Adaptive Filter for varying frequency noise

[nlosc]

In our accel and strain data acquisition, we are seeing a varying frequency noise on the signal. It is guessed that it is some wireless communication signal getting at the wires. I would like to propose an adaptive filter using a LMS approach. Or whatever is suggested. It does not need to be real time.

I want to keep it simple, but also I want a chance at it working correctly at reducing the noise and keeping integrity of the data. I don't really have the time to make it a PhD study either. I would like some advice on a no nonsense adaptive filter algorithm approach. I use matlab and simulink.

I am usually not into these forums, but I have found some extra time and have been reading some of the threads. And have posted a couple threads.

 

[GregLocock]

Well, I don't want to waste your valuable time, but how would you differentitate between signal and nosie, for this adaptive filter?

Cheers

Greg Locock

 

[sms]

I am not real sure what your test set up is like, or what you are actually trying to measure, and I would really like a more in depth description of this "varying frequency noise", but I would suggest that you ring out the wiring rather than trying an adaptive filter. Are you sure the noise is not off your ground? Have you tried a different ground? Are you using high quality low noise cables and connectors? Are you sure that the transducers and cables are isolated? Some types of cable are very sensitive to vibration, are you sure this is not from the cables whipping about? What kind of dynamic range are we talking? Are you trying to resolve a micro g level signal? Have you unplugged everything to make sure it is not being generated internally to your analyzer? Are you useing the right impedance setting or input on the analyzer?

In my opinion, you risk throwing out valuable data with any filter. It is better to optimize the measurement setup.

 

[sms]

Oh, here are a couple more, how are your accels mounted? Do you have some base strain issues? Have you torqued them properly if you are using a screw connection? I have also seen accels show strange thermal sensitivity, is something blowing on them that would cause them to see a fluctuating temperature?

 

[IRstuff]

This is cross-posted here:

http://www.eng-tips.com/threadminder.cfm?spid=236&newpid=236

TTFN