Patent Application
20090027075
The present invention relates to testing analog driver circuitry using digital scan-based test methodologies such as level-sensitive scan design (LSSD) or any flip-flop based scan technique. Digital scan-based test methodologies typically define a number of Shift Register Latches (“SRLs”) that are electrically coupled in series and may be set deterministically by digital test equipment, such that application of a test vector defines the state of the device under test. Separate SRLs may be driven by outputs of the device under test. After the test is complete, the test equipment scans the output vector out of the SRLs and compares it against an expected output vector. SRLs which sense the output of a circuit under test can only interpret digital states and cannot interpret an analog state. Digital scan-based testing methodologies are well known in the art, and in particular, the implementation of a digital scan architecture into a design, the generation of patterns for testing the design, and the stimulation and observation of circuits within the design is well known. For example, see U.S. Pat. No. 4,298,980, entitled LSI Circuitry conforming to level sensitive scan design (LSSD) rules and method of testing same, by Hajdu et al. Use of SRLs for scan-based testing is well known to those skilled in the art. For example, see Miron Abramovici et al., “Digital Systems Testing and Testable Design”, IEEE PRESS, Ch. 9.3-9.9, 1990.
Analog driver circuitry typically requires functional testing to ascertain whether the driver functions as expected. As technology integration has become more complex, for example the integration of analog and digital circuits in the same design, testing of analog circuitry has become more difficult and expensive. There are several disadvantages associated with functional testing, such as: (1) it is more difficult to debug a functional test of a large analog/digital circuit as compared to digital scan-based logical paths; (2) it is more expensive to package a device for functional test than to test the device at the wafer level; and (3) more generally, development cost, difficult debugging and the high cost of test hardware make functional testing unappealing. For example, as operating voltages continue to decrease (e.g. 1.2V and below) and as the precision of digital-to-analog converters (DACs) increase, it is extremely difficult to digitally test the least significant bits (LSBs) of DACs where the DAC current output is converted to a voltage with a very low impedance (e.g. 50 ohms). For illustrative purposes, consider a seven-bit DAC which is operated at 1.2V. The output differential amplitude of the DAC for the LSB would be 9 mV at an output termination impedance of 50 ohms (1.2V /127). The offset voltage of a differential receiver is typically much larger than the LSB of the DAC, thus making the receiver's interpretation of the driver output unreliable. It is therefore desirable to provide a system and apparatus which is capable of testing an analog driver circuit of a mixed-signal design at the wafer-level, avoiding the extra cost associated with analog test equipment or on-chip analog BIST circuits.
See the following US patents for examples of analog test equipment, which are typically expensive and add to circuit test time: U.S. Pat. No. 5,610,530, entitled Analog interconnect testing, by Whetsel; U.S. Pat. No. 5,481,471, entitled Mixed signal integrated circuit architecture and test methodology, by Naglestad et al.; U.S. Pat. No. 5,107,208, entitled System for partitioning and testing submodule circuits of an integrated circuit, by Lee; and U.S. Pat. No. 6,011,387, entitled Analog autonomous test bus framework for testing integrated circuits on a printed circuit board, by Lee. See the following references for examples of analog BIST circuits implemented on chip, which are expensive in design time and chip area: U.S. Pat. No. 5,861,774, entitled Apparatus and method for automated testing of a programmable analog gain stage, by Blumenthal; and Karim Arabi, Bozena Kaminska, and Janusz Rzeszut, “BIST for D/A and A/D converters,” IEEE Design & Test of Computers v. 13 n. 4 Winter 1996, pp. 40-49.
The present inventors believe that the prior art has at least the following drawbacks:
An object of the invention is to digitally test an analog driver circuit.
Another object of the invention is to digitally test an analog driver circuit using a scan-based testing methodology.
Another object of the invention is to digitally stimulate and observe an analog driver circuit from shift-register latches.
Another object of the invention is to digitally stimulate and observe the least-significant-bit of an analog driver circuit by skewing the output termination impedance of an analog driver circuit.
Another object of the invention is to digitally test an analog driver circuit by converting the differential output of an analog driver circuit into a common mode signal by skewing the output termination impedance of the analog driver circuit, detecting the skewed signal with a differential receiver circuit and converting the detected skewed signal into a single-ended signal for observation.
Another object of the invention is to digitally test an analog driver circuit by detecting a skewed signal with a differential receiver circuit, wherein the receiver circuit incorporates a comparator circuit with built-in offset voltage for reliably detecting the state of the driver circuit.
Another object of the invention is to either skew or match the output termination impedance of a differential driver circuit in response to a test enable signal, wherein the impedance is skewed when the test enable signal is active and matched otherwise.
