4071 DVM apparent measurement error, voltage measurement drops with test system run time.

[terryatsbe]

I am tasked with supporting a capacitor voltage withstand test system built by an outside contractor. The system measures the voltage across 6 capacitors under test, these are very large value capacitors, say up to 10mF and up to 2KV. The system uses a PXI chassis, 4071 DMM, with a 2527multiplexer. The voltage across the capacitors is sensed via LEM sensors and transformed to a maximum output current ~50mA through precision 100ohm resistors. These voltages are fed into the multiplexer. The system has been on line and operational since 2010. We have seen in the past months a gradual deterioration in voltage accuracy and repeatability. Discussions with LEM educated us to the fact that LEM had problems with the original design voltage sensors, new ones were available with those issues "designed out". Replacing the sensors, and installing linear power supplies provided much improved uniformity of measurements with less random variation in measured voltage. When the test system is turned on, it works as designed. As production lots of 6 capacitors are tested, the measured voltage falls equally on all 6 parts. The programmable Lambda supplies are sourcing the exact voltages requested. It seems that this is a DMM problem. Before I submit a purchase request for a ~$3500 4071 module, I'm open to any other possible ideas to resolve this issue. I am completely unfamiliar with how to program NI systems. There are no error messages displayed on the tester GUI, the tester is controlled by a generic desktop PC. The software runs without problems. Any ideas? I sure thank any folks for thinking about this issue. terryatsbe

[hfarley2]

Hi terryatsbe,

Before assuming that the DMM is at fault I would check to make sure that the signal is still good on the other side of your 2527 multiplexer. In order to do this I would simply probe the wires that connect your multiplexer to your 4071 with an external DMM that you know works properly. If the voltage level with this is correct and does not exhibit the drop you are seeing this tells you it is not the 4071 that is the issue bit rather the 2527 multiplexer or some other external circuitry. If the signal getting fed into your 4071 from your 2527 multiplexer is correct then that means your DMM is not working properly. 

There is a chance the issue is realted to your 2527 multiplexer module as well given the age of your hardware. In order to figure this out you will want to isolate all components of your system and see where the issue arises.

Best,

hfar2

[terryatsbe]

Thanks for the ideas!  Turns out that my "poking around" appears to have "fixed" the problem,  as in "it's working better than in months".   Coming from the point of view that

before getting my MSEE I did over 15 years in the home entertainment service business,  I know for a fact that the problem will return.  I just hope it is one of those

"really long term intermittent" situations.....  Since the tester is "working",  any further diagnose effort is futile.

One thing I tried was to input directly to the tester DVM the correct voltage that represented that which the tester would accept as "good".  Since only one part out of 6

was being tested, and the multiplexer looks at all 6 test stations,  the software error handling picked up inappropriate voltages at the unused test stations and initiated

an emergency shutdown, including the control computer.  Although complicating [!] a diagnose process,  I would argue that it was an entirely appropriate response

to provide another layer of operator safety.

When trying your suggestion to monitor the input with a DVM,  I learned that all hand held DVM test leads are not created equal.  The test leads I cannabalized

to implement your suggestion had pins just a crumb too small to connect with the input pins of the Fluke hand held I was using.  I may be nearing 70, but I

can still be trained.....  I hope retirement remains many years away!

Cheers!

Terry