How to reduce the fluctuations in the acquired signal

[GiridharYadav]

Dear Sir,

In one of my application I am acquiring 0 to 10VDC voltage signals and 4 to 20 mA current signals using NI 9201 and NI 9203 modules respectively installed in the cRIO 9073 system.

But I am unable to acquire stable signals. Both voltage and current signals have fluctuations ranging between 0.01 to 0.03 with which the output scaled signal is varying very much. But the same signals connected to their respective indicators were indicating the stable signal without any fluctuations.

For your information I have used the shielded cable between the sensors and the DAQ modules. And I have also taken proper measures in grounding the signals as suggested in ni.com. Of course the environment from where the signal acquisition is some what harsh (where 50 HP motor is placed). If the noise is induced in the cables then the value displayed in the indicator should also fluctuate, which is not happened here. I have also noticed these fluctuations before I make ON the motor.

From this I got to a conclusion that the indicators are having something which the DAQ modules won't.

I had also tried averaging the few (10) samples and displayed, after which only the fluctuation came down to 0.01 to 0.03. So there is no much affect over the signal. If I try to increase the no. of samples for averaging, the update rate of the signal is getting effected.

In the multimeter and in another DAQ system (eSOMAT) also I didnt find any fluctuations.

So In this regard I request NI as well as the senior professionals to help me out and let me know whether should I use something before the DAQ module which makes the signal fluctuations free or stable signal.

Thanks and Regards

Giridhar

[Nitin_Ahuja]

Dear Giridhar I was also facing the same problem with NI 9203 and cRIO 9073 and the fluctuation in the signal is up to 0.04 mA I got a solution here. I used VI in FPGA functions FPGA maths and analysis. I found that this fluctuation in signal is because of 50 Hz. frequency componet available with the signal. Then I used Butterworth low pass filter with 48 Hz. cut-off frequency and i received the signal from filter is stable up to 4 decimal. I also tried Notch filter with notch frequency 50 Hz. and a bandwidth of 2 Hz. to 4 Hz. & i received siganl is stable up to 3 decimal. For my application i need a slow update rate so both filter are fine but notch filter is more preferable just to avoid the loss of higher frequency components in the signal. For your applcation i think first you have to check what is the major freqency component of noise by taking the FFT of signals, I hope it is helpful for the selection of filter for each channel independently. Best Regards Nitin

[tryu]

Hi Giridhar, 

There are few points which can be said and done in this regard..

First of all what are the kind of sampling rates that you are handling in this application .. and more importantly , what kind of update/sampling rates does the application demand. What kind of sensors are you using for these inputs ? whats the typical output rate of those sensors ?These factors would play a deciding factor for suggesting the right solution.

Also , as you said .. the data in the indicators is averaged and wont show the sensitivity to noise , the sampling rates are different in that.

It is advisable to reduce the sampling rates of the modules/ the FPGA code in case you dont need very high sampling.

Its important to justify the choice of timing (sampling rates) with the application at hand. 

Regards
Rishi 

[Yixiao-uncle]

Hi to everybody.

I need to come back on this issue.

Im using a c-RIO system in scan engine mode to acquire data from a small power plant test rig and to controll some parameters for safe pourpose.

Fot this reason i need to scan channels every 10 ms, and by this way also very stable signals look foll of noise (more than 20% fs).

This behavior is not acceptable for safe loops, like fast switch off of plant.

I improved the acquisition and control by installing capacitors between AIx and COM terminals of c-RIO 9203 boards.

Does anybody have experience in this sense? how correct is to use capacitors without interfering too much with real signals?

In your opinion a more stable Vdc supplier can improve the situation from noise point of view??

Thak You very much

Antonio

[GiridharYadav]

Hi all,

Using of Lapp cables has drastically reduced the noise levels.

Thanks and Regards

Giridhar

[Jon_Z]

Hello all,

   I am developing a program for cRIO and i have the same problem, i must measure a pressure sensor (4-20 mA) with NI 9203 and do some control every 10 ms but my analog signal is fluctuating.

   I used time averaging filter and the signal quality improved. 100 samples per second is not enough for a good signal, usually i use cDAQ faster than 1000 samples/second and then average samples for reducing measurement noise.

  With cRIO in scan mode I can not sample signals fast enough so now i am programming the FPGA for acquiring signals at 2400 samples/sec and calculating average every 48 samples. This way i am averaging signal every 20 ms improving noise cancellation for 50Hz noise from electrical plug.

  Best regards,

Jon

[Mn-pdp]

Hi,

I use the NI 9227 to receive the analogue signal from 4 different pressure sensors and the signal is 4-20 mA. When I read the signals, all 4 readings have the fluctuation problem.

I have tested the sensors and they are good, The wiring is good. I used a current source instead of the sensors and I got the same result (fluctuation).

I replaced the NI 9227 module with another one and the result was the same (fluctuation).

I used different program and NI examples to read the signals and the result was the same (fluctuation).

I am really confused. I am wondering if the reason is known and if there is any solution  for that.

Thank you!

[Guru786]

Hi,

We have encountered the similar issue with NI-9203 where we have required to acquire 4-20 mA signal from two flow and two pressure sensors. We have resolve the issue by analyzing the frequency and noise components using FFT and then we have used the Low pass Butterworth filter. In our case, we have set the lower cut-off frequency to 50 Hz and mostly it all the case and got stable result up to 4-5 decimal place. Attached a demo program for your quick reference. 

Thanks ,

Sarafaraz Alam

[AmitKumarPal]

Same problem here. Somehow readings from DAQ won't stabilize enough, when compared to Indicators.Can anybody Help?