Another object of the invention is to receive a skewed common mode signal with a differential receiver circuit and convert the detected signal into a single-ended signal in response to a test enable signal.
Another object of the invention is to disable an up-channel receiver, enable the up-channel receiver for up-channel signaling, or enable the up-channel receiver for digital scan-based testing, all in response to a test enable signal.
Programmable termination impedance network 120 sets the termination impedance at differential output nodes ZP and ZN. During digital scan test mode (e.g. LSSD testing), programmable termination impedance network 120 skews the termination impedance at differential output nodes ZP and ZN in response to signals RESCTRL_P<5:0> and RESCTRL_N<5:0>. When the termination impedance is skewed at differential output nodes ZP and ZN, the differential output of driver 110 (DOP and DON) is skewed, and thus, up-channel receiver 130 is capable of detecting the full dynamic range of driver 110. Skewing the output termination impedance serves to covert DOP and DON into a common-mode signal. During normal operation, programmable termination impedance network 120 programs the termination impedance at output nodes ZP and ZN in response to signals RESCTRL_P<5:0> and RESCTRL_N<5:0> to the appropriate termination resistance for optimum transmission (e.g. matched 50 Ohm impedance at ZP and ZN).
Up-channel receiver 130 may be any receiver circuit with voltage offset such that when inputs to the receiver are equal, the receiver reliably outputs a low voltage. Alternatively, up-channel receiver 130 may incorporate hysteresis for improved noise immunity during up-channel signaling. Up-channel signaling is well known in the art. For example, see U.S. patent application Ser. No. 10/604025, entitled Data Transceiver And method for Equalizing The Data Eye of a Differential Input Data Signal, assigned to IBM Corporation. In the preferred embodiment, up-channel receiver 130 is a differential receiver with a built-in offset voltage and hysteresis, and is capable of receiving signals transmitted up-channel (i.e. in the opposite direction as the data being transmitted). Such receivers are incorporated in high speed serializer-deserializer (SERDES) circuit designs for up-channel signaling, and thus, circuit area may be minimized when an up-channel receiver with hysteresis is already present in a design. Up-channel receiver 130 is used to complete the observation path from the output of driver 110 to a SRL or a Primary output during digital scan-based testing. During normal operation, up-channel receiver 130 may sense a signal as low as 70 mv. However, there will be some ambiguity at these low levels. Since digital scan-based test mode requires deterministic results of either a logic ‘1’ or ‘0’ without ‘x’ propagation, the area of ambiguity must be removed. This is accomplished by changing the characteristics of up-channel receiver 130 during test mode. When up-channel receiver 130 is placed in test mode by LSSD_EN, offset and gain characteristics of the receiver are changed such that the receiver now requires a 150 mv signal or the output will result in a ‘0’. This increased voltage offset (−70 mv) guarantees that when all DACs are turned off in driver 110, there will be a logic ‘0’ output from up-channel receiver 130 at node Z. Additionally, by skewing the termination impedance at nodes ZP and ZN during test mode, the LSB of driver 110 can be observed as a logic ‘1’ at the output of up-channel receiver 130.
LSSD Test Mode Control Block 140 enables scan-compatible driver circuit 100 of
During digital test mode, up-channel receiver 130 is enabled for digital testing by LSSD_EN and powered on by PDWN. Up-channel receiver 130 functions as an Analog-to-Digital Converter (ADC) and is capable of passing a digital value Z, which represents the output of differential driver 110, to a shift register latch (“SRL”) which is observable during digital testing. During digital test mode, the termination impedance is skewed, thus enabling up-channel receiver 130 to detect the full dynamic range of differential driver 110 even though the least-significant-bits (LSBs) effect a much smaller change than up-channel receiver 130 could detect during normal operation. Detecting the LSB of differential driver 110 enables the digital-to-analog converters (DACs) residing within differential driver 110 to be fully digitally modeled (e.g. as a number of n-bit OR gates), thus enabling an automatic test pattern generator to generate test patterns for the entire DAC circuit. If the termination impedance is not skewed, the voltage offset of up-channel receiver 130 will be much larger than the LSB of driver 110, making the receiver's interpretation of the output of driver 110 unreliable. By skewing the termination impedance, the LSB of the DAC may be tested using a scan-based digital test methodology. During normal operation, up-channel receiver 130 is taken out of digital test mode by disabling LSSD_EN. Additionally, up-channel receiver 130 may be powered down by activating PDWN or may remain powered on by leaving PDWN deactivated.
