Patent Grant
8886987
Embodiments according to the present invention relate to a data processing unit and a method of processing data.
Some embodiments according to the invention relate to a tester for testing a device under test (DUT).
A device under test (also designated briefly as “DUT”), for example an integrated circuit, is tested in order to ensure a correct operability of the device. Usually, a DUT is tested with an automatic test equipment (ATE). Test patterns or test stimuli are applied to the DUT by the ATE, which is in the following also briefly named “tester”, and electrical output signals from the DUT are compared to expected data. With this method, different DUT parameters can be proven or characterized, and if necessary the electronic devices (DUT) are split into one or more different categories, for example pass or fail categories or performance categories. In the production test, the influence of different parameters on a proper operation of the DUT can be checked or proven. Such parameters may be, among others, temperature, humidity, over-voltage or, for example, tight timing parameters.
The output timings of digital signals sent from a DUT to the tester can be very tight. The digital signals sent from a DUT to the tester may be coupled to (or synchronized with) a central master clock from the tester, or may be source-synchronous signals. For source-synchronous signals, the bus clock can be generated and provided by the DUT.
If a DUT output arrival time (for example a time at which a data signal from a DUT arrives at an input of a tester, for example referenced to a master clock of the tester) is not known exactly, or if the DUT output arrival time is not stable during a test, the ATE does not know exactly where to place a strobe time (or strobe point-in-time) in order to sample the received data from the dut, for example to compare the received data from the DUT with expected data.
For example, the capture of data from a source-synchronous bus with clock or data strobe signal can be difficult, when clock and/or strobe (or clock/strobe) and data are separated by long propagation delays (or propagation delay differences), e.g. handled by different chips or boards, for example in an automated test equipment (ATE). In other words, in some cases it is a challenge to strobe the data in a correct manner in such an environment.
A data strobe channel may for example comprise a (programmable) routing in order to adjust a propagation delay (or a propagation delay difference) between a strobe signal and a data signal. However, this programmable routing of the data strobe channel (which may process the data strobe signal) may sometimes lead to delays that are too long to still capture the data safely.
A tester may comprise a strobe signal channel and a plurality of data signal channels. In a conventional way, the data paths for the plurality of data signals may comprise buffer delays in order to compensate a propagation delay of the strobe signal routing. The required delay matching between the strobe signal and the plurality of data signals can be very difficult, in particular across multiple chips, for example across multiple chips on a channel board for a test system, or across multiple chips connected to so-called dut-board or load board.
For example, in order to reduce the production test time, a plurality of DUTs on a respectively configured load board can be tested in parallel. The propagation delay (or propagation delay differences) between the different strobe and data signals can be compensated, at least partially, by the above-mentioned buffer delays.
In the US patent application US 2004/0022196 A1, another conventional method for compensating delays, for recovering a bitstream from a signal, is described. Said patent application describes an oversampling bit stream recovery, which is applicable to embedded clocks, where data and clock are embedded on one physical signal.
In view of the above, there is a need for a reliable concept for extracting data from a sequence of data in response to a clock or strobe signal.
According to an embodiment, a data processing unit may have a time information provider for processing a clock or a strobe signal, and at least one data extraction unit. The time information provider is configured to provide a digitized clock or strobe time information on the basis of the clock or strobe signal. The at least one data extraction unit is coupled to the time information provider and is configured to select data from a sequence of data samples of a data signal depending on the digitized clock or strobe time information, wherein the clock-or-strobe signal is physically different from the data signal; wherein the time information provider has a time-to-digital converter configured provide, as the digitized clock-or-strobe time information, a digitized time stamp describing a time of a transition of the clock-or-strobe signal, wherein the digitized clock or strobe time information is represented as a multi-bit value encoding the clock or strobe time information or wherein the digitized clock-or-strobe time information is represented as an increment-decrement-information, wherein the increment-decrement-information describes a difference between a timing of a currently detected transition of the clock-or-strobe signal and a timing of a previous transition of the clock-or-strobe signal.
The clock-or-strobe signal may optionally be physically separate from the data signal. For example, the clock-or-strobe signal may be received from a different pin of a device-under-test than the data signal, if the data processing unit is applied in a test system.
According to another embodiment, a tester for testing a device under test may have a master clock signal line, a strobe signal generator, a comparator unit, a device under test (DUT) clock generator, a data processing unit as mentioned above and a device under test interface. The tester is configured such that the strobe signal generator generates a strobe signal for the data processing unit on the basis of the master clock signal, and such that the DUT clock generator generates a DUT clock signal coupled to the DUT interface on the basis of the master clock signal. The data processing unit is coupled via at least one data signal line to the DUT interface to receive the data signal. The data processing unit is also coupled to a comparator unit. The comparator unit is configured to compare the selected data from the data processing unit with expected data and to provide a test result based on the comparison of the selected data with the expected data.
