Data processing: measuring – calibrating – or testing – Testing system – Of circuit
Reexamination Certificate
1999-03-01
2001-11-27
Wachsman, Hal (Department: 2857)
Data processing: measuring, calibrating, or testing
Testing system
Of circuit
C324S537000
Reexamination Certificate
active
06324486
ABSTRACT:
FIELD OF THE INVENTION
The present invention pertains generally to automated testing techniques, and, more particularly, to a method for adaptively learning test error sources in an automated test system in order to reduce the number of measurements that are taken in real-time.
BACKGROUND OF THE INVENTION
Automated test equipment is used to perform a wide variety of tests that might otherwise be performed manually at a slower rate and/or greater cost. Automation of a test typically involves the reading of measurements from a device under test from which calculations are performed to determine whether the device under test is within or outside of acceptable test limits. For example, in the large-scale production of electronic circuits, automated test equipment is used for performing tests on each circuit board of a run of circuit boards. A run is a testing sequence of the same type of assembly with no intervening different types of assemblies. A typical automated circuit tester includes a test configuration circuit, a bed-of-nails fixture, and a set of programmable relay matrices and internal measurement busses. A circuit under test is positioned on the bed-of-nails fixture such that the nails electrically connect to nodes of the circuit under test. The measurement paths from the circuit under test to the test configuration circuit are connected by programming the appropriate relays in that path to close. While the relays in the measurement path are closing, the test configuration circuit is set up and prepared to take actual measurements. After the relays have closed, measurements are read from the circuit under test. Calculations are performed based on the measurements to determine whether the circuit under test passed the test.
In the prior art, automated testers obtain each required measurement in real-time-that is, each measurement that is required in the calculations that determine whether the test passed or failed are physically acquired on each test of a device under test that is part of the same test run. In many applications, however, several of the required measurements do not change significantly over time. This may be due to system-level error sources that generally remain constant for a particular test run. The taking of these stable error source measurements results in unnecessary test time overhead. Accordingly, a need exists for a method for reducing the number of real-time test measurements over a run of devices under test.
SUMMARY OF THE INVENTION
The improvements of the present invention over prior art automated testing techniques is achieved in several ways. First, the measurements required to compensate for error sources that do not change significantly between test runs or over time are measured and stored as historical measurements for use in calculating the component under test values in subsequent test runs. The measurements required to compensate for the error sources described above that do significantly change between test runs or over time are measured in real-time on a per test basis.
Secondly, since error source compensations differ on a per test basis, the measurements required to compensate for those error sources also differ on a per test basis. Therefore, the historical measurements that are measured and used for error compensation and test calculations are stored with each individual test.
Third, the stored historical measurements used for error compensation and test calculations are updated (i.e., re-measured and re-stored) on a periodic basis, first as they begin to affect the component under test value calculations and secondly as the physical number of times that those same stored historical measurements used in the component under test value calculations begin to increase over time.
Accordingly, the technique of the invention effectively adaptively learns the measurements and thus the error sources that do not change with time. By adaptively learning which measurements used in error source compensation do not change with time, and substituting historical values for those measurements, the number of measurements required to be taken in real-time is significantly reduced, and thus the speed of the test (i.e., its execution time) is significantly improved.
In accordance with the invention, the set of real-time measurements taken in real-time is reduced using an adaptive learning technique. First, a current set of real-time measurements is selected. The current set of real-time measurements is a subset of a full set of required measurements that are required to determine whether the test passes or fails. Once the current set of real-time measurements are selected, these measurements are measured from the circuit under test in real-time. Values of a current set of historical measurements comprising historical measurement values for each of the required measurements that are not included in the current set of real-time measurements are obtained. A determination is made, based on the current set of measured real-time measurements and the current set of historical measurements substituted for those required measurements that are not included in the current set of real-time measurements, as to whether said test passed or failed. If the test fails, a current retry set of real-time measurements comprising a subset of the required measurements different than the current set of real-time measurements is selected. The current retry set of real-time measurements is then measured in real-time. A current retry set of historical measurements comprising corresponding historical measurement values for each of the required measurements that are not included in the current retry set of real-time measurements is obtained, and a determination is made, based on the current retry set of measured real-time measurements and the current retry set of historical measurements substituted for those required measurements that are not included in the current retry set of real-time measurements, as to whether said test passed or failed.
In one embodiment, if the test fails using the current retry set of real-time measurements and current retry set of historical measurements, a next retry set of real-time measurements is selected and the test is reexecuted using these sets. This process may be repeated until either the test passes or the current retry set of real-time measurements comprises the full set of required measurements. If the test passes during a reexecution of the test, the corresponding historical measurement values of each of the current retry set of real-time measurements may be updated with the corresponding measured real-time measurements. In addition, if the test passes during a reexecution of the test, the current set of real-time measurements may be set to the current retry set of real-time measurements for use in testing the next circuit under test. An optional feature that may be performed if the test passes during reexecution of the test is reevaluating the current set of real-time measurements by determining a new set of real-time measurements and using the new set of real-time measurements as the current set of real-time measurements. Another optional feature that may be performed if the test passes during reexecution of the test is an audit function, in which the set of historical measurements is updated with current measurement values.
REFERENCES:
patent: 5202639 (1993-04-01), McKeon et al.
Journal of Electronic Testing: Theory and Applications 9.9-18(1996), XP 000636621, “Selecting Measurements to Test the Functional Behavior of Analog Circuits”, J. Van Spaandonk and T.A.M. Kevenaar, pp. 9-18.
Crook David T.
List Steven K.
Rozum Stephen P.
Williamson Eddie L.
Agilent Technologie,s Inc.
Pretlow Demetrius
Wachsman Hal
LandOfFree
Method and apparatus for adaptively learning test error... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for adaptively learning test error..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for adaptively learning test error... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2575222