Circuit Configuration Employing a Compare Unit for Testing Variably Controlled Delay Units

1. Field of the Invention

This invention relates to the testing of electrical delay circuits which have controllable delay characteristics. The invention also relates to clock generator circuits employed within, for example, microprocessing units.

2. Description of the Relevant Art

An electrical delay element is typically associated with an output signal which is an image of the input signal delayed by a certain amount of time. Most delay elements have either a fixed delay time or a delay time that varies in a specified way in accordance with a control signal. Testing such delay elements for manufacturing defects can often be simply a matter of comparing their delay with a known reference delay.

A plurality of identical variable delay elements that form a delay chain may be employed within a clock signal generator to synthesize, for example, an internal microprocessor clock signal. Although it is relatively simple to fabricate virtually identical delay elements on a single integrated circuit chip, batch-to-batch variations due to processing variations make it very difficult to fabricate a delay element in which a certain control input value will always yield the same fixed time delay. Fortunately, with respect to the above-mentioned clock signal generators, it is only necessary that the relative delays of the variable delay elements within a particular clock generator be essentially identical. The actual delay values are not critical. In view of this, it may not be possible or may not be cost-effective to test this class of delay element by simple comparison with a known reference delay.

A further consideration with respect to the test of delay elements is the type of possible defects. In an integrated circuit manufacturing test, it is assumed that the fundamental design of the delay element is sound and that only failures introduced by manufacturing defects remain. A manufacturing defect will not necessarily cause the delay element to fail entirely but may rather subtly or dramatically affect the relationship between the control input and the time delay of the delay element. Complete failure of the delay element is easily detected, but subtle manufacturing defects in delay elements are very difficult to detect.

Accordingly, a test configuration and method are desirable wherein variably controlled delay elements embodied within, for example, a clock generator circuit may be readily tested for subtle defects without the need for costly calibration of individual delay elements.

The problems outlined above are in large part solved by a test configuration and method for testing a plurality of variably controlled delay units according to the present invention. In one embodiment, a test configuration is provided which allows a plurality of variable delay units within a delay chain to be compared with respect to one another. The delay chain is employed within a clock generator circuit that generates internal clock signals of a microprocessor. During normal operation, a set of multiplexers interposed within the delay chain are configured such that the plurality of variable delay units are electrically coupled in series with respect to one another. As a result, a pulse signal may be provided to an input line of the delay chain such that it propagates consecutively through each variable delay unit to an output line of the delay chain. Since each variable delay unit is associated with an identical delay, a tap point within each variable delay unit may be utilized to generate a clock signal having a precisely controlled duty cycle.

During a test operation when it is desired to test the variable delay units for possible defects, the four delay units are electrically separated from one another by setting the multiplexers in a test mode. A common test signal is then driven through two or more of the variable delay units simultaneously, and a compare circuit coupled to the output of each variable delay unit determines whether a transition in the common pulse signal propagated through each variable delay unit at essentially the same time. If no manufacturing defects are present, the four outputs of the variable delay units should be virtually indistinguishable from one another. Similar tests may be conducted throughout the entire operating range of the variable delay units. The test configuration and method advantageously allow the variable delay units within a delay chain of a clock generator circuit to be tested for subtle defects without the need for costly calibration of individual delay elements and without the need for testing the actual delay of each delay element.

Broadly speaking, the present invention contemplates a variably controllable delay circuit comprising a plurality of variable delay units each including an input line, an output line, and a control line for controlling an electrical delay associated therewith, and a switching unit interposed between the plurality of variable delay units. The switching unit is capable of electrically coupling the plurality of variable delay units in series during a first mode of operation, and is further capable of electrically decoupling the plurality of variable delay units such that a test signal can be simultaneously provided to the input line of each of the plurality of variable delay units. The delay circuit further comprises a compare unit coupled to the output lines of at least a pair of the variable delay units, wherein the compare unit is capable of determining whether a transition of the test signal provided simultaneously to the input lines of the pair of variable delay units has propagated to the output lines of the pair of variable delay units within a relative time range.

The present invention further contemplates a method for testing a delay chain including a plurality of variable delay units capable of being electrically coupled in series with respect to one another. The method comprises the steps of electrically decoupling an output line of a first variable delay unit from an input line of a second variable delay unit, providing a test signal simultaneously to the first variable delay unit and to the second variable delay unit, and causing a signal transition to occur in the test signal. The method comprises the further steps of monitoring an output line of the first variable delay unit while monitoring an output line of the second variable delay unit, and determining whether a corresponding transition at the output line of the first variable delay unit occurs within a relative time range of when a corresponding transition at the output line of the second variable delay unit occurs.

