the second signal.

96. The component of claim 15, wherein at least one of the first signal and the second signal is coupled to respective external termination loads.

97. The component of claim 15, wherein at least one of the first signal and the second signal is coupled to respective internal termination loads.

98. The component of claim 97, wherein at least one of the internal termination loads is calibrated to an external reference value.

99. The component of claim 98, wherein at least one of the internal termination loads is adjustable.

100. The component of claim 99, wherein at least one of the internal termination loads is adjusted using a value in an internal register.

101. An electronic system comprising: at least one semiconductor component configurable to perform testing functions, wherein the at least one semiconductor component has at least two operational modes comprising a normal mode and a test mode, the at least one semiconductor component comprising, a transmitter; a receiver; and a test circuit, wherein in the test mode the transmitter is configurable to test the receiver and the receiver is configured to test the transmitter.

102. The system of claim 101, wherein in at least one semiconductor component is a component configured to operate at high frequencies.

103. The system of claim 102, wherein the semiconductor component is configured to operate at a frequency of at least 1 Gigahertz.

104. The system of claim 102, wherein the semiconductor component is configured to operate at a frequency of at least 3 Gigahertz.

105. The system of claim 101, wherein in the test mode, the transmitter and the receiver are on a same semiconductor component.

106. The system of claim 101, wherein in the test mode, the transmitter and the receiver are on different interfaces of a same semiconductor component.

107. The system of claim 101, wherein in the test mode, the transmitter and the receiver are on different semiconductor components.

108. The system of claim 101, wherein the at least one semiconductor component comprises a memory interface on a memory component and a memory controller interface on a memory controller component, and wherein the memory component and the memory controller component are coupled via an interconnect in the test mode and in the normal mode.

109. The system of claim 108, wherein the interconnect is a high frequency interconnect.

110. The system of claim 108, wherein the interconnect is configured to operate at a frequency of at least 1 Gigahertz.

111. The system of claim 108, wherein the interconnect is configured to operate at a frequency of at least 3 Gigahertz.

112. The system of claim 101, wherein the at least one semiconductor component is coupled to a plurality of reference signals, including a voltage reference signal, VREF, a receiver clock reference signal, CLKRREF, a transmitter clock reference signal CLKTREF, and a current reference signal, IREF, and wherein the reference signals are scanned in the test mode, wherein scanning comprises sweeping a value of a reference signal to determine an event on another signal.

113. The system of claim 112, wherein a differential comparator receives the VREF signal and compares the VREF signal to a signal voltage on the interconnect as the VREF signal is scanned to produce a test output signal.

114. The system of claim 112, wherein the plurality of reference signals further comprise a time reference signal, TREF, and wherein the test circuit further comprises an analog sampling circuit controlled by the TREF signal so as to control sampling times of a signal voltage on the interconnect in the test mode independent of individual interface parameters, wherein the analog sampling circuit produces a time test voltage signal.

115. The system of claim 114, wherein the test circuit further comprises: a test pattern circuit; a register that holds the test output signal and the time test voltage signal; and a selection circuit coupled to the test pattern circuit, that selects a value from the register to be compared to a test pattern.

116. The system of claim 112, wherein the plurality of reference signals further comprise a test current reference signal ITL-REF that controls a test current generated to allow a pin of one of the at least one semiconductor components to test itself.

117. An electronic system comprising: at least one semiconductor component configurable to perform testing functions, wherein the at least one semiconductor component has at least two operational modes comprising a normal mode and a test mode, the at least one semiconductor component comprising, a transmitter; a receiver; and a test circuit, wherein in the test mode the transmitter is configurable to test the receiver or the receiver is configured to test the transmitter.

118. A method for a semiconductor component to be tested by another semiconductor component, the method comprising: generating a test pattern; storing the test pattern in the semiconductor component; providing a reference signal to the semiconductor component; generating an operational signal by processing a data signal in the semiconductor component; scanning the reference signal; comparing the operational signal to the reference signal, while scanning the reference signal; and comparing a result of the comparison with the test pattern.

119. The method of claim 118, wherein at least one of the semiconductor component and the other semiconductor component is configured for operation at high frequencies.

120. The method of claim 119, wherein at least one of the semiconductor component and the other semiconductor component is configured for operation at a frequency of at least 1 Gigahertz.

121. The method of claim 119, wherein at least one of the semiconductor component and the other semiconductor component is configured for operation at a frequency of at least 3 Gigahertz.

122. The method of claim 118, wherein the method is performed by the semiconductor component on the other semiconductor component, and wherein the semiconductor component and the other semiconductor component are on at least one semiconductor wafer.

123. The method of claim 122, wherein the semiconductor component and the other semiconductor component comprise a transmitter and a receiver, and wherein the plurality of signals further comprises a time reference signal that decouples semiconductor component transmitter timing parameters from semiconductor component receiver timing parameter, the method further comprising scanning the time reference signal to define a output eye region for the transmitter, including: creating a series of timing events, including an active clock edge that causes the transmitter to output data; determining a time after an active clock edge when the output data becomes valid; and determining a time after an active clock edge when the output data becomes invalid.

