NI-9203 generates noise pulses when acquired

[smacire]

Hello,

The capacitor in parallel worked for me.

"A simple solution is to put 22µF (or more) capacitor in parallel from that channel to the ground and forget about this problem. "

[OfirHarari]

Thanks Alex,

I would like to sample the pressure reading at 500Hz. How do I know that using the capacitor will not effect my readings? which capacitor to use? When I measure I'm looking for sharp pressure variation to detect vortex.

Thanks,

Ofir

[Alex2012]

Well, make a RC filter, add resistor in series and capacitor in parallel.

The cutoff frequency (in Hz) and corresponding values for resistance and capacitance you can find in following table:

		C, µF
R, Ohm	0,1	2	10	22	40
5	318310	15915	3183	1447	796
10	159155	7958	1592	723	398
20	79577	3979	796	362	199
33	48229	2411	482	219	121

The higher the capacitance the better attenuation of parasitic noise there will be. So, 22 or 40µF is a good choice.

You need to experiment a little. You will need to decrease capacitance if you see that you loose sensitivity for your pressure measurements/vortex detection. In addition to RC or C filter, perhaps you need to sample data with 5kHz speed and apply some light moving average or median filtering or even FFT filtering. It is unfortunate problem with NI-9203 and there might be no simple and single solution for your particular case, but a s little more complex solution might work just fine. Let us know your final solution and post it here please. Good luck Ofir!

[OfirHarari]

I am replacing those modules with 9253. As I understand they do not have the noise issue. 

What do you think? do you have any experience with 9253 module?

thanks,

Ofir

[K.C]

placing 22uf will affect the sensitivity very much. It will supress the sharp peaks and important waveform  data present in the measuring waveform.

Wrong solution!

[Alex2012]

---

@K.C wrote:

placing 22uf will affect the sensitivity very much. It will supress the sharp peaks and important waveform  data present in the measuring waveform.

Wrong solution!

---

Well, given the fact the you have not providing any solution at all, your poor judgement about what is wrong is certainly 100% wrong! 😋

In the table from the above posts you may find a cutoff frequency values for RC filter. So, by choosing a combination of resistance and capacitance a smart engineer would select a suitable passive components allowing not to filter out "sharp peaks". One would need to find a compromise between the acceptable level of the noise (originating from NI-9203 itself!) and that cutoff frequency. One can also analyze that noise and, based on own measurement requirements, build a LC or RCL or even an active filter. A 22µF capacitor worked perfectly for my application, but for somebody else's application a smaller capacitance might need to be used, sure. What's the point with your comment?

Actually, I just found that in 2021 (I posted this issue in 2016) NI confirmed this issue (after 5 years!) and even issued a workaround solution: https://knowledge.ni.com/KnowledgeArticleDetails?id=kA00Z0000019RcsSAE&l=sv-SE

Funny, but they recommend a capacitor too! Or, to down-sample the measured data, yeah, your sharp peaks will be gone anyway... 🤣

So, dear @K.C, please tell to NI that they are wrong with the noisy input module design and their workaround is also wrong!

[Alex2012]

---

@OfirHarari wrote:

I am replacing those modules with 9253. As I understand they do not have the noise issue. 

What do you think? do you have any experience with 9253 module?

thanks,

Ofir

---

I have not tried this module myself, but I do hope that NI-9253 will not have such noise. I would recommend to contact NI and ask them if this module has any issues similar to https://knowledge.ni.com/KnowledgeArticleDetails?id=kA00Z0000019RcsSAE&l=sv-SE

Depending on your country and/or agreement with NI, you may ask them for demo module and test it. If it is noisy then return it!

Dear @OfirHarari, please let us know in this thread, whether the 9253 has noise or not. Thanks!

[Alex2012]

---

@OfirHarari wrote:

I am replacing those modules with 9253. As I understand they do not have the noise issue. 

What do you think? do you have any experience with 9253 module?

thanks,

Ofir

---

(something is odd with reply here, apologies if duplicated answer)
Unfortunately I have not tried 9253. Depending on your country and agreement with NI, perhaps you can ask them if 9253 suffers from the noise as 9203. And maybe NI can borrow a demo 9253 module so you can try it and quantify the noise.
Dear @OfirHarari please, let us know the results! Thanks!

[K.C]

I have used both the modules(9203 and 9252), where 9252 is an updated module don't have any noise issue, because which is simultaneous sampling inside, where as 9203 is multiplexed architecture.

Previously we used 9203 and switched to 9252. In my opinion the data acquisition is  "Acquire the signal as it is without any compromise." That is what the customers always say to us.

[Alex2012]

---

@K.C wrote:

I have used both the modules(9203 and 9252), where 9252 is an updated module don't have any noise issue, because which is simultaneous sampling inside, where as 9203 is multiplexed architecture.

Previously we used 9203 and switched to 9252. In my opinion the data acquisition is  "Acquire the signal as it is without any compromise." That is what the customers always say to us.

---

For many customers and applications there is a need to use acquired data as a feedback for example to a PID controller or other real-time steering algorithms/units. A noise will be very problematic. So raw noisy data is unacceptable there. One can use digital filtering of course but latency and performance might an issue and you dont want to waste precious LUTs/slices in your FPGA code. So often a simple capacitor connected in parallel is a good and cheap solution. Even NI recommends it as I posted above.