suppress power line noise without using notch/band stop filter for eddy current sensor

[Haristhota]

I have a requirement to suppress powerline interference (50Hz Noise) in voltage data of an eddy current sensor.

It has a stand-alone signal conditioner but still  I am getting 50Hz noise in the signal, due to this I could not able to get accurate measurements. I tried implementing zero phase band stop filter in the range of 48 to 52Hz but still I am losing accuracy in the certain range of readings. Basically, I don't want to use any filter. How to suppress the noise without using any filter?

[ARAO]

Hi Harshithota,

Try out the suggestions in these links

http://www.ni.com/white-paper/3344/en/

http://www.ni.com/newsletter/50262/en/

http://digital.ni.com/public.nsf/allkb/632470D5285E52328625629800537049

When you mention that you are losing accuracy during filtering, it means that the frequencies of interest for your application are close to the 50Hz noise. Is this correct? Or is the frequency of interest in some other range?

Cheers,

Rao

[Henrik_Volkers]

If you can't avoid the line hum, I suggest the tone detection.vi to eliminate the line frequency and it's n harmonics.

The tone detection vi allow a frequency search region, start with about line frequency and with the exact line frequency measured detect the higher harmonics.  The vi has the option to just remove the detected frequency 🙂

One starting point migth be the vi I posted here

(latest version https://forums.ni.com/t5/Signal-Conditioning/Filtering-Noise-amp-Its-subharmonics/m-p/3327695#M8543)

If you have more questions, post a vi with a diagram containing the data (or a tdms/lvm file) ...  and tell us what you need to quantify 🙂

[Haristhota]

Hi, Rao

Thank you for sharing the links, i must follow those suggestions by NI to suppress the noise.

[Haristhota]

Hello Henrik,

The posts suggested was actually posted by me only. I was using that VI for separation of harmonics & Non-harmonics for spindle run analysis. It is working very well, thank you for your suggestions.

This problem is arising when I am running spindle at 3000 RPM. At this condition, the fundamental peak in FFT should appear at 50Hz, due to filtering effect this component is getting suppressed so, I am unable to detect the actual speed.

I should find alternative technique to suppress the HUM

[Henrik_Volkers · Accepted solution]

So use a spare channel, capture the line freq&phase (litlle transformer and a resistor divider)

do  'zero' measurements, measure Hum of your setup  (&line) , before and after test 

compensate the hum 🙂

your 'zero' measurements need to be as close to reality as possible ... so engine turns at 3000rpm , tool installed (magnetic field) ??   Sensor covered??    

usually there is some slip , so there should be a small difference in line and engine frequency.  If you have a power converter running the engine and controlling the speed: measure sligthly below and higher than line freq and you migth find that there is no difference 🙂

[johnsold · Accepted solution]

You have already been given some good advice. 

One thing I consider important when dealing with power line interference is to determine how it is getting into the measurement. Often the solution becomes apparent when the source coupling mechanism is identified.  If you can keep the interference out of the measurement, then you do not need to do any sophisticated signal processing later.

One common coupling mechanism is ground loops. Another is inductive coupling. Teh solutions for these cases may be quite different.

Lynn

[Haristhota]

Problem Solved. Thank you all for your valuable advice. Especially to Volkers & Johnsold

[Henrik_Volkers]

Good 🙂  , can you tell us (or others, for the future) how you did it?

Always curious 😉   😄

[Haristhota]

As per johnsold advise, first I figured out the grounding issues with equipment.

The spindle was connected to a heavy 3 Phase motor and entire setup was placed on a cast iron test bed.

I ensured proper grounding of test bed, motor, and my sensor mounting nest.

then I was able to see reduced amplitudes of 50Hz, after that, i moved my signal conditioning setup away from the electrical cabinet and power supply lines of the motor. Now the 50Hz component is not that significant.