Each paralleled output stage has a separate predriver, driven from the same driver logic (XNOR circuits 220 and buffer circuits 240). The number of segments per differential data bit determines the drive strength for that differential data bit. The more segments, the higher the drive strength for the programmed current. Any segment may be disabled via the SEG_DISABLE<x:0> signal. The output drive strength of output drivers 260A-260D is set by current DAC (IDAC) circuit 270. IDAC circuit 270 comprises IDACs 270A-270D. The value of each IDAC 270A-270D is set by TX COEFFX<x:0>. As illustrated in
Resistor circuit 310 provides termination impedance to output node ZP of differential driver 110 of
During normal operation, signals RESCTRL_P<5:0> and RESCTRL_N<5:0>program transistor switches 314 and 324 so that the output termination impedance at nodes ZP and ZN match. This occurs when RESCTRL_P<5:0>=RESCTRL_N<5:0>. For illustrative purposes, consider a differential driver that comprises a seven-bit DAC where: the LSB of the 7 bit DAC=1.2V/127=9 mV in terms of driver output differential amplitude at 50 ohms and 90 mV at 500 ohms. LSB of 4/5/6 bit DAC=19 mV of Driver output differential amplitude at 50 ohms and 190 mV at 500 ohms. Up-channel receiver 130 of
First built-in offset voltage comparator 408 comprises p-fet transistors TEL, TPL, and TNL. The comparator has a built-in offset voltage (“Vth−”). The value of Vth− is dependent upon the ratio of TNL to TPL as well as the biasing current IBIAS. In general, the higher the ratio or higher the biasing current, the higher Vth− becomes. As previously stated, this offset voltage provides improved noise immunity. Without this offset voltage, a comparator would output a high when VINP>VINN, output a low when VINP<VINN, and output an unknown value when VINP=VINN. Such a comparator has zero noise margin due to the fact that a minimal amount of noise triggers a response. First built-in offset voltage comparator 408 addresses Low to High transitions at VINP and VINN. VINP corresponds to DOP and VINN corresponds to DON as illustrated in
Second built-in offset voltage comparator 410 comprises p-fet transistors TER, TPR, and TNR. The comparator has a built-in offset voltage (“Vth+”). The value of Vth+ is dependent upon the ratio of TNR to TPR as well as the biasing current IBIAS. In general, the higher the ratio or higher the biasing current, the higher Vth+ becomes. As previously stated, this offset voltage provides improved noise immunity. Second built-in offset voltage comparator 410 addresses High to Low transitions at the inputs VINP and VINN. Differential amplifier circuit 412 comprises n-fet transistors TINP, TINN and TBIAS. Output stage 414 comprises p-fet transistor TNR2, n-fet transistor T2, and inverter circuit 416. Transistors TNR2 and T2 supply an input signal to inverter circuit 416 and the inverter inverts that signal. Differential amplifiers, output stages and inverters are all well known to those skilled in the art, and therefore, no further description is necessary. Common level shifter 404 transitions the output of inverter circuit 416 from VTT to VDD and outputs a signal Z.
Up-channel receiver 400 is either disabled, enabled during up-channel signaling (e.g. while performing equalization), or enabled during digital scan-based testing. The receiver is disabled when VBIAS is a logic ‘0’. VBIAS is an inverted version of the PDWN signal as illustrated in
When up-channel receiver 400 is enabled during up-channel signaling, both first built-in offset voltage comparator 408 and second built-in offset voltage comparator 410 are enabled. First built-in offset voltage comparator 408 is enabled when LSSD-EN signal is deactive (logic ‘0’). Second built-in offset voltage comparator 410 is always enabled because the gate of p-fet transistor TER is tied to ground. In this mode, up-channel receiver 400 outputs a logic ‘1’ at Z when VINP−VINN >Vth+, outputs a logic ‘0’ when VINP−VINN <Vth−, and remains in its previous state when Vth− <VINP−VINN<Vth+. First built-in offset voltage comparator 408 and second built-in offset voltage comparator 410 both have memory, which is desired for improving noise immunity, but not permitted during digital scan-based testing.
