Calibration of Keysight Electronic Calibration Kits

Technical PapersSummaryThis article outlines and explains the process of calibrating or verifying a Keysight E-Cal module and the associated calibration certificate that is given after such a cal is performed at a Keysight service centreDescriptionOverview of the Re-certification Process Annual re-certification is recommended to insure that an E-cal module delivers accurate calibrations.  The solid state nature of an E-cal makes annual calibration (re-certification) especially important. Because of the complex, precision nature of the tasks involved, re-certification of an E-cal module requires highly specialized equipment and hours of processing time. The E-cal test systems are precision test beds specifically designed for the characterization and testing of E-cal modules. Precision calibration of an E-cal test system: Before an E-cal test system is used to re-certify an E-cal module, a precision calibration specific to the E-cal connector type and frequency range must be performed.A precision calibration is performed on the test system PNA using a characterized devices (CD) calibration kit.  The CD calibration kits are themselves directly characterized using primary standards.The measurement accuracy provided by the E-cal test system is equivalent to the level provided by a national metrology laboratory.  The test systems are maintained in a carefully controlled temperature environment (+/-1.0 C).Essentially a double calibration is done to insure that the test system is properly calibrated.  The results of two different calibrations with separate calibration standards are compared at the E-term level.  Precision consistency between the two calibrations insures an accurate calibration of the test system. The process of E-cal re-certification involves: Mechanical inspection: a precision mechanical check of the connectors is performed on the E-cal module.  Defective or out-of-tolerance connectors are adjusted or replaced.As Received test:  a calibration is performed on the test system PNA using the E-cal.  The resulting E-terms are compared to the E-terms from the precision calibration previously performed on the test system.  An As-Received Tolerance Report is generated which lists the differences in E-terms.Characterization update:  the E-cal module is then characterized and the state definition information is stored in the E-cal.  The memory in the E-cal is updated to reflect its current state.Final check:  the updated E-cal is used to calibrate the test system PNA once more.  The resulting E-terms are compared to the E-terms from the precision test system calibration.  The E-term differences must be very small (typically less than 3 milli-units) in order for this final check to “pass.” Reports provided: As Received & Post Characterisation Tolerance Reports: lists the differences between a calibration performed using the E-cal module and a precision test system calibration.  Worst case differences are reported by E-term and frequency band.  Pass/fail limits are provided for each frequency band.  There are two versions of this report, As Received and Post Characterization (final check). If an E-cal module is repaired or if it has drifted beyond its specifications since it’s last characterization (as indicated by the As Received test), the E-cal module will be temperature cycled between -40 and +70 C for ten hours to enhance stability.  In general, about 80% of E-cal modules require temperature cycling during the re-certification/repair process. Because the states in a given E-cal module do not have expected values (unlike mechanical calibration standards), a “bad” connection during characterization is not easily recognized.  Therefore the test process must insure the connections to the E-cal module are very good.  The re-certification process includes precision “connection checking” to insure the E-cal characterizations are accurate. The “specifications” for a calibration kit or E-cal indicate the accuracy of a calibration that may be performed using the kit or E-cal.  These “specifications” are not the same as the accuracy requirements for measuring the individual standards or states.  But the “specifications” do, of course, influence these accuracy requirements. The accuracy of a calibration using an E-cal depends on the accuracy of the state values stored in the E-cal memory.  In order for an E-cal to meet its specifications, the stored state values must be within a small tolerance of the actual values Details on the tolerance reports produced for E-cal re-certifications. There are four Tolerance Reports available for an E-cal re-certification: As Received Tolerance Report: reports worst case points, a summary reportAs Received Detailed Tolerance Report: reports worst case points within each frequency bandPost Characterization Tolerance Report: reports worst case points, a summary reportPost Characterization Detailed Tolerance Report: reports worst cast points within frequency bands (this report is only provided to customers upon special request) High level process for As Received testing of an E-cal: A PNA system is calibrated using a CD (characterized devices) cal kit.  This calibration produces a set of error terms (ET_cd) in the PNA. The same PNA system is calibrated using the E-cal under test in an as-received condition.  This calibration produces a second set of error terms (ET_ecal_ar) in the PNA. An “Uncertainty” for the E-cal under test is calculated as a function of the differences between the two sets of error terms (ET_cd and ET_ecal_ar). These differences are considered the actual residuals for the E-cal in the as-received condition.  The “Uncertainty” is calculated as a function of the directivity, source match, and reflection tracking residuals.  It is a measure of the difference between the CD cal kit and the E-cal under test.  