According to another embodiment, a tester for testing a device under test (DUT) may have: a comparator unit; a device-under-test (DUT) interface; and a data processing unit as mentioned above; wherein the data processing unit is coupled to the DUT interface to receive from the DUT interface the data signal and the clock-or-strobe signal, and to select data from the sequence of data samples of the data signal, in response to the clock-or-strobe signal; and wherein the comparator unit is configured to compare the selected data provided by the processing unit with expected data, and to provide a test result based on the comparison of the selected data with expected data.
According to another embodiment, a method of processing data may have the steps of processing a clock or strobe signal to provide a digitized clock or strobe time information on the basis of the clock or strobe signal and selecting data from a sequence of data samples of a data signal depending on the digitized clock or strobe time information, wherein the clock-or-strobe signal is physically different from the data signal; wherein providing the digitized clock or strobe time information has providing, as the digitized clock or strobe time information, a digitized time stamp describing a time of a transition of the clock-or-strobe signal; wherein the digitized clock or strobe time information is represented as a multi-bit value encoding the clock or strobe time information or wherein the digitized clock-or-strobe time information is represented as an increment-decrement-information, wherein the increment-decrement-information describes a difference between a timing of a currently detected transition of the clock-or-strobe signal and a timing of a previous transition of the clock-or-strobe signal.
Another embodiment may have a computer program for performing a method of processing data as mentioned above, when the computer program runs on a computer.
Some embodiments according to the invention may be partly or entirely implemented in hardware.
Some embodiments may alternatively or additionally be partly or entirely embodied or supported by one or more suitable computer programs, which can be stored on or otherwise provided by any kind of data carrier. The computer program may run on a computer or any suitable data processing unit.
Various embodiments according to the invention will be explained below in more detail with reference to the accompanying figures, wherein:
a to 4c show schematic representations of a concept of digitizing a strobe time and selecting data on the basis of the digitized strobe time, according to embodiments according to the invention;
With reference to the accompanying
In the following, a conventional approach of compensating a propagation delay difference between a data signal and a data strobe signal will be described in order to facilitate an understanding of the inventive concept.
In order to compensate the delay caused by a data strobe routing, a conventional data processing unit may comprise a couple of buffer delays. A block schematic diagram of such a conventional data processing unit 10 is shown in
The data processing unit 10 comprises a strobe channel 1a configured to receive a strobe signal S. The strobe channel 1a may comprise buffer delays 6, 7 in order to shift or adjust (e.g. time-shift or time-adjust) the strobe signal S relative to a data signal. In
Achieving the needed delay matching in a conventional data processing unit is very difficult, in particular across multiple chips and a plurality of data signals D1 to Dn. For example, providing a long delay may result in a distortion of the signals. The data signals may be send from a device under test. The usual way to match different data signals and clock/strobe signals is adjusting the propagation delay with tunable delay lines. One of the drawbacks of this conventional approach is the time-consuming, complicated and less flexible way of performing this compensation.
The embodiment shown in
The data processing unit 20 may optionally comprise a plurality of data extraction units 24 configured to select data from a sequence of data samples of corresponding data signals. The one or more data signals D1, together with the one or more strobe or clock signals, may form a bus or a link. A test system or a tester (ATE) may optionally comprise a plurality of such busses or links. Each of the data extraction units may be coupled to a separate time information provider and may be configured to select data from the sequence of data samples of the respective data signal depending on the digitized clock or strobe time information provided by the corresponding time information provider.
The clock-or-strobe time information may be forwarded to one or more data extraction units 24 using a “Daisy Chain” chain configuration. For example, the clock-or-strobe time information may be forwarded from the time information provider to a first data extraction unit in a first cycle, from the first data extraction unit to a second data extraction unit in a subsequent second cycle, and so on. Alternatively, a programmable routing may be provided to route the clock-or-strobe time information from the time information provider to one or more data extraction units. For example, a data path from the time information provider to a first data extraction unit may be enabled in a first cycle, to transfer the clock-or-strobe time information to the first data extraction unit. Subsequently, a data path from the time information provider to a second data extraction unit may be enabled in a second cycle, to transfer the clock-or-strobe time information to the second data extraction unit. Alternatively, a star configuration may be used for the transport of the clock-or-strobe time information from the time information provider to a plurality data extraction units, such that, for example, the clock-or-strobe time information can be transferred from the time information provider to a plurality of data extraction units simultaneously. However, other configurations are naturally possible.
The clock or strobe time information STI may be forwarded as a digital word. In some embodiments, the digitized clock or strobe time information may be represented as one or more discrete coded values. The digitized clock or strobe time information forwarded from the time information provider 22 to the at least one data extraction unit 24 may be represented as a multi-bit value encoding the clock or strobe time information. The at least one data extraction unit 24 may be coupled to the time information provider 22 in a daisy chain, a parallel or in a serial configuration. This means that the digitized clock-or-strobe time information may be forwarded to the at least one data extraction unit in parallel or in a serial way, i.e. for example during several clock periods applied to the data processing unit.
