DMM High Value Capacitance Issues Using Fluke 5520A

[Hawaii596]

Thought I'd toss this issue out to see if anyone has similar problems, and perhaps knows the solution.

I calibrate an HP 973A DMM using Fluke 5520A Calibrator.   One value in OEM manual is 900 uF.   On this model (I have a few of them that all have the same problem) a 900 uF output from the Fluke 5520A yields about 655 - 690 uF reading.   I connect other DMM's to the 5520A for the same measurand and get good values (in-tolerance).   I measure an electrolytic at approximately 900 uF with another LCR meter (HP 4261A, Fluke 187 DMM), determine approximately 900 uF C. V. , and measure it on the HP 973A and get in-tolerance readings.   

Using the same experimental technique, I determined the problem begins around 100 - 200 uF and higher.

I hypothesize that the Fluke 5520A synthesized capacitances and the test signal from the HP 973A DMM are incompatible.   I believe the test signal from the HP 973A DMM is too low to function properly with the Fluke 5520A.

Anyone have any thoughts on this?  Is this is common problem?  Does Fluke have any White Papers, App Notes, or notices about this?  Or does everyone out there just give in and use discrete caps for such readings?

Thanks all for any thoughts.   -Jerry E.

[flew-da-coup]

Thought I'd toss this issue out to see if anyone has similar problems, and perhaps knows the solution.

I calibrate an HP 973A DMM using Fluke 5520A Calibrator.   One value in OEM manual is 900 uF.   On this model (I have a few of them that all have the same problem) a 900 uF output from the Fluke 5520A yields about 655 - 690 uF reading.   I connect other DMM's to the 5520A for the same measurand and get good values (in-tolerance).   I measure an electrolytic at approximately 900 uF with another LCR meter (HP 4261A, Fluke 187 DMM), determine approximately 900 uF C. V. , and measure it on the HP 973A and get in-tolerance readings.   

Using the same experimental technique, I determined the problem begins around 100 - 200 uF and higher.

I hypothesize that the Fluke 5520A synthesized capacitances and the test signal from the HP 973A DMM are incompatible.   I believe the test signal from the HP 973A DMM is too low to function properly with the Fluke 5520A.

Anyone have any thoughts on this?  Is this is common problem?  Does Fluke have any White Papers, App Notes, or notices about this?  Or does everyone out there just give in and use discrete caps for such readings?

Thanks all for any thoughts.   -Jerry E.

This is a problem that I have seen in other model DMMs. I also agree with your Hypothesis.

[LarryH]

I seem to recall a Fluke service note on this issue.  The DMM's use a charging voltage and time measurement to calculate the capacitance.  This voltage is not always capatible with the synthesized capacitors in the 5520A.  I'm going off of memory, so I might be off a little on my theory.

Use plug-in capacitors or a good decade capacitor.

[scottbp]

We have Fluke 5500As, a 5520A, and a Wavetek 9100 in our inventory, and we've found that the 9100 can calibrate capacitance meters that the 55XX calibrators won't, but oddly enough, we can't get a readout of the capacitance of a 9100 using our GenRad 1689M RLC Digibridge, which we use to verify the Fluke 5500As.

We haven't tried to verify the 5520A yet, the capacitance goes beyond the capability of the 1689M, but Fluke did publish a paper on how to do it using a Fluke 5700A and a HP3458A.
Go to http://us.fluke.com/usen/support/appnotes/default.htm, scroll down and click on Fluke Technical Papers, and look for the one titled "Supporting new calibrators with existing equipment". (I think you have to register to download that one). The method described uses the 5700 to charge the capacitance in the 5520, and the 3458 to measure the I*dV/dt to get C (for those who know a little calculus).

[Hawaii596]

In my massive (nearly obsessive) web searches last night, I did read the paper mentioned in the previous post about calibrating the 5520A.   Thanks for that input.

In an attempt to act like I know what I'm talking about, remembering that capacitance is current multiplies by [charge voltage divided by charge time].   I BELIEVE the 5520A actually synthesizes that through common circuitry used to produce other electrical outputs. 

so in my OVER-analysis of my problem, I connected an Infinium oscilloscope in parallel with the HP 973A connected to the Fluke 5520A output while the 5520A was outputting 900 uF (knowing the scope applies a parallel capacitance path - at 1 megohm setting).   I tested a Fluke 187 meter (obviously they make sure their own products work properly together), which read right on the money at 900 uF.

There were some differences between the test signals of the Flk 187 and HP 973A.   My final opinion is that the HP 973A outputs a lower amplitude signal for its capacitance measurements, which in turn doesn't hit thresholds to interact adequately with the Fluke 5520A's synthesized capacitance "simulation. "  Therefore, the culprit I believe is both:  The HP 973A has a low quality signal and the Flk 5520A has threshold problems.

So to get the unit out the door, I had a tech measure an 850 uF electrolytic (close enough to the OEM's prescribed value) using an HP 4261A LCR meter, and alter the data sheet (we take full data on everything).   

