Spikes on the PXI-4071 DMM

[joseospina]

I'm performing extensive tests to familiarize with the PXI-4071 DMM. The program takes several periodic measurements of "0" DC voltage accross one P2 probe connected as a shortcircuit between the DMM input. I then graph and analyze the samples for different configuration parameters of the multimeter.

Everything was as expected until I manually configured the aperture time to 1ms.

All test I've done with this aperture time, independently of the settle time or anything else show periodic spikes on their measurement. The only pattern I've seen is that spikes tend to be of 4 uV of amplitude. Anything else is just guessing, for instance, I guess they are related with the sample time, the power line frequency and some aliasing effect, but i'm nou sure how, since their period is 32 seconds! which is too many PLCs...

I've done different tests for different sample times, and aperture times. I attach the tests results in PDF for those interested in finding the spikes source, or suggesting different test configurations. If I find anything I'let you know.

Thanks
Jose Ospina
Torino, Italy (PLC -> 20ms)

[ChrisR.]

I see no such spikes on my 4071 configured the same as yours. I have a couple of suggestions to help find the problem.

1. Do the same test with a much higher sample rate. That is, keep the aperture the same, but sample as fast as you can consistent with that aperture. You might want to do a "multipoint" acquisition to speed things up a bit. I can get around 900 S/s with a 1msec aperture with multipoint, as the settle time is optimized away for all but the first sample. If you sample fast enough, you can see if the actual frequency of the spikes is higher than you realize, and is just aliased with your 0.5Hz sample rate.

2. Since I don't see the problem here, I assume the noise is being picked up inductively in your probe leads. Can you try your test with a shorting bar across the input, or at least as short of a connection as you can make between the two banana jacks?

Chris

[joseospina]

Hi Chris,

Thanks for your answer.

I have taken tests for sample periods different than 2s. In the PDF file of my original post, starting from test 01-0050, you can find tests for periods of 2.1s, 1.9s, 3s, 1s and 0.1s. I can only arrive to a little more than 100S/s because I'm taking single point measurements. Since my final application will not require waveform acqusition I only want to try it (rearrange my code) if I find it necessary. This is, if taking samples at 900S/s will reveal somthing 100S/s cannot reveal.

I've just done new tests for sample times of 2s, 1s, 0.1s and 0.01s. (0.5S/s 1S/s 10S/s and 100S/s respectively). I've also changed the NI probe that was shor-circuiting the DMM input for a smaller (3cm) copper cable. Spikes still there. I attach the graph of recent tests. Watch out the first graph since temperature effects are big.

The time between spikes actually change. It becomes smaller as the sampling rate is increased. However it seems the "number of samples" between spikes remain somewhat constant, which is not nice since it suggest that it is something the DMM is doing each number of samples without being requested to.

Question: Does your powerline frequency is 50Hz?

Notice also that these spikes are found the same, but of 4mOhms of amplitude, if the DMM is configured to measure resistance in the 100 Ohms range for apperture time of 1ms.

For my application: highly stable temperature measurements, these spikes are not welcome at all...


Thanks
Jose Ospina

[ChrisR.]

I still don't see the spikes here. I've tried the 100mV DCV range and the 100 ohm 2WR range. Can you clarify which range you're using? Your pdf said 100 ohms, but the results were in uV, not milliohms. To answer your question, I live in 60Hz-land, but NI-DMM uses the powerline frequency only to set the aperture when it's specified in PLCs. Since you're setting the aperture to 1msec, powerline frequency is ignored. Of course you could be seeing 50Hz noise coupling in somehow. I have a couple of suggestions:

1. Try putting the PXI-4071 in a different slot in your chassis. I have heard of issues with noise coupling into 407x products from the power supply in the SCXI/PXI "combo" chassis. Which chassis and slot are you currently using?

2. Try one of the waveform measurement examples to sample at 1kS/s. If the spikes are at 50Hz, you should be able to see that. If you have trouble getting the examples to do what you need, I can try posting a vi. Are you using LabVIEW? What version?

Chris

[joseospina]

You are right, range is 100 mV, I made a typing mistake...
I'm using Labwindows/CVI 8.1.

I'll try what you are suggesting. May be the chassis thing...

Unfortunately next week i wont be working so I'll do it in two weeks.

Just in case, may you try with an aperture time of 0.83333 ms? The proportional for a 60Hz period...

Thanks! see you soon.
Jose Ospina








Message Edited by joseospina on 07-11-2008 11:29 AM

[ChrisR.]

I tried the 0.83333msec aperture. No change.

Chris