Alternatively, the clock-or-strobe time information may be forwarded in the form of a digitized information describing a deviation between an initial clock-or-strobe time and an updated clock-or-strobe time. For example, a digitized signed numerical value may be transferred, which represents a change of the clock-or-strobe time with respect to an initial value (or a previously used value). In some embodiments, an increment/decrement information may be transferred from the time information provider to the one or more data extraction units. Thus, two signal lines may be sufficient in order to transfer an information indicating whether the clock-or-strobe time should be increased, decreased or held constant. In other words, a information comprising only three states (increase clock-or-strobe time, decrease clock-or-strobe time and hold clock-or-strobe time constant) or even only two states (increase clock-or-strobe time, decrease clock-or-strobe time) may be used to communicate the digitized clock or strobe time information STI from the time information provider to the one or more data extraction units.
In some embodiments, a rate at which the digitized clock-or-strobe time information (STI) is transferred from the time information provider to the one or more data extraction units may be synchronized with a master clock. For example, a digitized clock-or-strobe time information may be transferred once per master clock cycle.
In other embodiments, it may be sufficient to transfer the digitized clock-or-strobe time information only once per I master clock cycles (or clock-or-strobe signal periods), wherein I>1. In some embodiments, I may be larger than 10, or even larger than 100 or larger than 1000. Accordingly, a significant amount of energy (which would be needed for a permanent exchange of the clock-or-strobe signal) can be saved in some cases.
In some embodiments, the digitized clock-or-strobe time information may be transferred from the time information provider to the data extraction unit only once per J data positions of the data signal, wherein J>1. For example, a data position may designate a time interval of the data signal during which a single value or a single bit is transmitted via the data signal.
In some embodiments, digitized clock-or-strobe time information (or an information indicating a change of the clock-or-strobe time information) may be transferred in response to a change of the clock-or-strobe time information.
The clock-or-strobe time information (STI) may be transferred synchronously or asynchronously with respect to the master clock. For example, the digitized clock-or-strobe time information may be transferred using a bus system which is independent from the master clock. Also, there are many different methods for obtaining the digitized clock-or-strobe time information in the time information provider. In general, the time-information-provider may be configured to provide a digitized time stamp indicating a time of a transition in the clock-or-strobe signal. Devices for providing such a digitized time stamp describing a timing of an edge are also known as time-to-digital converters (TDC). In other words, time-to-digital converters (TDC), which may for example be used in automated test equipment, time-stamp selected events (for example an edge or transition) in a signal (for example from a device-under-test). For example, a time-to-digital converter may be configured to measure a time of a transition relative to a fixed (or stable) tester clock.
Different architectures may be used to provide the digitized clock-or-strobe time information. For example, an ADC sampling may be used. The ADC sampling may interpolate samples of an analog-to-digital converter around zero-crossings or perform a digital down-conversion to track phase changes.
Alternatively, a ramp interpolation may be used to provide the digitized clock-or-strobe time information. The ramp integration may integrate a current during a measured time and digitize the end voltage using an Analog-to-digital converter (ADC).
Alternatively, a delay line may be used to provide the digitized clock-or-strobe time information. A delay line TDC may inject a measured transition into a chain of gate delay elements and capture the chain's state with the next coarse clock.
Alternatively, a sampling offset may be used to provide the digitized clock-or-strobe time information. Sampling offset (SO) TDC achieves a fine resolution by exploiting a random sampling offset mismatch between multiple parallel flip-flops.
Alternatively, a vernier oscillator may be used to provide the digitized clock-or-strobe time information. A vernier oscillator TDC may use two counted oscillators with slightly different periods τs, τf, where τs>τf, to achieve a fine resolution of Δτ=τs−τf. A measured event (for example an edge or transition of the clock-or-strobe signal) may trigger the slow oscillator. Time t later, a next coarse clock (for example a master clock) triggers the fast oscillator. After c=t/Δτ slow cycles the fast oscillator has cycled more often than the slow oscillator.
Alternatively, a vernier delay line may be used to provide the digitized clock-or-strobe time information.
Alternatively, a DLL-stabilized vernier delay line may be used to provide the digitized clock-or-strobe time information.
Alternatively, a double-resolution time quantizer, a vernier oscillator array or a pulse shrinking delay line may be used to provide the digitized clock-or-strobe time information on the basis of the clock-or-strobe signal.