The most frustrating part of this is that we actually sent the unit out for repair, figuring my calibrator should do everything it says it will do.   

To keep myself out of trouble for saying negative things about Fluke - I have very high regard for their products and their integrity.   So any of you newbies out there, please understand I believe for it's intended range of applications, the Fluke 5520A is a very good calibrator and our 6000+ unit per year lab uses it constantly with very little difficulty.   So when picking out multi-function calibrators for mid-range meters, I still wouldn't use anything else.

[OlDave]

Thanks for the indepth analysis and explanation. I'll try to store this away in my feeble, forgetful mind in case I ever run into a similiar situation. It's the sharing of information, problems and solutions like this that really makes this forum valuable.

[CalLabSolutions]

Keep you SS-32 standard caps. .   The 5500 and 5520 are not that accurate on capicatance.   

One big mistak a lot of techs make is the 5520 is not that accurate.   Unless you are calibrating them every 90 days you are not able to make 4:1 accuracies on a lot of handheld meters.

If you are calibrating Fluke 87s with the 5500 \ 5520, you better check your uncertanties before the auditors do.   
 

[USMCPMEL]

What are you talking about? Please give a specific example? Last time I checked the 5520 was more than capable of calibrating almost every handheld meter I have ever heard of.

[Hawaii596]

I got a little nervous reading the earlier post and ran a few numbers.  I checked a Fluke 87 DMM.  There is, I believe, a typo in their manual (first measurement point listed as 5 uF, which I believe should be 5 nF.

Anyway, here are some of my numbers for capacitance on Fluke 5520A:

READING at 0.0500 uF Range = 0.0470 uF (Tolerance: 0.0462 - 0.0478)
  Standard Limit:  0.0467825 - 0.0472175
  Ratio (rounding to Fluke 87 Resolution): 4:1

READING at 0.500 uF Range = 0.470 (Tolerance: 0.462 - 0.478
  Came out the same as above, but move decimal one place left.

READING at 5.00 uF Range= 1.00 uF (Tolerance: 0.96 - 1.04)
  Standard Limit:  0.9965 - 1.0035 uF
  Ratio:  11.43:1

We do a number of other handheld DMM's of other varieties.  I'll check a few more.  But my recollection was that the 5520A was designed to be just good enough.  However, as the 4:1 ratio is on the borderline, It's reasonable to assume some of them may be a little less than 4:1.

[USMCPMEL]

Thanks. He also goes on to say that it is not accurate enough for ALOT of hand held meters? I have never come across a handheld that it was not accurate enough for a few benchtop meters but no handhelds. I guess I should also state that I am using the Fluke 5520A.

[CalLabSolutions]

Here is a more detailed example of one specific calibration most people would use the 5520 or 5500 and never check the specifications ..

**Please Note when we are calculating the pass/fail limits based on the manufacturer's specifications not the rounded pass fail limits in the manual that are some times wrong.  We recalculate, check and double check every test point.  The software our customers use is engineered automatically rounds down based on the UUT's resolution when calculating the Pass / Fail limits.  So out numbers will not always match the numbers in the manual.

In this example the 4.7 nf test point whos unrounded limits are (4.623 to 4.777 nF) are rounded down to  (4.623 to 4.777) because the UUT can not read .001 nF it's resolution is .01 nF.  However keep in mind with .077 / .0335 you get 2.298507 to 1 and with .08 / .0335 you get 2.38806 to 1. 

Using the Fluke 83, 85, 87 Series III Service Manual PN 688645 Apr 1998

with the 5500 / 5520 on one year interval 
Step 3.24.2 0.470 µF the TAR is 3.5 to 1
     +/-.007 µF /  .002175 uF (.25% of IV +.001 uF)

Step 3.24.3 .0.0470 µF the TUR is 3.5 to 1
     +/-.007 µF / .0002175  uF (.25% of IV +.1 nF)

Step 3.24.4 4.7nF the TUR is 2.3 to 1  (with the 5500)
     +/- .07 nF / .0335nF (.5% IV + .01 nF)
Step 3.24.4 4.7nF the TUR is 3.5 to 1  (with the 5520)
     +/- .07 nF / .02175nF (.25% IV + .01 nF)


or with the wider specifications in the manual
Step 3.24.2 0.470 µF the TAR is 3.6 to 1
     +/-.008 µF /  .002175 uF (.25% of IV +.001 uF)

Step 3.24.3 .0.0470 µF the TUR is 3.6 to 1
     +/-.008 µF / .0002175  uF (.25% of IV +.1 nF)

Step 3.24.4 4.7nF the TUR is 2.4 to 1  (with the 5500)
     +/- .08 nF / .0335nF (.5% IV + .01 nF)
Step 3.24.4 4.7nF the TUR is 3.6 to 1  (with the 5520)
     +/- .08 nF / .02175nF (.25% IV + .01 nF)