Unusual voltage distortion on NI-9202 (AI, 10KS/s sampling) - thoughts?

[AndyClegg]

Hi,

We're seeing a bit of unusual distortion in a voltage being measured on an NI-9202. I'm 95% sure its not coming from the LV code reading the IO module, but since it is reading via FPGA, being passed through DMA FIFO to RT side (raw data is then used in FFTs/spectra generation) there is always a possibility. However, the same acquistion works fine without any similar distortion with a NI-9229. Probably related to wiring / grounding, especially as the bumps happen at 50Hz (UK mains frequency), but not obvious.

1) NI-9202 module sample rate = 10KS/s, NI-9229 module sample rate = 50KS/s, FPGA is sampling all data at 10KHz. There are 22 AI channels captured (12 from NI-9202 and 10 from three NI-9229 modules) and multiplex and written into DMA on a point by point basis on FPGA. The RT side reads from the DMA in blocks of 2048 data points (i.e. 22 x 2048) which are demultiplexed. All the channel ordering is preserved, no buffer full conditions, etc etc.
2) The voltage signal applied (sinewave) is from a signal generator, and the frequency / ampltude of the raw data captured on the RT side are all as they should be. The built-in filtering on the NI-9202 is configured to have a "Notch at fs", and seems as it should be per the module data sheet.
3) However, we see "bumps" starting to appear in the recorded data (growing to size in figure) occurring at 50Hz. These bumps/distortion looks more like sudden deviations and then a recovery of the voltage over 10-12 samples, rather than any sort of superimposed 50Hz signal. The image shows a section of raw data for an applied 100Hz sinewave (incl dotted line for the sinewave as it should be). Applying a different frequency of signal, the bumps stay at 50Hz. Also, the bumps are always on the negative side.
4) Perhaps most puzzling or informative thing is that when we connect the signal generator to the AIs on a NI-9229 IO module then these bumps are never seen.

My thinking is that this is due to some increased sensitivity of the AIs on the NI-9202 (compared to the NI-9229) to electrical noise with the wiring arrangement used. Though the shape of the "bump" is making me question that.

Any thoughts / comments gratefully received.

Andy

[ZYOng]

What is the signal source and how did you wire the signal?

Different signal source types and input configurations require different wiring strategies. See Table 1 in Field Wiring and Noise Considerations for Analog Signals

[AndyClegg]

Hi,

The source is a standard signal generator. The field wiring guide is something we are familar with. Unfortunately the panel that that the cRIO is in wasn't accessible when I got this data, so we were not able to confirm exactly how it is wired. The main query is that when connected to an AI on the NI-9229 there are no such bumps at all, but which appear when connected to the NI-9202 module. Also, the shape of the bumps seems unusual compared to what you might expect to see form normal electrical interference / ground loops etc. We'll do more investigation when in next, but it was just in case somebody had seen something similar.
Thanks,

Andy

[VictorZoratti]

Hi, i have the same voltage distortion. Did you get how to solve this problem?

[AndyClegg]

Hi Victor - we just changed to a different IO module (a NI-9239, which has same input circuitry as the NI-9229) and the "bumps" went away (no other changes to wiring from the instruments). We didn't have time to do a more detailed assessment of the original IO module, but I suspect it is to do with the sensitivity of the input circuit of that particular IO module to noise (EMI) on that particular module (even though there isn't anything obvious in specifications).

Have a look at the different wiring options for input channels between the 9202 and 9229 and you will see some subtle differences, and that may give you a clue how to change the wiring connections if you need to stick with the 9202. I didn't have chance to explore at the time.

Hope this helps.