Some of the above mentioned methods for providing the digitized clock-or-strobe signal are summarized in the article “Fully-Digital Time-to-Digital Converter for ATE with Autonomous Calibration” by J. Rivoir (published: IEEE International Test Conference (ITC), Santa Clara (Calif.), USA, Oct. 24-26, 2006, Paper 6.3). Some possibilities for obtaining the digitized clock-or-strobe signal (i.e. for providing a digitized information describing a timing of an event, like a transition) are also described in the references cited in said article.
In the embodiment shown in
In the embodiment depicted in
This means that, depending on the digitized clock or strobe time information, which is forwarded from the time information provider 22 via the connection 25 to the data selection unit 22c, data E1 (for example one bit) is selected from the sequence of data samples of the data signal D1, which sequence is temporarily stored in the second bit history unit 24b.
It should be noted that the terms “STI”, “transition time information” and “digitized clock or strobe time information” are used interchangeably within this text.
The digitized clock or strobe time information may be forwarded to further data extraction units connected to the transition detection unit in order to select data from other data signals.
As it is shown in
In order to influence the selection of data from the sequence of data samples, the correction unit 30 may be coupled to the connection 25 to receive the digitized clock or strobe time information. The correction unit 30 may be configured to change or modify the digitized clock or strobe time information provided on the electrical connection 25, to obtain a modified digitized clock or strobe time information, and to provide the modified digitized clock or strobe time information to the multiplexer 24c. The modification of the digitized clock or strobe time information performed by the correction unit 30 may therefore have an influence on the selection of the data E1 out of the sequence of data samples of the data signal D1 temporarily stored in the second bit history unit 24b. In other words, assuming that the digitized clock or strobe time information is a digitally encoded (e.g. binary-encoded, BCD-encoded, ternary-encoded, or encoded according to any other digital data representation) information describing a time at which an edge or transition of the clock or strobe signal S occurs, the correction unit 30 may modify said digitized clock or strobe time information, for example by increasing or decreasing a value represented by the digitized clock or strobe time information. The increased or decreased value may be used to encode, which bit (or which bits) out of a plurality of bits of the data sequence stored by the second bit history unit 31 is (are) forwarded as the data E1.
In another embodiment, the correction unit 30 may be coupled to the second bit history unit 24b and/or to the data selection unit 24c. The correction unit may be configured to change or shift the sequence of the data samples stored in the bit history unit 24b (for example with respect to an original position). For example, if a correction information indicates that an additional shift of two sample periods should be introduced, the correction unit may cause the bit history unit to shift the samples provided by the second sampling unit 24a by two more bit positions, in addition to transferring said samples into the bit history unit 24b.
Alternatively, or in addition, the correction unit may be configured to change or influence the selection of data from the sequence of the data samples stored in the bit history unit 24b.
To summarize the above, there are various different possibilities, how the selection of the data E1 out of the sampled data sequence provided by the second sampling unit 24a can be influenced in dependence on a digital value represented by the digitized clock or strobe time information.
In some embodiments, the data select unit 24c may comprise a multiplexer configured to select data (e.g. one bit) from the sequence of data samples temporarily stored in the second bit history unit 24b depending on the digitized clock or strobe time information, to obtain certain selected data E1 (for example a selected bit). This means that the digitized clock or strobe time information provided by the time information provider (which may for example represent a digitized value) may for example be used to set or select a multiplexer selection address, such that a certain data sample (for example a certain bit) temporarily stored in the bit history register (out of the plurality of bits stored in the second bit history register 24b) is selected.
The correction unit 30 may for example change a digital word—representing the digitized clock or strobe time information—such that a different multiplexer selection address is applied to the multiplexer 24c. Hence a different data sample is selected (when compared to a simple embodiment in which the correction unit is absent or inactive).
It is also possible that the correction unit changes directly the data selected by the multiplexer (for example by shifting the data input to the multiplexer) in order to influence the selection of the data E1 to be selected.
The correction unit 30 may be configured to influence the selection of the selected data sample E1, such that a propagation delay of the clock or strobe signal S and the data signal D1 (or a propagation delay difference between these signals) is compensated. A correct “sampling” of the data signal D1 (for example a correct choice of a point-in-time in a temporal evolution of the data signal D1, at which the data signal D1 is evaluated), or a correct selection of a data sample of the data signal D1, can be achieved by adding to the digitized clock or strobe time information or subtracting from the digitized clock or strobe time information a certain (for example predetermined) digital value, the digital value (for example designated as “Corr”) representing the delay (or delay difference) between the clock or strobe signal S and the data signal D1. Thus, the correction unit may obtain a corrected digitized clock or strobe time information, for example for controlling the multiplexer 24c.
The correction unit 30 may be formed as an adder or subtracter in order to change the digitized clock or strobe time information (STI) in such a way that the propagation delay (or propagation delay difference) between the clock and strobe signal and the data signal is compensated. The correction unit 30 may change the digitized clock or strobe time information, and the changed digitized clock or strobe time information may be used for the selection of the data sample E1.
In the following, some further details will be described.
