PXI-5102 Noise Dependent on Sampling Frequency

[VI_Joe]

I have captured a pulse that's at least 5s in duration with a PXI-5102 Digitizer.  In the attachment, I have zoomed in on the noise, and I'm comparing results taken with a sampling rate of 10K vs. 100K.  The display taken with the 100K sampling rate is less noisy.  Since this pulse is of such long duration, I don't understand why the higher sampling rate makes a difference.  Is it a function of the rise time? Can anyone explain?

Thanks in advance.

[Sarah_Y]

Hi Allen,

After speaking with you on the phone, I just wanted to make some updates about what we discussed regarding your setup and application.

You are trying to measure a 28.5 V signal that is being read across a 1 Ohm resistor.  You are taking 7 seconds worth of data.  Your vertical range is set to 32, sample rate to 10k, and buffer to 70.  You are using an example VI that you modified slightly.  You have 2 5102 cards, one in slot 2 and one in slot 3.  You only see this problem for the card that is in slot 2.  You have tried running this on another test set up (different chassis, different 5102) and saw the same result, always with the card in slot 2.  You see this noise at the 10k sampling rate, but not 100k.  You mentioned that you might be using a 1050 chassis but you weren't sure.

Please post if any of this information is incorrect.  I am going to try and replicate this issue and see if I can find any documentation that would explain why you are seeing this behavior.

[VI_Joe]

Sarah,

• Buffer size = 70k.
• I see this noise at the 10k sampling rate, but the noise is reduced at 100k.  Please see the attachment that was part of the original posting.
• The chassis is PXI-1050.

Something else - I'm using a 10x probe.

[Aaron O]

Allen,

From my understanding, you are measuring a ~28V signal that is attenuated 10X to ~2.8V. Have you tried testing the 10kHz and 100kHz in the Scope Soft Front Panel ( Start Menu » All Programs » National Instruments » NI-SCOPE )? This would help to eliminate any possible configuration problems in the code and a good general troubleshooting step. 

What is generating the signal that you are acquiring? I noticed that Sarah mentioned that you were measuring across a 1Ω Resistor. Can you explain the measurement setup more thoroughly?

The second slot in the PXI-1050 is the Star Trigger Slot. Have you tried both digitizers in different slots, such as 4 and 5?

Any additional information would be great about you're application. Thank you!

Regards,

Aaron

[VI_Joe]

I am measuring 28V directly from my power supply – not across a 1 ohm resistor.

I have not used the Scope Soft Front Panel since I can’t control sampling rate with the Scope Soft Front Panel.  Back to my original question - why does sampling rate have an impact on the displayed noise?

I have not tried slots 4 & 5.

[Sarah_Y]

Hi Allen,

I mentioned that I was going to try and replicate this issue, but after a bit of troubleshooting I found the only PXI 5102 that we have in our department is broken.  While I am working on getting another one to test this issue out, I have a few more questions.  

Do you have any grounding on your input?

Would it be possible to take an fft of the signal you are reading?  There are several possible sources of noise that I am thinking may be causing this problem, the controller, the controller power supply, and the external power supply to name a few and an fft would help determine if it is one of these that is leaking into your signal.  

Do you have any power spectral density information?

You said that you have seen this on multiple chassis, were you using the same power supply when you did both of these measurements?  Do you have another power supply that you could try?

Any additional information will help.  Thanks in advance.

[Sarah_Y]

Hi Allen,

I just wanted to give you an update since we keep missing each other on the phone.  I was able to get a PXI 5102 and tested it out in a 1044 chassis.  I did not see the noise that you were seeing.  However, there were a few differences in our test set ups.  First, as I mentioned, I am using a 1044 chassis.  Second, I am using a 2 V signal supplied from an arbitrary waveform generator, PXI 5412 connected with a BNC cable instead of a power supply and 10x probes.  I have my vertical range set to 10 instead of 30.  I tried using the 10 kHz sample rate and 100 kHz sample rate and saw no difference.  I tried the measurements on both channels.

Based on what I saw I have some more questions about your set up.  

1.  What controller are you using?  Do you have access to a MXI controller and if so can you try using that instead?

2.  What probes are you using?  Model number and any detail you can provide will be helpful. 

3.  Can you post your code?  I did this test running an NI Scope example, but I would like to try it with your code if possible.

4.  Did you use the same probes and power supply each time you saw this behavior?

Look forward to hearing your answers and talk with you soon.

[VI_Joe]

Sarah,

The scope probes being used are most likely the source of the noise.  Co-ax cable was run from the 5102 digitizer to a switching matrix.  From the switching matrix, discreet (non-twisted) wires were run to a test interface box where the measurement is being made.  I would like to know how bad the noise really is.  Is it the measurement taken with the sampling rate set at 10k or 100k?

[Sarah_Y]

Hi Allen,

In order to really know which one is the correct signal, I would need to look at the power spectral density.  There is a function in LabVIEW that will measure it for you (http://labview.ni.com/docs/proposals/6.1/HTMLHelpProposal/HTML%20Conference/WebHelp/LvwaveFFT_Power_...)  If you can take a sample for the same amount of time at both sample rates and then take the power spectral density, it should help figure out what you are really seeing.  Let me know if you have any issues getting this data and I'd be happy to help.  After you have it I can start to look through it to get some more details about where this noise is actually coming from and which sample rate is giving you the "real" signal.