Standard versus Custom calibration of a Fluke 5720A

Transcription

1 Standard versus Custom calibration of a Fluke 5720A How many calibration points suffice to be sure your instrument is completely within specifications? Preface In this document I would like to share an experience I had with the calibration of our Fluke 5720A Multifunction Calibrator. It will reveal a case in which the so called Standard calibration showed the unit to be within specifications, while a Custom calibration showed the unit to be out of specifications. Disclaimer It is not my intention to criticize either the calibration provider, their calibration methods and policies, or the performance of the Fluke 5720A Multifunction Calibrator. Introduction A Multifunction Calibrator like the Fluke 5720A is meant for calibrating measuring devices. To be sure that the values generated by the Multifunction Calibrator are traceable and accurate within their published specifications, the Multifunction Calibrator itself has to be calibrated too. During this calibration, the generated output signals of the Multifunction Calibrator are being measured and eventually the measured values compared to the published specifications. In this way it is possible to determine if the Multifunction Calibrator is within specs or out of specs. Remember however, that this statement only holds true for the points that have actually been measured! Hence, the more points a calibration provider measures, the more coverage you have, up to the point that you could state that the specific unit is completely (= at all values in all functions and ranges) within specifications. Since it is impossible to measure an infinitely number of points, there s always the dilemma of how many, and which calibration points suffice to be sure that the Multifunction Calibrator is completely within specifications. Of course there s a tradeoff between cost (effort) and confidence in this case. Here is where the so called Standard calibration comes in sight. A Standard calibration tries to cover all generally used functions and ranges of an instrument, while keeping costs down by reducing the number of calibration points as much as possible. One of the methods to accomplish this is to measure only one value halfway each range and function (the mid-range point). While this definitely reduces costs, it raises the question: Does it give the user enough confidence that the Multifunction Calibrator can be considered to be completely within specifications? TCEM 1208 Standard versus Custom calibration of a Fluke 5720A Page 1 of 6

2 Custom calibration Things become even more stringent when the Multifunction Calibrator is being used as a reference standard on which the Calibration and Measurement Capability (CMC) of an ISO accredited calibration laboratory is based. In this case the calibration of the Multifunction Calibrator must have enough coverage to calculate the CMC, and comply to the general rule that interpolation is allowed in the calculation of the CMC, while extrapolation is not allowed or only to a modest extent. This implies that you must have at least two widely spread calibration points on each function and range that are part of the CMC. This often disqualifies a Standard calibration for the purpose of deducing the CMC of an ISO accredited calibration laboratory. Some laboratories rely on the assumption that a Multifunction Calibrator is completely within specification when a Standard calibration is performed and all measured points are found to be within specs. But is this a safe assumption? Calibration of the Fluke 5720A We asked a calibration provider to perform a Custom calibration on our Fluke 5720A Multifunction Calibrator and supplied them with a list of calibration points we would like to have measured: two points per range (mostly around 10% and 90% of range) on all ranges and functions we are planning to use. The calibration provider apparently started with a Standard calibration and soon reported that our unit was found to be within specs, without any adjustments being performed. So far, so good. After about a week we got another, different message, telling us that during the Custom calibration it turned out that our unit was out of specs at four measured points and passed conditionally at another four measured points. [NB: Conditional Pass being that the measurement result is within the specification limit by a margin less than the measurement uncertainty: it is therefore not possible to state compliance based on the stated level of confidence. However the results indicate that compliance is more probable than non-compliance with the specification limit.] Our unit was subsequently adjusted and returned to us and we received the 5720A with no less than 4 calibration certificates: Calibration certificate with the results of the Standard calibration before adjustment Calibration certificate with the results of the Custom calibration before adjustment Calibration certificate with the results of the Standard calibration after adjustment Calibration certificate with the results of the Custom calibration after adjustment On the next pages the most interesting results of the Standard calibration and the Custom calibration are shown. TCEM 1208 Standard versus Custom calibration of a Fluke 5720A Page 2 of 6

3 Calibration results of the Standard calibration Before Adjustment, Calibration Date = DC Voltage 1 22 V V V V ± 0.5 ppm 28% Pass 2 22 V V V V ± 0.5 ppm 8% Pass V V V V ± 0.6 ppm 40% Pass V V V V ± 0.6 ppm 38% Pass AC Current pa No points measured on this range 6 22 ma ma 55 Hz ma ma ± 21 ppm 1% Pass 7 22 ma ma 400 Hz ma ma ± 21 ppm 1% Pass 8 22 ma ma 1 khz ma ma ± 21 ppm 1% Pass 9 22 ma ma 5 khz ma ma ± 21 ppm 3% Pass ma ma 10 khz ma ma ± 21 ppm 7% Pass TCEM 1208 Standard versus Custom calibration of a Fluke 5720A Page 3 of 6