The first sampling unit 22a may be configured to digitize or sample the clock or strobe signal S with a first sampling frequency fs1, which may be n-times (n≧2) higher than a first maximum frequency fmax1 of the clock or strobe signal S.
The second sampling unit 24a may be configured to sample or digitize the data signal D1 with a second sampling frequency fs2, which may be m-times (m≧2) higher than the second maximum frequency fmax2 of the data signal D1, wherein the frequency fmax2 may be defined as an inverse of a minimum period duration of a data symbol of the digital signal D1. This means that the clock or strobe signal S may be n-times over-sampled, and that the at least one data signal D1 may be m-times over-sampled. The first and the second sampling unit can be considered as over-sampling units in an embodiment. In some embodiments, n may be different from m, while n may be equal to m in some other embodiments.
m and n may be integer numbers in some embodiments. However, it is not required that m and n are integer numbers.
In some embodiments according to the invention, the time, information provider 22 may comprise a plurality of flip-flops (e.g. one or more clocked shift register chains) to temporarily store the sequence of samples of the clock or strobe signal S provided by the first sampling unit 22a. The at least one data extraction unit 24 may also comprise a plurality of flip-flops (e.g. one or more shift-register chains) to temporarily store the sequence of data samples of the data signal D1 provided by the second sampling unit 24a.
In
The first and the second sampling units 22a, 24a can be clocked with the n-fold frequency compared to the first and the second bit history units. This means that the first and the second bit history unit 22b and 24b may be configured, as indicated by the arrow 27 in FIG. 3, to store temporarily at least n of the samples of the clock/strobe signal S or data signal D1, respectively. Therefore, in some embodiments n may be an integer value larger than 1.
The transition detection unit 22c in the time information provider 22 may be configured for detecting signal transitions in the (over-)sampled clock or strobe signal S temporarily stored in the bit history unit 22b.
The first bit history unit 22b and the second bit history unit 24b may be formed as a shift register which can at least store n-fold samples of the respective clock/strobe-signal or data signal.
However, other embodiments may optionally use bit history units which are clocked with a shift clock having the same clock rate as the sampling clock of the sampling units. Also, in some embodiments, the shift clock applied to the first bit-history unit 22b may be different in frequency from the shift clock applied to the second bit history unit 24b.
In some embodiments, the bit history unit 24b may optionally be configured to cover a longer period of time than the bit history unit 22b. For example, the bit history unit 24b may be configured to store a longer bit sequence than the bit history unit 22b. Alternatively, the bit history unit 24b may be clocked with a lower clock frequency than the bit history unit 22b. For example, the bit history unit 24b of the data extraction unit may be configured to store a plurality of periods of the data signal D1, while it may be sufficient to store a single period of the clock-or-strobe signal S in the bit history unit 22b. Thus, a propagation delay resulting from the forwarding of the digitized clock-or-strobe time information may be compensated. In other words, a propagation delay in forwarding the clock-or-strobe time information from the time information provider 22 to the data extraction unit 24 may be balanced by increasing the length of the bit history unit 24b of the data extraction unit.
Also, in some embodiments the correction unit 30 may optionally be used to (or configured to) compensate for a propagation delay in forwarding the clock-or-strobe time information from the time information provider 22 to the data extraction unit 24.
a to 4c show a schematic representations of the sampling of a clock or strobe signal S and a data signal D1.
In the embodiment shown in
The same applies to the data signal D1, which may also be 8-fold over-sampled to obtain a sequence of 8 data samples in time intervals A′-H′, which time intervals A′-H′ may or may not be identical to the time intervals A-H. In some embodiments, a temporal synchronization is provided, such that there is a predetermined temporal relationship between the time intervals A-H and A′-H′. However, in some embodiments there may be a time shift between time intervals A-H and A′-H′, for example due to propagation delays in the distribution of a synchronization signal. In other embodiments, time intervals A-H and A′-H′ may be concurrent.
The clock or strobe signal S and the data signal D1 may be analog signals or digital bit streams. One bit of such a bit stream may be n-times over-sampled (n≧2), for example 8-fold, as it is shown in
The clock or strobe signal S may perform a transition 46 from a low state to a high state during the sample time interval B. This transition is indicated by the arrow 42 and may be detected by the transition detection unit 22c. For example, the transition detection unit 23c may find that the transition 46 occurs during the j-th sample, relative to the (repetitive) timing reference 45. In the given example, the transition detection unit may find, on the basis of an analysis of the samples of the signal S, that the transition occurs during the second sample interval “B” after the timing reference 45. Consequently, the transition detection unit may provide a digitally encoded information describing the temporal position of the transition 46 relative to the (typically repetitive or periodic) timing reference 45, to obtain the digitized clock or strobe time information. In the given example, the digitized clock or strobe time information may encode the value j, for example j=2.