124. The method of claim 123, further comprising using the time reference signal to control an analog sampler that samples the output data and creates a time offset between when the output data becomes valid and when the output data is sampled as input data by the receiver.

125. The method of claim 124, further comprising: setting the time offset using the time reference signal such that an input eye region of the receiver occurs within the output eye region defined for the transmitter; and scanning a receiver clock signal to determine whether a setup time specification and a hold time specification for the receiver are met.

126. The method of claim 118, wherein the method is performed by the semiconductor component on the other semiconductor component, and wherein the semiconductor component and the other semiconductor component are in at least one component package.

127. The method of claim 126, wherein the method is performed on the high-speed receiver circuit, the method comprising: scanning a voltage reference signal; and comparing the scanned voltage reference signal with a normal operational signal of the receiver to determine a high input voltage and a low input voltage.

128. The method of claim 127, further comprising scanning a transmitter clock reference signal to determine whether a setup time specification and a hold time specification for the transmitter are met.

129. The method of claim 118, wherein the semiconductor component and the other semiconductor component comprises a high frequency transmitter circuit and a high frequency receiver circuit.

130. The method of claim 129, wherein the high frequency transmitter circuit and the high frequency receiver circuit operate at a frequency of at least 1 Gigahertz.

131. The method of claim 129, wherein the high frequency transmitter circuit and the high frequency receiver circuit operate at a frequency of at least 3 Gigahertz.

132. The method of claim 129, wherein the method is performed on the high-speed transmitter circuit, the method further comprising: scanning a voltage reference signal and; comparing the scanned voltage reference signal with a normal operational signal of the transmitter to determine a high output voltage and a low output voltage.

133. The method of claim 132, further comprising scanning a receiver clock reference signal to determine whether a setup time specification and a hold time specification for the receiver are met.

134. A method for using one component to test and characterize another component, comprising: receiving at least one first signal having a first frequency; receiving a plurality of second reference signals having a plurality of second frequencies, wherein at least one of the second frequencies is lower than the first frequency; generating and storing a test pattern; scanning one of the second reference signals; comparing an operational signal to one of the second reference signals, while scanning one of the reference signals, wherein the operational single comprises a data signal processed by a circuit during operation; and comparing a result of the comparison with the test pattern.

135. A semiconductor circuit component comprising: a transmit interface comprising, a first transmitter; and transmitter test circuitry; and a receive interface comprising, a first receiver: and receiver test circuitry, wherein the first transmitter drives a first signal according to predetermined transmitter specifications, the first receiver samples a second signal according to predetermined receiver specifications, and wherein the component is configurable to operate in a test mode in which, the first signal is coupled to the second signal such that the first receiver evaluates transmitter specification parameters of the first transmitter; and the first signal is coupled to the second signal such that the first transmitter evaluates receiver specification parameters of the first receiver.

136. The component of claim 135, wherein the receive interface and the transmit interface are each configured to transfer data at rates equal to or greater than 3 GHz.

137. The component of claim 135, wherein the transmitter test circuitry comprises a first pattern source element, and wherein a first pattern set transmitted from the first pattern source element to the first transmitter, and is driven by the first transmitter onto the first signal during operation in the test mode.

138. The component of claim 137, wherein the transmitter test circuitry further comprises a first pattern storage element coupled to the first pattern source element, wherein the first pattern storage element stores the first pattern set as the first pattern set is generated.

139. The component of claim 137 in which the first pattern set is generated by circuitry in the first pattern source element.

140. The component of claim 137 in which the first pattern set is generated by circuitry external to the component and communicated to the first pattern source element.

141. The component of claim 137 in which the first pattern set is generated when the first pattern source element is controlled by at least one first external sideband signal whose transfer rate is lower than that of the first signal.

142. The component of claim 137, wherein the receiver test circuitry comprises: a delay element coupled to the first pattern source element; and a compare element coupled to the delay element, wherein the first receiver samples a second pattern set on the second signal, the first pattern set is delayed by the first delay element to form a third pattern set, and wherein the compare element is compares the first pattern set to the third pattern set.

143. The component of claim 137, wherein the transmitter test circuitry further comprises a memory array coupled to the first pattern source element, wherein the first pattern set is stored in the memory array.

144. The component of claim 143, wherein at least one pattern set is written from the first pattern source element to the memory array.

145. The component of claim 143, wherein at least one pattern set is read from the memory array to the first transmitter and driven onto the first signal during operation in the test mode.

146. The component of claim 143, wherein the receiver test circuitry further comprises: a delay element coupled to the memory array; and a compare element coupled to the delay element, wherein the first receiver samples a second pattern set on the second signal such that the first pattern set is delayed by the delay element to form a third pattern set, and wherein the compare element compares the first pattern set to the third pattern set.

