A2LA G112 - Guidance on 5700A/5720A/3458A Artifact Calibration

(Approved by the A2LA Criteria Council on: May 5, 2011)

5720A ARTIFACT CALIBRATION MAC WORKGROUP STUDY

5700A/5700A
March 5, 2008

Purpose of Study
This MAC workgroup was formed to focus on the performance enhancement offered by
artifact calibration for the Fluke 5700A and 5720A calibrators. A2LA presently accepts
Fluke’s claim for specifications based on the calibration cycle of its artifact calibration,
provided the instrument is given a classical external calibration at least bi-annually. The
task group’s intent is to determine what should be done, if anything, to help verify those
claims. Task group members are: Ray Kletke, S. Dizor, D. Deaver, and M. Duffin. Bill
Sorrells volunteered later.

Description of Fluke Artifact Calibration

The referenced Fluke calibrators contain a dc reference standard, dc resistance standards,
switching matrix, a high resolution DAC, and the capability to make precision
measurements. During calibration, these components are used to automatically set up the
zeros and gains of all the DC active components as well as the precision dividers used to
control the output values of the output DC Voltage, DC current and Resistance relative to
the internal standards. Additionally, during Artifact Calibration, three calibrated external
standards including a 10 VDC source, a 1 Ohm standard resistor and a 10 kOhm standard
resistor are applied to the unit’s output terminals. The internal Artifact Calibration
system calibrates the internal standards against the three external standards. The result is
traceable calibration for all dc parameters.

AC calibration includes two functional steps. First, the internal amplifier gains and
divider networks used for AC Voltage and AC Current generation are set at DC just like
the DC components described above. Then the AC-DC difference for these components
is measured using a special internal multirange thermal voltage converter (TVC)
standard. The resulting difference data is used to adjust the flatness of these components
across the specified bandwidth. Actually, Fluke assumes the low frequency performance
of these networks below 1 kHz will be essentially the same as for DC. Therefore, the
flatness correction data is measured using a low-frequency-AC to AC comparison.

AC calibration during Artifact Calibration differs from DC calibration in that no external,

traceable AC standard is applied. This clouds Fluke’s claim that Artifact Calibration is a
valid, traceable process for all DC and AC parameters.

Evaluation of Artifact Calibration by National Laboratories

An evaluation of the Fluke 5700A Series II Calibrator was performed by 3 National

Metrology Institutes including NMi Van Swinden Laboratorium B.V., the Netherlands,
SP Swedish National Testing and Research Institute, Sweden, and the Phsicalisch
Technische Bundesanstalt, Germany. A document describing their analysis of the
performance and acceptability of Artifact Calibration is documented in a book published
November, 1999, titled “Artifact Calibration – An evaluation of the Fluke 5700A Series
II Calibrator” (ISBN 90-9013322-4.

L:\Guidance\G112 - Guidance on 5700A/5720A/3458A Artifact Calibration (5/5/11)

Page 1 of 5
Important final observations from this study include the following:

x For Direct Voltage, Direct Current and Resistance, the Artifact Cal procedure
makes a traceable calibration of the calibrator. Thus for these functions, there is
an unbroken traceability chain.
x The Artifact Cal procedure does not realize a traceable calibration of the AC
functions in the Artifact Cal process. DC external reference standards are used
and an internal AC/AC sensor. The later is not calibrated to an external reference
during the Artifact Cal procedure. However, the Artifact Cal procedure does
increase the confidence in the AC part of the calibration. It does adjust the
calibrator within the 24 hour specification of the calibrator provided that the
characteristics of the internal AC/AC transfer sensor have not changed
significantly since the last external verification.
x The calibration report produced by the Artifact Cal procedure accurately reflects
the changes at the binding posts of the calibrator due to the adjustment, except for
the opposite sign (in voltage and current). Thus, the Artifact Cal calibration
report is very well suited as reference document for building up history of the
calibrator.
x Strictly speaking, Artifact Cal is not equal to a (conventional) calibration since it
does not produce measurement values. (shifts are mentioned in the Artifact Cal
calibration report) and uncertainties. Furthermore, it does not make a separate
uncertainty evaluation for each measurement it performs. However, in practice
Artifact Cal measurement values can be deduced using the values of the constants
in the raw constants report. Furthermore the specifications may be regarded as
the final total uncertainty of the complete Artifact Cal process. Viewed from this
side, Artifact Cal is very similar to a calibration (combined with an adjustment).

Recommendations by National Lab Study

Important recommendations from this study include the following:

x Full calibrations should be made at least every two years

x The Artifact Cal report complete with shift data for all parameters should be
stored by the user
x The user of an instrument should be able to justify the calibration interval he
has chosen.

The National Lab study presented the following recipe as an example of how to collect
data that could justify a longer overall calibration interval. (A) is Artifact Calibration and
(C) is full external calibration.

Page 2 of 5
 Table 1: National Lab Guideline for 5700A/5720A Calibration Scheduling
EA
1 CAC CAC CAC CAC CAC CAC
year
Fluke
1 AC A AC A AC A
year
Study
1 AC AC A AC A AC
year
Study
90 AC AC A AC A A A AC A A A A A A A AC A A A A A
Day
0 3m 6m 9m 1yr 3m 6m 9m 2yr 3m 6m 9m 3yr 3m 6m 9m 4yr 3m 6m 9m 5yr

Note that when the laboratory wants to use the 90d specifications (last row), the
calibration results are verified against the 90d specifications, even when this calibration is
only performed once a year.