In an embodiment, the bit history unit 22b may be configured to cooperate with the first sampling unit 22a such that a position, at which bits provided by the first sampling unit 22a are stored in the bit history unit 22b, is dependent on a relative timing relationship between the bits provided by the first sampling unit and the timing reference. For example, the bit history unit may be configured to take over bits provided by the sampling unit 22a in response to the timing reference. Alternatively, an evaluation of the bits stored in the bit history unit 22b (for example by the transition detection unit) may be triggered in response to the timing reference.
The digitized clock or strobe time information obtained by the transition detection 22c may then be forwarded as a digital word, for example as a discrete binary coded value, to the data select unit (not shown in
In a source-synchronous bus of a DUT, a strobe signal S may be transferred together with a data signal in order to indicate for example valid data. In this case the strobe time can be digitized and transmitted. In subsequent steps the digital strobe time can be used to select data from a choice of data samples of the data signal, or to indicate to another device, for example an evaluating unit or a tester, which data sample is valid and represents the correct value of the data signal.
In the following, a situation will be described in which a data signal D1 is time shifted with respect to a corresponding clock signal S at a processing unit.
b shows a graphical representation 470 representing both the signal S and the signal D1 as sampled at the processing unit. It will be assumed here, for the sake of simplicity, that the first sampling unit 22a and the second sampling unit 24a comprise a common timing reference. An abscissa 472 describes the time, and an ordinate 474 describes levels or values of the signals S, D1 as seen at the processing unit 20. For example, an extra delay introduced to the data signal D1 when compared to the delay introduced to the clock-or-strobe signal S when forwarding the signals S, D1 from the source to the processing unit may cause that the edge 468 of the signal S does not coincide with a central portion 469 of the data signal D1 at the processing unit, as can be seen from the graphical representation 470. Assuming now that samples of the clock-or-strobe signal S are taken in time intervals A, B, C, D, E, F, G, H (the time intervals being defined relative to the common timing reference), an edge detection can be performed, for example, by the transition detection unit 22c. The edge detection unit may, for example, indicate that an edge occurs in the time interval “B”.
Assuming further that the data signal D1 is sampled at time intervals A′, B′, C′, D′, E′, F′, G′, H′, relative to the common timing reference, wherein A=A′, B=B′, . . . , it can be seen that sample values in concurrent time intervals are not associated with corresponding points of the signals S, D1 at the source (device-under-test). Rather, a correction may be needed. For example, it can be seen from
The multiplexer 24c may, for example, select the sample value of the data signal D1 in the time interval F′ as the selected data E1 in response to the corrected transition time information 31 provided by the correction unit 30.
A proper setting of the correction unit 30 may, for example, be obtained in a calibration procedure. For example, if a calibration is performed, which allows the extraction of the time shift between S and D1 introduced by the forwarding of the signals S and D1 from the source (device-under-test) to the processing unit, a correction value may be extracted therefrom.
In the following, a situation in which the time references for the sampling of the clock-or-strobe signal S and for the sampling of the data signal D1 are shifted with respect to each other, will be described taking reference to
Taking reference now to the graphical representation 490, an abscissa 493 describes a time, and an ordinate 494 describes the sampled values of the signals S, D1. Taking reference to the graphical representation 490, an edge detection on the basis of the samples of the clock-or-strobe signal S may indicate that an edge 468 is present in a sample interval E. However, the sample interval E of the clock-or-strobe signal S does not correspond to the sample interval E′ of the data signal D1 due to the shift of the timing reference for the sampling of the signals S and D1. Thus, the sample interval E of the clock signal S rather corresponds to the sample interval C′ (referenced to the timing reference for a sampling of D1).
Accordingly, the correction unit 30 may receive a transition time information indicating that the edge was detected in the sample interval E of the clock-or-strobe signal S, and the correction unit 30 may provide a corrected indication, indicating to select a bit sampled in the sample interval C′ of the data signal D1 as the selected data E1.
For example, if the sampled intervals are designated by integer numbers, the sample interval E may be represented by the integer number of 5 (with the sample interval A being represented by the integer number 1). The sample interval C′ may, for example, be represented by the integer number 3, with the sample interval A′ being represented by the integer number 1. Thus, the correction unit 30 may receive the integer number 5 (representing the sample interval E) and subtract the integer of 2 from the received value, to obtain a value of 3, representing the sample interval E′ of the data signal D1.
To summarize the above, a concept has been described taking reference to
In some embodiments, the correction unit 30 may optionally be used to correct for the propagation delay of the transfer of the digitized clock-or-strobe time information from the time information provider to the data extraction unit. For example, a transfer of the digitized clock-or-strobe time information from the time information provider to the one or more data extraction units may need, for example, one or more periods of the first sampling frequency fs1, or of the second sampling frequency fs2. Accordingly, the correction unit may be used to compensate for this propagation delay. Also, if the propagation times for the digitized clock-or-strobe time information between the time information provider and a plurality of data extraction units are different, the correction units 30 of the data extraction units may be set to different values.