4 Calibration results of the Custom calibration Before Adjustment; Calibration Date = DC Voltage V V V V ± 0.5 ppm 28% Pass V V V V ± 1.8 ppm 167% Fail V V V V ± 0.5 ppm 8% Pass V V V V ± 1.8 ppm 174% Fail V V V V ± 1.8 ppm 107% Fail V V V V ± 2.1 ppm 61% Pass V V V V ± 1.8 ppm 98% Pass V V V V ± 2.1 ppm 69% Pass AC Current pa pa 20 Hz pa pa ± 200 ppm 98% Conditional Pass pa pa 40 Hz pa pa ±200 ppm 91% Pass pa pa 1 khz pa pa ± 200 ppm 82% Pass pa pa 5 khz pa pa ± 200 ppm 42% Pass pa pa 10 khz pa pa ± 200 ppm 7% Pass pa pa 20 Hz pa pa ± 150 ppm 98% Conditional Pass pa pa 40 Hz pa pa ± 150 ppm 90% Conditional Pass pa pa 1 khz pa pa ± 150 ppm 18% Pass pa pa 5 khz pa pa ± 220 ppm 2% Pass pa pa 10 khz pa pa ± 600 ppm 2% Pass ma ma 20 Hz ma ma ± 56 ppm 112% Fail ma ma 40 Hz ma ma ± 36 ppm 46% Pass TCEM 1208 Standard versus Custom calibration of a Fluke 5720A Page 4 of 6

5 31 22 ma ma 1 khz ma ma ± 36 ppm 28% Pass ma ma 5 khz ma ma ± 36 ppm 9% Pass ma ma 10 khz ma ma ± 36 ppm 1% Pass ma ma 20 Hz ma ma ± 56 ppm 99% Conditional Pass ma ma 40 Hz ma ma ± 36 ppm 61% Pass ma ma 1 khz ma ma ± 36 ppm 33% Pass ma ma 5 khz ma ma ± 36 ppm 17% Pass ma ma 10 khz ma ma ± 36 ppm 4% Pass Within specifications out of specifications It is interesting to determine why the Multifunction Calibrator was judged to be within specs according to the results of the Standard calibration (see page 3) while judged to be out of specs according to the results of the Custom calibration (see page 4). Therefore we compared the measurement results from both calibrations. It is obvious that in the 22 ma AC Current range the Fail result at 2.2 ma@20 Hz (Row 29) wasn t detected because the lowest measured point of the Standard calibration was 10 ma@55 Hz (Row 6). Notice that even 2.2 ma@40 Hz (Row 30) measured well within specs. The two Fail results in the 22 V DC Voltage range are a completely different story. This range has been measured, but only at the mid-range points of 10 V and -10 V. [NB: In fact 10 V and -10 V are the low-range points of this range, because these values would normally be generated in the 11 V DC Voltage range.] Looking at the measured values of these points (Row 1 and 2), there is no hint whatsoever that would indicate the Multifunction Calibrator to be out of specifications on this range. However, when measured at the high-range points of +20 V and -20 V (Rows 12 and 14) there is no doubt that the unit is out of specifications by quite a margin. To a lesser extent the same applies to the Fail result on the 220 V DC Voltage range. Measured only at the mid-range points of 100 V and -100 V (Rows 3 and 4) the measurement results look okay and again there is no reason to believe that the Multifunction Calibrator would be out of specifications on this range. But then again, when measured at the low-range points it shows the unit to be just out of specifications at 20 V (Row 15) and very close to its spec limits at -20 V (Row 17). Regarding the three Conditional Pass results on the 220 pa range it is clear that they weren t detected during the Standard calibration because this range hasn t been measured at all. Evaluation of the Standard calibration The 5720A got a Pass result on all ca. 150 measured points There were no measured points with a Fail or even a Conditional Pass result TCEM 1208 Standard versus Custom calibration of a Fluke 5720A Page 5 of 6

6 There are no hints whatsoever that would indicate the unit to be out of specifications on any measured function or range Based on the results of the Standard calibration, the calibration provider could (would) decide not to adjust the Multifunction Calibrator because no out of tolerance values were found Based on the results of the Standard calibration, the user of the Multifunction Calibrator would be tempted to judge the 5720A to be completely (= at all values in all functions and ranges) within specifications Conclusions A Standard calibration not necessarily covers all functions and ranges During a Standard calibration often only mid-range calibration points are being measured Sometimes mid-range calibration points fail to show offset and gain errors to their full extent A Standard calibration may not suffice to judge if a certain instrument is completely (= at all values in all functions and ranges) within specifications Guidelines Offset errors are best detected by choosing a calibration point as low as practical in each range, for example 0% - 10% of range Gain errors are best detected by choosing a calibration point as high as practical in each range, for example 90% - 100% of range Amplitude errors over the frequency range are best detected by choosing calibration points at strategically spread frequencies, especially in the frequency extremes Recommendations Define your own Custom calibration points for the calibration of your Multifunction Calibrator: o Include calibration points on all ranges that you are (probably) going to use o Take at least one low-range and one high-range calibration point and eventually an additional mid-range point in all these ranges o For AC ranges choose strategically spread calibration frequencies, especially at the frequency extremes o Have a linearity check performed on (at least) one range These recommendations also apply to other instruments like Digital Multi Meters R.J. Timmerman MiPlaza, Philips Innovation Services TCEM 1208 Standard versus Custom calibration of a Fluke 5720A Page 6 of 6