Task Group Recommendations

The Task Group recommends that Artifact Calibration be accepted provided that the
recommendations made above by the National Lab Study are followed. It is not
necessary, however, to follow the above recipe exactly. What is important, is to schedule
A and AC so that the performance at the desired calibration interval is confirmed by the
“As Found” data from C. Furthermore, the “Final “data should confirm that Artifact
Calibration has pulled the instrument back into better than 90 day performance limits. For
example, if it is desired to perform Artifact Calibration every 180 days and claim 180 day
performance (as published in the Fluke spec sheet), a full external calibration (C) should
be performed 180 days after an Artifact Calibration has been performed. The “As
Found” data collected during (C) must fall within the specification limits published for
the Fluke calibrator for all parameters. Furthermore, the “Final” data should confirm
operation well inside the instrument’s 90 day specifications. These validations should be
completed before 180 day performance specifications are accepted by the A2LA assessor.
Furthermore, it should be repeated at every external calibration (C). An example of
appropriate calibration scheduling is given in Table 2.

Table 2: MAC Workgroup Guideline for 5700A/5720A Calibration Scheduling

CAL PERIOD
12 Mo AC AC A AC
6 Mo AC A AC A A A AC
3 Mo AC A A A AC A A A A A A A AC
Mo 0 3 6 9 12 15 18 21 24 27 30 33 36
Yr 0 1 2 36

Another important requirement is that the artifact calibration Cal Shift report, complete
with shift data for all parameters, should be collected after every Artifact Calibration and
used to confirm operation within the specifications claimed by the user. It should then be
properly stored. Care must be taken during the calibration to set the 5700A/5720A to the

Page 3 of 5
proper specification. For example, if Artifact Calibration is performed every 90 days and
the 90 day, 99% CL performance is claimed for the 5700A or 5720A standard by the user
laboratory, make sure the 90 day, 99% CL performance is selected during the Artifact
Calibration. This is necessary since the reported shifts in parameter values are given as a
percentage of specification. The user must also be aware that the specification used in
the Artifact Calibration Shift Report will always be the 99% CL specification even if the
95% CL is chosen by the user. Therefore, the user must be careful to either claim
only 99% CL performance, or must change the Artifact Calibration Shift Report by
multiplying the reported Shift (% spec) by the factor 1.3, which is the coverage
factor for 99% CL divided by the coverage factor for 95% CL. Or alternatively, the
accepted shift limit should be reduced from 100% to 77%. A sample from a Cal Shift
Report is attached.

The user should also be aware that the report called Artifact Calibration Report, often
issued after artifact calibration, is a summary report that only includes the data for the
parameter that has shifted the most. Care must be taken to obtain the required Cal Shift
report which contains shift data for all parameters.

It also is important to verify that the “As Found” data collected during the full external
calibration (C) be checked against the performance claimed by the user (90 days, 95% or
whatever). Normally, this data is only compared to the one year, 99% CL specification in
the calibration supplier’s calibration report.

Page 4 of 5
 ATTACHMENT 1
Sample Page from Calibration Shift Report

EXTERNALLY CALIBRATED INTERNAL REFERENCES

REFERENCE Most Recent Previous Shift (ppm)

6.5 V DC Reference 13 6.911804 6.911804 o.oc
V DC Reference 13.819903 13.819903 O.oc
Resistance Reference 0.9999824 0.9999824 O.OC
1.0 Ohm Resistor 1.0000267 1.0000267 O.OC
1.9 Ohm Resistor 1.9000295 1.9000295 o.oc

------------------------------------------------------------------------
DC VOLTAGE OUTPUT SHIFTS
------------------------------------------------------------------------
RANGE Point Zero Shift Full Scale Shift Spec Shift
(+/-) (% spec)

220 mV +FS 0.00000 mV 0.00006 mV 0.28 ppm 11 . 27 2.45

ppm
-FS -0.00000 mV -0.00006 mV -0.28 ppm 11.27 ppm -2.45
2.2V +FS 0.0000000 V 0.0000000 V 0.00 ppm 6.36 ppm 0.03
-FS -0.0000000 V -0.0000000 V 0.00 ppm 6.36 ppm -0.03
11V +FS 0.000000 V 0.000000 V 0.00 ppm 4.27 ppm 0.04
-FS -0.000000 V -0.000000 V 0.00 ppm 4.27 ppm -0.04
22V +FS 0.000000 V 0.000000 V 0.00 ppm 4.23 ppm 0.00
-FS 0.000000 V 0.000000 V 0.00 ppm 4 . 23 ppm 0.00
220V +FS 0.00000 V 0.00000 V 0.00 ppm 6.23 ppm 0.03
-FS -0.00000 V -0.00000 V 0.00 ppm 6.23 ppm -0.03
1l00V +FS 0.0000 V 0.0000 V 0.00 ppm 8.45 ppm 0.02
-FS -0.0000 V -0.0000 V 0.00 ppm 8.45 DDm -0 O?

Page 5 of 5