Other objects and advantages of the invention will become apparent upon reading the following detailed description and upon reference to the accompanying drawings in which:

FIG. 1 is a block diagram of a clock generator circuit that employs a delay chain including a plurality of variable delay units.

FIG. 2 is a block diagram of a portion of the clock generator circuit wherein the delay chain is modified to include a set of multiplexers interposed within the delay chain for allowing a test of the variable delay units in accordance with the present invention.

FIG. 3 is a block diagram of a test circuit that allows the comparison of selected variable delay units.

FIG. 4 is a schematic diagram that illustrates one implementation of a compare unit employed within the test circuit of FIG. 3.

FIG. 5 is a block diagram that illustrates another implementation of a compare unit employed within the test circuit of FIG. 3.

FIG. 6 is a schematic diagram that illustrates an exemplary delay chain employed within the compare unit of FIG. 5.

While the invention is susceptible to various modifications and alternative forms, specific embodiments thereby are shown by way of example in the drawings and will herein be described in detail. It should be understood, however, that the drawings and detailed description thereto are not intended to limit the invention to the particular form disclosed, but on the contrary, the intention is to cover all modifications, equivalents and alternatives falling within the spirit and scope of the present invention as defined by the appended claims.

Referring now to the drawings, FIG. 1 is a block diagram of a clock waveform generator circuit 100 for generating a clock signal within, for example, a microprocessor. The waveform generator 100 comprises a pulse generator 102 coupled to a delay chain 104 including a plurality of variable delay units 106A-106D. Delay chain 104 is further coupled to a clock synthesis unit 108 and a delay control logic unit 110. The clock waveform generator of FIG. 1 receives an externally generated clock signal labeled INCLK at an input line of pulse generator 102, and generates a clock output signal at an output line 111 of clock synthesis unit 108.

It is noted that while the externally generated clock signal INCLK is associated with stable frequency characteristics, variations in its duty cycle may occur. The clock waveform generator circuit illustrated in FIG. 1 is provided to generate a clock output signal at line 111 having both stable frequency characteristics and a stable duty cycle. For the embodiment of FIG. 1, the clock output signal at line 111 may be either the same frequency or twice the frequency of signal INCLK, though any integer (or half integer) multiple of the frequency of signal INCLK is possible if the number of variable delay units 106A-106D incorporated within the clock waveform generator is varied.

The clock waveform generator of FIG. 1 conforms generally to that described within the co-pending, commonly assigned patent application entitled "Digital Clock Waveform Generator and Method for Generating a Clock Signal" by Horne, et al., Ser. No. 011,068, filed Jan. 29, 1993. This patent application is incorporated herein by reference in its entirety. In the interest of simplicity and clarity, only those portions of the digital clock waveform generator of FIG. 1 as they relate to the present invention are discussed herein. Additional details regarding an exemplary digital clock waveform generator may be found within the above-referenced patent application.

Generally speaking, the variable delay units 106A-106D are configured identically. During operation, pulse generator 102 generates a pulse at a fixed frequency which passes sequentially through each variable delay unit 106A-106D and into delay control logic unit 110 via line 114. Delay control logic unit 110 includes circuitry that monitors the time at which a particular pulse exits the variable delay unit 106D relative to the time at which a subsequent pulse generated by pulse generator 102 enters variable delay unit 106A. Delay control logic unit 110 responsively adjusts the delays of the variable delay units 106A-106D such that the overall delay of delay chain 104 equals one period of the pulse signal generated by pulse generator 102.

A set of timing reference signals labeled TREF1-TREF4 are derived at corresponding locations within the variable delay units 106A-106D. As a signal transition (induced by pulse generator 102) propagates down the delay chain 104 from the input of variable delay unit 106A to the output of variable delay unit 106D, corresponding signal transitions sequentially appear within the timing reference signals TREF1-TREF4. Since each variable delay unit 106A-106D has the same delay with respect to the others, and since the timing reference signals TREF1-TREF4 are derived at corresponding tap points within each variable delay unit 106A-106D, clock synthesis unit 108 can utilize the timing reference signals TREF1-TREF4 to generate a clock output signal having a precisely controlled duty cycle. That is, since the four timing reference signals TREF1-TREF4 represent four equally spaced taps from the delay chain 104, the timing reference signals may be used by clock synthesis unit 108 to trigger transitions in the clock output signal. The timing reference signals TREF1-TREF4 are equally spaced in the sense that signal TREF2 produces a rising edge one-fourth of an INCLK period after signal TREF1 produces a rising edge. Similarly, signal TREF3 produces a rising edge one-fourth of an INCLK period after signal TREF2, and signal TREF4 produces a rising edge one-fourth of an INCLK period after signal TREF3 produces a rising edge. As a result, the clock synthesis unit 108 is capable of generating a clock output signal having a frequency twice that of the INCLK signal with a duty cycle of 50 percent.