In some embodiments, the correction unit may optionally be used to (or configured to) provide a pre-trigger functionality. Thus, a sample of the data signal D1 may be selected as the output data E1 (in response to a corresponding setting of the correction unit 30), such that the selected sample of the data signal describes a status of the data signal temporally before the occurrence of the corresponding transition on the clock-or-strobe signal.
Another embodiment according to the present invention is depicted in
Integrated circuits or devices may comprise up to several hundred data signals (for example, external data signals), which may, for example, form different busses or links with defined strobe or clock signals. In order to test such electronic devices, a tester may also comprise several hundred test channels. One or more of the tester channels may comprise time information providers and/or data extraction units. A tester or ATE may therefore comprise one or more data processing units, as described above.
In some embodiments, a single unit may be configured to be switchable to operate as a time information provider or as a data extraction unit. For example, a combined unit may comprise a sampling unit (for example a sampling unit 22a) and a bit history unit (for example a bit history unit 22b) coupled to the sampling unit (for example as described above). Moreover, the combined unit may comprise a transition detection (for example a transition detection 22c) and a multiplexer (for example a multiplexer 24c). In the combined unit, the transition detection and the multiplexer may be coupled to a common bit history unit. Thus, if the transition detection of the combined unit is active, the combined unit may serve as a time information provider. Further, if the multiplexer of the combined unit as active, the combined unit may serve as a data extraction unit. Accordingly, the combined unit may provide a maximum flexibility, such that a channel, to which the combined unit is coupled, may selectively operate as a clock-or-strobe channel or as a data channel.
Taking reference to
The different data extraction units 24, 26, 28 may each apply individual correction values to the common transition time information received from the timing information provider 22. Naturally, identical correction values could also be applied. Also, rather than distributing a common transition time information to the data extraction units 24, 26, 28, individually corrected transition time information could be provided by the time information provider 22 for the different data extraction units 24, 26, 28.
In embodiments, the strobe signal generator 64 and/or the DUT clock generator 68 may include a frequency divider or an adjustable delay line in order to time-shift the DUT clock signal 68a and/or the clock-or-strobe signal S relative to the master clock signal to generate a DUT clock signal and/or a strobe signal S.
In some embodiments, the frequency of the DUT clock signal 68a or the frequency of the clock-or-strobe signal S may be different from the frequency of the master signal.
Taking reference to
The tester may furthermore optionally comprise a master clock 72 and a timing circuit 74.
The timing circuit 74 may, for example, be configured to provide one or more timing reference signals 74a to the processing unit 20, for example, to adjust the timing reference for sampling the signals S and D1. Moreover, the timing circuit 74 may optionally provide a timing signal 74b to the comparator unit 66.
In another embodiment of the present invention, the tester may comprise at least one phase-locked loop (PLL) or delay-locked loop (DLL) in order to generate the test clock signal or a clock signal for the data processing unit.
The tester 60 may furthermore (optionally) comprise a data generator 76 coupled to the DUT interface 70 and to the comparator unit 66. The data generator unit may be configured to generate data signal 76a (for example, as a stimulus for a device-under-test) and to provide the data signals to the DUT interface. Also, the data generator may be configured to generate information 76b regarding the expected data. In the embodiment shown in
In general, a tester (for example, the tester 80) may comprise a data processing unit (20) with a data strobe-or-clock channel (for example, the time information provider 22) and a data channel (for example, the data extractor 24) each of them over-sampled (e.g. using the sampling units 22a, 24a) and configured to feed pipelines (e.g. the bit history units 22b, 24b) with the over-sampled bit history of the respective bitstream. A two-fold or higher over-sampling may be used to provide enough choices for the safe extraction of valid information or data. The transition time information (for example, of a clock or strobe time signal) is extracted from the over-sampled bit history of the data strobe signal and forwarded as a digital word. Data (for example, valid data) is extracted from the over-sampled bit history (for example, of the data signal) at a point in time that corresponds to the forwarded transition time of the data strobe signal. As it is described in the context of
The processing 80 may be performed in such a way that the clock or strobe signal is sampled with a first sampling frequency fs1 which is n times higher than a maximum frequency fmax1 of the clock or strobe signal. The sampling frequency fs1 may be n-times higher than a first bandwidth frequency fband1 of the clock-or-strobe signal if the clock-or-strobe signal is an analog signal. The selection 82 of data from a sequence of data samples of a data signal may include sampling the data signal with a second sampling frequency fs2, which is m-times higher than a maximum frequency or data rate fmax2 of the data signal. The second sampling frequency fs2 may be m-times higher than a maximum bandwidth frequency fmax2 of the data signal (n and m≧2). The sampling of the clock-or-strobe signal and of the data signal may be performed with the same sampling frequency, i.e. fs1=fs2. In other words, the sampling rates for the sampling or over-sampling of the clock-or-strobe signal and of the at least one data signal may be equal.