When up-channel receiver 400 is enabled during digital scan-based testing, first built-in offset voltage comparator 408 is disabled and second built-in offset voltage comparator 410 is enabled. First built-in offset voltage comparator 408 is disabled when LSSD_EN signal is active (logic ‘1’). By disabling first built-in offset voltage comparator 408, hysteresis is turned off, thus satisfying the digital scan-based methodology requirement of not having memory. When first built-in offset voltage comparator 408 is disabled, second built-in offset voltage comparator 410 functions as a built-in offset-voltage comparator. Therefore, up-channel receiver 400 outputs a logic ‘1’ if and only if VINP−VINN>Vth+. Due to process, temperature, and supply variation, Vth+ will vary with some range (e.g. Vth+_lower to Vth+_upper). For illustrative purposes, Vth+_lower=70 mV and Vth+_upper=150mV. This variation in Vth+ is addressed by skewing the output termination impedance as previously described and as illustrated in
MUX 502 selects either NOMINAL_RESCTRL_P<5:0> or hard-wired value ‘0111111’ in response to LSSD_EN. MUX 504 selects either NOMINAL_RESCTRL_N<5:0> or hard-wired value ‘111101’ in response to LSSD_EN. During normal operation, LSSD_EN is deactivate (logic ‘0’) and MUX 502 selects NOMINAL_RESCTRL_P<5:0> and MUX 504 selects NOMINAL_RESCTRL-N<5:0>. As previously described, NOMINAL_RESCTRL_P<5:0> and NOMINAL_RESCTRL_N<5:0> program the output termination impedance so that the impedance matches at nodes ZP and ZN of
MUX 510 selects the drive strength signal to be supplied to power DAC 280 of
MUX 512 selects either NOMINAL_SEG_DISABLE<7:0> or a decoded signal supplied from decoder 514 in response to LSSD_EN. During digital testing, only one segment of the differential driver may be tested during a test sequence (e.g. load test patterns, run system clock, unload test results). Test patterns that represent which segment is to be tested during a test sequence are provided to decoder 514 from SRLs 516 during digital testing. Decoder 514 may be any standard 3 to 8 decoder and decodes the signals provided by SRLs 516. For a differential driver that contains eight segments, three SRLs (i.e. ) supply test patterns to decoder 514. These three inputs to decoder 514 enable the decoder to provide eight outputs to the differential driver when being tested, each output representing which one of eight segments within the differential driver will be activated during a particular test time. Test patterns are loaded into the SRLs, or “scanned in”, each time a different segment is to be tested. For example, when
=0 0 0, SEG_DISABLE<7:0>=<11111110>; when
=0 0 1, SEG_DISABLE<7:0>=<11111101>; when
=0 1 0, SEG_DISABLE<7:0>=<11111011>; etc.; and finally when
=1 1 1, SEG_DISABLE<7:0>=<01111111>. The decoded signal is selected by MUX 512 and transmitted to the differential driver when LSSD_EN is active. During normal operation, MUX 512 selects NOMINAL_SEG_DISABLE<7:0>. The number of SRLs required to provide test input to the decoder and the size of the decoder depend upon the number of segments contained within the differential driver. Thus, any size decoder and number of SRLs for inputting test signals to the decoder are within the scope of this invention. OR circuit 518 is optional and allows the differential driver to be placed in tristate mode when TS is active. Placing a driver in tristate mode is well known in the art, and thus, is not discussed in further detail.
Logic circuit 520 either inhibits or gates NOMINAL_PDWN signal depending upon the value of LSSD_EN. When in digital scan-based test mode, LSSD_EN is active and inverted to a logic ‘0’ at the input to logic circuit 520, thus inhibiting NOMINAL_PDWN and resulting in PDWN being a logic ‘0’. This assures that up-channel receiver 130 of
During digital scan-based testing, the decoder determines which segment is to be activated and activates that segment at 640 in accordance with the present invention and as previously described. The differential driver circuit is stimulated from SRLs (e.g. LSSD or GSM) at 650. During digital scan-based test modeling, the differential driver is modeled, for example as a number of n-bit OR gates, and digital test patterns are automatically generated (ATPG) for testing the components of the driver for stuck-at ‘1’ or ‘0’ faults. These test patterns provide stimulus to the driver circuit at 650 during digital scan testing. The stimulus represented by the TX_COEFFx<x:0> signals, which are generated during ATPG, are stored in SRLs and stimulate the differential driver circuit during digital scan-based testing. The four bits of differential data input signal AP<3:0> as illustrated in
In an alternate embodiment, up-channel receiver 130 of
The description of the present invention has been presented for purposes of illustration and description but is not intended to be exhaustive or limited to the invention in the form disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiment was chosen and described in order to best explain the principles of the invention and the practical application, and to enable others of ordinary skill in the art to understand the invention for various embodiments with various modifications as are suited to the particular use contemplated.
This application is a divisional application of a U.S. patent application Ser. No. 10/708,788, entitled “SYSTEM AND METHOD OF DIGITALLY TESTING AN ANALOG DRIVER CIRCUIT”, and filed Mar. 25, 2004.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 10708788 | Mar 2004 | US |
| Child | 12189226 | US |