As stated previously, it is necessary that the relative delays of the variable delay units 106A-106D be essentially identical. Thus, there is a need to test the variable delay units 106A-106D to ensure that, for example, no manufacturing defects are present. It is noted that if one of the variable delay units 106A-106D has a manufacturing defect which affects its relative delay, the duty cycle of the generated clock output signal at line 111 will not be associated with a desired duty cycle of 50 percent.

Therefore, referring next to FIG. 2, a block diagram is illustrated of a portion 200 of the clock generator circuit of FIG. 1 modified to include a test configuration according to the present invention. Circuit portions that correspond to those of FIG. 1 are numbered identically. It is noted that selected circuit blocks incorporated within the clock generator of FIG. 1 have been omitted from FIG. 2 for simplicity and clarity.

As illustrated in FIG. 2, a set of multiplexers 202A-202D are interposed within the delay chain 104 such that the variable delay units 106A-106D can be electrically decoupled from one another during a test mode. In this embodiment, multiplexer 202A is configured to control the selection of an input to variable delay unit 106A, multiplexer 202B is configured to control an input to variable delay unit 106B, multiplexer 202C is configured to control an input to variable delay unit 106C, and multiplexer 202D is configured to control an input to variable delay unit 106D.

During normal operation of the clock generator circuit, multiplexers 202A-202D are selected such that the pulse input signal at line 210 (from the pulse generator 102 of FIG. 1) is electrically coupled through the delay chain 104 from an input 212 of variable delay unit 106A, through each variable delay unit 106A-106D, and to the output 214 of variable delay unit 106D. It is noted that during this mode of operation, multiplexers 202A-202D are selected such that variable delay units 106A-106D are electrically coupled in series with respect to one another. This allows the generation of the timing reference signals, as described previously.

During testing, the multiplexers 202A-202D are selected such that the variable delay units 106A-106D are no longer electrically coupled in series with respect to one another. Multiplexer 202B electrically decouples the output of variable delay unit 106A from the input of variable delay unit 106B, multiplexer 202C electrically decouples the output of variable delay unit 106B from the input of variable delay unit 106C, and multiplexer 202D electrically decouples the output of variable delay unit 106C from the input of variable delay unit 106D. Multiplexer 202A similarly decouples the input of variable delay unit 106A from the pulse generator 102 (of FIG. 1). Instead, during this mode of operation, a test input signal at line 204 is allowed to be simultaneously coupled through each multiplexer 202A-202D and through each variable delay unit 106A-106D independently. An output signal from each variable delay unit 106A-106D may thus be monitored at a respective tap point (test tap#1-test tap#4) derived from the output line of each variable delay unit 106A-106D, as will be described in greater detail below.

When multiplexers 202A-202D are set in the test mode, the test input signal at line 204 is driven with a step transition such that a rising signal edge propagates simultaneously through each variable delay unit 106A-106D. If no manufacturing defects are present, the times at which the rising edge transitions appear at the four test tap outputs (test tap#1-test tap#4) should be virtually indistinguishable.

FIG. 3 is a block diagram of a test circuit that allows the four test tap outputs (test tap#1-test tap#4) of FIG. 2 to be compared against one another. The four test tap outputs are provided to a multiplexer 302 which channels a selected two of the test tap outputs to a compare unit 304. It is noted that the incorporation of multiplexer 302 allows the use of only one comparator circuit 304.

During operation, multiplexer 302 is selected such that test tap#1 and test tap#2 are first channeled through the multiplexer to the outputs of multiplexer 302 labelled Compare(l) and Compare(2), respectively. Subsequently, when a rising edge transition occurs in the test input signal at line 204 (FIG. 2), the transition propogates through variable delay units 106A and 106B and through multiplexer 302 to compare unit 304. The time at which a corresponding transition occurs at the output of each variable delay unit 106A and 106B may thereby be compared to the other by compare unit 304. If the transitions do not occur at essentially the same time, compare unit 304 indicates a bad result. On the other hand, if the signal transitions occur at essentially the same time, the compare unit 304 indicates a good result.

Multiplexer 302 is subsequently selected such that test tap#2 is compared against test tap#3. Another rising edge transition to the test input signal at line 204 is again induced, and the transition propagates through variable delay units 106B and 106C. Compare unit 304 then determines whether corresponding transitions occurred at the outputs of the variable delay units 106B and 106C at essentially the same time. A similar test is thereafter applied with respect to test tap#3 and test tap#4, and possibly with respect to test tap#4 and test tap#1.