The method of processing data may comprise, as it is schematically shown in the flow chart of
In some embodiments, the method of processing data may comprise sampling a clock or strobe signal and temporarily storing a sampled clock-or-strobe time information based on the clock-or-strobe signal, for example, in a register. Furthermore, the method comprises sampling at least one data signal, temporarily storing data samples from the at least one sampled data signal, and selecting data samples temporarily stored, based on (or independence on) the clock-or-strobe time information.
In a further embodiment, the method may comprise distributing the digitized clock-or-strobe time information to at least one data select unit in order to select a data sample from the data signal.
Using the method of processing data, a propagation delay compensation between one or more (different) data signals and the strobe-or-clock signal can be achieved by sampling the respective signals and forwarding or distributing a strobe point-inactive, coded as a digital word.
The method of processing data may be performed in such a way that a clock-or-strobe time or a data signal is sampled depending on the clock-or-strobe signal. This means that, the clock or strobe signal determines the sampling time.
In another embodiment, the sampling of the clock or strobe signal and the data signals is performed in relation to a time reference provided by a master clock.
The invention can alternatively or in addition be partly or entirely embodied or supported by one or more suitable computer programs which can be stored on or otherwise provided by any kind of data carrier, and which might be executed in or by an suitable data processing unit, for example a computer. The methods as described above of processing data can be performed when a computer program runs on a computer.
Embodiments of the present invention reveal that, instead of strobing data signals directly, a strobe time is digitized, transmitted and used to select data from a choice of digitized data samples. The transition time information can be forwarded in a daisy chain or by other means, for example in a parallel or serial manner. The transport of the digitized strobe time information—the transition time may need multiple pipeline clocks (instead of the vertical buffers).
Embodiments according to the invention comprise the following method: Digitize Strobe time; transmit strobe time; use digitized strobe time to select data from a choice of data samples.
In other embodiments of the present invention, a data strobe channel and data are each over-sampled and feed pipelines with over-sampled bit history. An over-sampling which is, for example, at least two times higher than the bit rate of a bitstream representing the clock or strobe signal or the data signal (“2× or higher over-sampling”) provides enough choices for safe extraction of valid information. A transition time information (for example the digitized clock or strobe time information) is extracted from the over-sampled bit history of the data strobe signal and forwarded as a digital word. Data is extracted from their over-sampled bit histories at a point in time that corresponds to the forwarded transition time of the data strobe signal, optionally corrected for individual signal skew. The point in time that corresponds to the forwarded transition time of the data strobe signal can be optionally corrected individually for each data signal.
While the foregoing has been particularly shown and described with reference to particular embodiments thereof, it will be understood by those skilled in the art that various other changes in the form and details may be made without departing from the spirit and scope thereof. It is to be understood that various changes may be made in adapting to different embodiments without departing from the broader concept disclosed herein and comprehended by the claims that follow.
While this invention has been described in terms of several embodiments, there are alterations, permutations, and equivalents which fall within the scope of this invention. It should also be noted that there are many alternative ways of implementing the methods and compositions of the present invention. It is therefore intended that the following appended claims be interpreted as including all such alterations, permutations, and equivalents as fall within the true spirit and scope of the present invention.
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/EP2008/007912 | 9/19/2008 | WO | 00 | 4/26/2011 |
| Publishing Document | Publishing Date | Country | Kind |
|---|---|---|---|
| WO2010/031417 | 3/25/2010 | WO | A |
| Number | Name | Date | Kind |
|---|---|---|---|
| 5247161 | Actis et al. | Sep 1993 | A |
| 7113560 | Huang et al. | Sep 2006 | B1 |
| 20010018751 | Gresham | Aug 2001 | A1 |
| 20030061564 | Maddux | Mar 2003 | A1 |
| 20040022196 | Rivoir | Feb 2004 | A1 |
| 20050265440 | Sohn | Dec 2005 | A1 |
| 20080034265 | Kang | Feb 2008 | A1 |
| 20080068355 | Chen et al. | Mar 2008 | A1 |
| Number | Date | Country |
|---|---|---|
| 2010031417 | Mar 2010 | WO |
| Entry |
|---|
| Kazamaki, T., “Milestones of New-Generation ATE”, IEEE Design & Test of Computers, 2(5), pp. 83-89, Oct. 1985. |
| Rivoir, J., “Fully-Digital Time-To-Digital Converter for ATE with Autonomous Calibration”, IEEE International Test Conference, 2006, paper 6.3. |
| Number | Date | Country | |
|---|---|---|---|
| 20110197086 A1 | Aug 2011 | US |