The “Uncertainty” is calculated for combinations of frequency and reflection coefficient. An “Uncertainty Limit” is calculated for a given frequency and reflection coefficient, where the Uncertainty Limit is a function of the Directivity, Source Match, and Reflection Tracking.  A guard band on the Uncertainty Limit is calculated.  It is based on the residuals of the CD cal kit with two sigma coverage.  This guard band is the uncertainty contribution from the CD cal kit.  It may be considered the uncertainty on the “Uncertainty” value mentioned above.  High level process for Post Characterization testing of an E-cal: Note: After an E-cal is re-characterized (data for state definitions has been updated), a post characterization test is performed.  A PNA system is calibrated using a CD (characterized devices) cal kit.  This calibration produces a set of error terms (ET_cd) in the PNA. The same PNA system is calibrated using the E-cal under test.  This calibration produces a second set of error terms (ET_ecal_pc) in the PNA. An “Uncertainty” for the E-cal under test is calculated as a function of the differences between the two sets of error terms (ET_cd and ET_ecal_pc). These differences are considered the actual residuals for the E-cal.  The “Uncertainty” is calculated as a function of the directivity, source match, and reflection tracking residuals.  It is a measure of the difference between the CD cal kit and the E-cal under test.  The “Uncertainty” is calculated for combinations of frequency and reflection coefficient. A post characterization “Uncertainty Limit” is calculated for a given frequency and reflection coefficient based on the “performance characteristics” of the E-cal model under test.  The Uncertainty Limit is a function of the Directivity, Source Match, and Reflection Tracking values given in the E-cal Reference Guide (N4693-90001 ).  (These are the “residuals” for the E-cal model based on the published performance characteristics.)  A guard-banded uncertainty limit is calculated as 25% of the Uncertainty Limit.  This tighter limit is used as the “Uncertainty Limit” in the Post Characterization Tolerance Report (the summary report).  This tighter limit is used to insure there is margin in the E-cal performance to allow for typical drift and other changes in the E-cal during twelve months of normal usage. All results must be PASS for an acceptable Post Characterization test. Results for Tolerance reports (for a given frequency and reflection coefficient): PASS – calibration using the E-cal in the as-received condition will produce a measurement uncertainty that is less than the measurement uncertainty magnitude derived from the performance characteristics for the E-cal model.PASS# (conditional pass) - calibration using the E-cal in the as-received condition will likely produce a measurement uncertainty that is less than or equal to the measurement uncertainty magnitude derived from the performance characteristics for the E-cal model.  (i.e. the Uncertainty value is between the Guard-banded Limit and the Uncertainty Limit)FAIL - calibration using the E-cal in the as-received condition will likely produce a measurement uncertainty that is greater than the measurement uncertainty magnitude derived from the performance characteristics for the E-cal model. Notes and explanations: The values for Limits and Uncertainties are linear magnitudes referenced to the reflection coefficient.  For example, if Uncertainty = 5.500m for Reflection Coefficient = 0.100, then for a measured reflection coefficient value of 0.100 the actual value is in the range of 0.100 +/- 0.0055. Residuals define the measurement uncertainty contribution from a cal kit or E-cal.  As used above, “uncertainty” represents the uncertainty expected for a reflection coefficient measurement for a VNA that has been calibrated using a cal kit or E-cal.  It is a function of directivity, source match and reflection tracking factors and the magnitude of the reflection coefficient.  It indicates the overall quality of the calibration result when using a cal kit or E-cal.  “Uncertainty” does not indicate the measurement uncertainty of individual measurements performed on the E-cal under test. Explanations for “As Received Tolerance Report” (summary): Worst Case Uncertainty – worst case “uncertainty” value relative to the Uncertainty Limit for the indicated reflection coefficient.Uncertainty Limit – Uncertainty Limit is a function of the Directivity, Source Match, and Reflection Tracking.. Explanations for “As Received Detailed Tolerance Report”: Uncertainty – worst case “uncertainty” value across the indicated frequency band.  (Multiple frequency points are tested within each band.)Guard-banded Limit – Uncertainty Limit less the uncertainty contribution from the CD cal kit.Uncertainty Limit - limit based on performance characteristics of E-cal model (as given in E-cal Reference Guide).  This limit is constant within a frequency band for a given reflection coefficient. Explanations for “Post Characterization Tolerance Report” (summary): Worst Case Uncertainty – worst case “uncertainty” value relative to the Uncertainty Limit for the indicated reflection coefficient.Uncertainty Limit – 25% of the uncertainty based on the performance characteristics defined for the E-----cal model in the E-cal Reference Guide (N4693-90001). Example test report from an “As Received Detailed Tolerance Report”: The picture below shows part of the calibration report where the columns can be interpreted as follows: Reflection coefficient: This is the nominal value for the parameterUncertainty : This is the “measured value” of the calibration, i.e. a measure of the uncertainty when using this E-cal kit to measure the parameter; reflection coefficientUncertainty Limit: This is the “specification” for the “measured value”.Guard-banded Limit : This is the “specification limit” that the “measured value” is compared against to make the PASS/FAIL statementUncertainty Limit minus Guard-banded Limit : The difference of the two columns is the “measurement uncertainty” in the verification of the E=cal kit.