CDAQ 9201 noise in comparison to PCI 6120

[ceties]

Hello,

I have problem with our cDAQ, more precisely with the 9201 voltage module. The resulting signal seems way too noisy especially in comparison to our 6120. Could somebody have a look into it and tell me if it is expected result.

I measured three things: AA battery, signal from manometer and square signal from signal generator. The white line is data measured from 6120 and the red one is the cDAQ module. I only removed the offset (mean) in the battery and manometer measurements so it is easier to see the noise amplitudes.

Thanks

Sampling rate: 10k

Range -10V;+10V

[Stefo]

Hi Ceties,

I can see you changed picture to krtek cool one.

In general, I can think of grounding. What confuses me a bit however, it seems the peaks measured by 9201 has approximately 100Hz. I would bet it is poor grounding. Even though 9201 should have islolated ground, try to graound cDAQ chassis. Input noise for 9201 is 5 LSB peak to peak, but i can see in pictures you provided it is bit more. I have seen in the past some incorrect readings caused by pood powering. What you could try is to measurel short AI to COM. Or, you could try to isolate power sypply - instead of using of the shelf powering, you could try battery (which might not be really possible for many measurements).

Moreover, 9201 according to operating instruction and specifications is used to measure grounded signal! I haven't found there it should be used for floating signal (your batery). To measure floating signals, try different modulse for better results.

Stefo

[ceties]

Krtek rocks! 🙂 And thanks Stefo as usual for your kind help!

5*(20/*2^12) = 0.02V this is the P-P noise level. Right?

Regarding measurement floating sources - my understanding is that it still should work although there might be possibility of more noise given by different ground potentials. That is case of the AA battery and our manometer (supplied by 9V battery).

In any way with the signal generator it should work as expected.

I tried to measure the AI0 against the COM – image 1. The noise level seems in the range given by 5LSB but nothing super amazing to be honest.

What I also tried is to measure the AI0 while I had a cable in COM. I started DAQ and then I touched the loose end of the cable. There was a raise of the amplitude of the noise. As soon as I released it went back to the previous level. I repeated it couple times. Is that expected behavior? – image2. Plus I grounded the chassis with no effect (screw on the right)

Message Edited by ceties on 06-16-2010 07:51 AM

[Stefo]

Hi Ceties,

fist of all, to the best of my knowledge if you get noise of 5LSB, it doesn't mean that 5*LSB Value, but rather noise can occupy 5 bits. This would turn your calculation into:

32*(20V/2^12)V = 0.156Vp-p, which is still quite bigger than what you can see. However, adding improper signal source (floating, non refferenced) could affect measured signal and put it out of specs.

With your 2nd connection i'm not quite sure what you mean. Did you have AI0 connected to COM, and when you touched cable you could see increase in the noise amplitude? If so, it could be still fine, as we are still not talking about referenced signal, and by touching the cable, you add some more circuity (capacitance) and you act as antena as well (as far as I remember, but I might be wrong).

Or did you connected one cable into AI and other cable into COM, and then touching cables? if so, than you would have floating source with high output impedance (human body), which would cause  signal not to be just noisy, but it could even flow over time.

Anyway, unless the signal is grounded (referenced to common ground), your measurement might be out of specs.

To verify the module works properly, lets try to measure ground referenced signal. Be sure your signal generator doesn't have isolated outputs.

Stefo

[ceties · Accepted solution]

Hi. So in the end the problem was in the chassis itself (not module). We had guy from NI here and tested it against his modules/chassis and we ended up with clear conclusion that there is some problem in our chassis.

Anyway thanks for trying to help!

[LuisO]

Ceties,

I'm glad you got your problem figured out.  Just to set the record straight, 5LSB's of noise does not mean 5 out of the 12 bits of resolution are noise.  This would leave you with a usable 7-bit board 🙂

The 9201 has a full scale input range of +/-10.53V, and it is a 12 bit board.  So, 1LSB = 21.06V/4096 = 5.14mV.  

So, 5LSB peak-peak is around 26mV peak-peak.

Luis

[ceties]

Thanks Luis. So it is as I thought. More or less...