By employing the test described above, it may be determined whether the variable delay units 106A-106D each have the same relative delay. If one or more of the variable delay units 106A-106D has a different delay characteristic with respect to the others, thus suggesting that a manufacturing defect may be present, the deviation in delay will be detected by compare unit 304 during the test. It is noted that multiplexer 302 may be configured and controlled such that different combinations of pairs of test tap outputs (from those chosen above) may be selectively compared while achieving the same result. It is further noted that the variable delay units 106A-106D may be tested throughout their entire operating range by adjusting their control inputs and repeatedly testing the relative delays of the various units.

FIG. 4 illustrates one implementation of a compare unit 304. The compare unit of FIG. 4 includes an exclusive OR gate 402 and an SR-latch 404. Upon preparation for the test of each selected pair of variable delay units 106A-106D, the reset signal is asserted, thus setting the output of latch 404 low. Subsequently, when the rising edge transitions at the Compare(1) and Compare(2) lines occur, exclusive OR gate 402 transitions high only if the signals at the two input lines are driven high at different times. That is, if the input lines of exclusive OR gate 402 transition from low to high at essentially the same time, the output line will not go high. On the other hand, if the input signals to exclusive OR gate 402 make transitions at significantly different times, the output of exclusive OR gate 402 will go high, thus setting latch 404. It is noted that a small time deviation with respect to the transitions of the two input signals may not be detected, as determined by the specific electrical characteristics (i.e., response time) of exclusive OR gate 402. The setting of latch 404 indicates a bad result.

FIG. 5 is a block diagram that illustrates another implementation of a compare unit 304 which may be employed to compare the relative delays of variable delay units 106A-106D. The compare unit of FIG. 5 not only allows for the determination of whether respective rising edge transitions from the variable delay units 106A-106D occur at essentially the same time, but further allows a relative measure of how different a particular delay unit is from another. This may allow the user to vary the strictness of the manufacturing test without changing the compare structure itself.

As illustrated in FIG. 5, the compare unit includes a delay chain 502, a delay chain 504, and a register 506. An input of delay chain 502 is connected to the Compare(l) line (from multiplexer 302 of FIG. 3), and delay chain 504 is connected to the Compare(2) signal. An output of delay chain 502 is coupled to a latching input of register 506. As illustrated within FIG. 6, a set of tap points labeled Taps (1-8) are provided from delay chain 504 to an input of register 506. The output of each tap is a digital signal that indicates whether a rising edge transition of an input signal (at line Compare(2)) has reached that point in the chain or not. It is noted that the system is reset to all zero outputs before each test and that each zero becomes a one as the input signal progresses through the chain.

Delay chain 504 is twice as long as the delay chain 502 (that is, the overall delay of delay chain 504 is twice that of delay chain 502). The output of delay chain 502 is used as a control signal to cause register 506 to latch the current values of taps (1-8) upon its rising edge. Accordingly, when a rising edge of the signal at Compare(1) line propagates through delay chain 502, register 506 latches a value that indicates how far a rising edge of the signal at the compare(2) line has progressed along the longer delay chain 504. Since the delay chains 502 and 504 have identical structures, it is expected that the register 506 will indicate that the rising edge of the second input signal progressed approximately half way down the longer delay chain 504 at the point at which the transition in the first input signal exited the delay chain 502. Any variations in the transitions of the signal at the Compare(1) and Compare(2) lines will result in a greater or lesser number of high values latched into register 506. The results of each test may be determined by reading the value within register 506.

It is noted that the compare circuit of FIG. 5 may be calibrated by providing a common input signal to the Compare(1) and Compare(2) lines and by subsequently reading the resulting value of register 506. With this information, it is known what value will be stored within register 506 when identical input signals are provided to delay chains 502 and 504.

It is further noted that the test configuration method described above may be advantageously employed within clock generator circuits that include delay chains having any number of variable delay units. Furthermore, although the variable delay units 106A-106D as described in FIG. 1 are digitally controlled delay units, the present invention may also be adapted to delay chains that employ analog controlled delay units. It is finally noted that the test configuration and method described above may be employed to compare variable delay units utilized within applications other than the application described above with respect to microprocessor clock generators.

A microprocessor that incorporates a test configuration and method as described above is disclosed within the co-pending, commonly assigned patent application entitled "Microprocessor Employing A Configuration and Method For Testing A Delay Chain Within A Clock Generator", by McMinn et al., Ser. No. 08/212,037, filed concurrently herewith, and issued Jul. 4, 1995 as U.S. Pat. No. 5,430,394. This patent application is incorporated herein by reference in its entirety.

Numerous variations and modifications will become apparent to those skilled in the art once the above disclosure is fully appreciated. For example, while the embodiment described above in conjunction with FIG. 2 employs multiplexers 202A-202D, other types of switching circuitry could be employed to electrically couple the variable delay units in series during normal operation and to electrically decouple the variable delay units during the testing operation. It is intended that the following claims be interpreted to embrace all such variations and modifications.