147. The component of claim 135, further comprising a first pattern source element, wherein the first receiver samples a first pattern set on the second signal, and a second pattern set is provided by the first pattern source element during operation in the test mode.

148. The component of claim 147, wherein the first pattern source element includes a first pattern storage element used to accumulate the first pattern set as the first pattern is generated.

149. The component of claim 147, wherein the first pattern set is generated by circuitry in the pattern source element.

150. The component of claim 147, wherein the first pattern set is generated by circuitry external to the component and is communicated to the pattern source element.

151. The component of claim 147, wherein the first pattern set is generated when the pattern source element is controlled by at least one first external sideband signal whose transfer rate is lower than that of the second signal.

152. The component of claim 147, further comprising a compare element coupled to the first receiver and the first pattern source element, wherein the compare element compares the first pattern set to the second pattern set.

153. The component of claim 147, further comprising a memory array coupled to the first pattern source element, wherein the second pattern set resides in the memory array.

154. The component of claim 153, wherein at least one pattern set is written from the first pattern source element to the memory array.

155. The component of claim 153, wherein at least one pattern set is read from the memory array.

156. The component of claim 153, further comprising a compare element coupled to the first receiver and the memory array, wherein the compare element compares the first pattern set to the second pattern set.

157. An electronic system comprising: a first component comprising a first transmitter configured to drive a first signal onto a first external interconnect according to a set of transmitter specification parameters; and a second component comprising a first receiver configured to sample the first signal from the first external interconnect according a set of receiver specification parameters, wherein, the first component and the second component are configurable to operate in a normal mode in which the first signal is configurable to be driven by the first component and sampled by the first receiver; the first component and the second component are further configurable to operate in a first test mode in which the first receiver evaluates the transmitter specification parameters of the first transmitter; and the first component and the second component are further configurable to operate in a second test mode in which first transmitter evaluates the receiver specification parameters of the first receiver.

158. The system of claim 157, wherein the first component and the second component are each semiconductor components configured to operate at frequencies of 1 GHz or greater.

159. The system of claim 157, wherein the first component and the second component are each semiconductor components configured to operate at frequencies of 3 GHz or greater.

160. The system of claim 157, wherein system evaluation of the first component is performed in the first test mode.

161. The system of claim 157, wherein system evaluation of the second component is performed in the second test mode.

162. The system of claim 157, wherein the first component further comprises a second receiver coupled with a second interconnect to the first transmitter, wherein the first component is further configurable to operate in a third test mode in which the second receiver evaluates the transmitter specification parameters of the first transmitter.

163. The system of claim 162, wherein the second interconnect is an internal interconnect of the first component.

164. The system of claim 162, wherein the second interconnect is an interconnect external to the first component.

165. The system of claim 162, wherein wafer evaluation of the first component is performed in the third test mode.

166. The system of claim 162, wherein package evaluation of the first component is performed in the third test mode.

167. The system of claim 157, wherein the second component further comprises a second transmitter coupled with a second interconnect to the first receiver, wherein the second component is further configurable to operate in a third test mode in which the second transmitter evaluates the receiver specification parameters of the first receiver.

168. The system of claim 167, the second interconnect is an internal interconnect of the second component.

169. The system of claim 167, wherein the second interconnect is external to the second component.

170. The system of claim 167, wherein wafer evaluation of the second component is performed in the third test mode.

171. The system of claim 167, wherein package evaluation of the second component is performed in the third test mode.

172. The system of claim 157, further comprising at least one third component coupled to the external interconnect, wherein the first component is further configurable to operate in a third test mode in which the first transmitter is coupled with a second external interconnect to a second receiver on the third component, so that the second receiver evaluates the transmitter specification parameters of the first transmitter.

173. The system of claim 172, wherein wafer evaluation of the first component is performed in the third test mode.

174. The system of claim 172, wherein package evaluation of the first component is performed in the third test mode.

175. The system of claim 172, wherein the third component is substantially the same as the first component.

176. The system of claim 157, further comprising at least one third component coupled to the external interconnect, wherein the second component is further configurable to operate in a third test mode in which the first receiver is coupled with a second external interconnect to a second transmitter on the third component, so that the second transmitter evaluates the receiver specification parameters of the first receiver.

177. The system of claim 176, wherein wafer evaluation of the second component is performed in the third test mode.

178. The system of claim 176, wherein package evaluation of the second component is performed in the third test mode.

179. The system of claim of claim 176, wherein the third component is substantially similar to the second component.

180. A device, comprising: a test circuit; a first element, coupled to the test circuit, including a transmitter; a second element, coupled to the first element and the test circuit, including a receiver; and means for testing the first element using the second element and the test circuit.

181. A device, comprising: a test circuit; a first element, coupled to the test circuit, including a transmitter; a second element, coupled to the first element and the test circuit, including a receiver; and means for testing the second element using the first element and the test circuit.