need help/advice on power line noise supression

[halls]

power line noise pick-up

hi!

I am developing a device for proximity sensing using a capacitive sensor of my own, made of two conducting surfaces and a common ground band, big enough for human proximity sensing.

The problem is that power line's 50 Hz are affecting the measures. In this device it is important to stay low power, so, for that, I just sample at very low frequencies (below the 50 Hz...). So I though of a technique which consists on subsampling at a frequency for which 50 Hz frequency's alias is in the high side of the spectrum, and thus, can be eliminated with a simple digital low pass filter...

Can anyone tell me what am I missing? It seemed to work fine, but actually I am still experimenting some background noise which I guess comes from the power lines noise's harmonics...

So, could anyone suggest a better way to reduce this noise, or tell me if I am missing something with this technique?

Thanks in advance!

 

[arunkumar446]

advice about power lines

if your supply is dc and it has ac component then connect a capacitor from that source to gnd. as it wont allow instantaneous changes in voltages across it. as change comes it conducts current. may be for your low power application it may be problem. but try this.

 

[halls]

power line noise

Hi,

I actually didn't explain myself as i see after re-reading my own post... my problem doesn't come from the supply. My device's supply is just a 3.6V battery, so it is not connected to the power line.

The power line noise i am getting is the one detected by the sensor itself, so that's why i was trying to bypass the problem using signal processing...

Thanks for the answer anyway

Still looking forward to reading any solution O

 

[pauloynski]

mains noise pickup during capacitive sensing

Try to synchronize your samplings with the zero crossings of the mains power instead.

 

[halls]

power line noise frequency

That's actually a problem, because my sampling rate is about 52 ms (i can't go lower than 40 ms because of low power consumption issues) and that's why i was trying some undersampling tricks...

 

[pauloynski]

measure power line noise

You need not to sample every time the mains crosses zero. Do a measurement every 3 or 4 crossings to save power. Only do the samples at this point.

 

[FvM]

supression one li...

You may want to explain your sensor principle and particular the involved signal processing more detailed. To my opinion, it's not said, that the interferences are actually caused by power lines, and if so, there may be very different means to fight them, not just synchronizing. Also, a device without a mains connection wouldn't be easily able to synchronize.

Apart from finding a possible solution, you should be able to verify by suitable tests, if power lines are the source of interferences.

 

[halls]

power lines noise

My sensor behaves like a variable capacitor, changes its capacity with the influence of something altering its dielectric (a person). When charged for a short but constant period of time it gets some voltage between its terminals, which is also variable depending on its own capacity. Then what I sample is the voltage at the time of measuring. Depending on the proximity of something altering the dielectric that voltage will increase or decrease.

So the basic process of sampling consists on charging the capacitor, measuring its charge and then discharging it again. This is done every 52 ms (19.5 Hz approx) for the reason explained in the first post.

My problem is not determining where the noise comes from. After many tests I know it is the power line. When left on a table the device works perfect, but when attached to a wall (where power line is going through) is when I see the noise.

As I said in the first post, I tried subsampling at some frequency where the 50 Hz would have an alias in the upper side of the spectrum. That's how I can get rid of the main noise frequency.

But line's harmonics appear to be the problem now... so that's why I'm asking for help now. I also though of developing a comb filter, but really not sure if it would work under an undersampling condition...

Thanks in advance to everyone one more time, as I see this post is getting rescued

 

[FvM]

line noise supression

As a rather general answer, you have to use a band limited detection method. This can't work by taking only samples at 19.5 Hz. You need a lowpass or band-pass filter as well. An ADC that integrates the input signal over the full measurement period (e.g. a dual-slope one) would also achieve a stronger attenuation of higher frequency components than sampling, but possibly not sufficient. You should also consider, that using a line frequency sub harmonics of 10 Hz most likely gives better results, at least in case of a stationary interference.

As another point, the sense amplifier in front of the filter must operate in linear range, without being overloaded by interfering voltages, may be either line voltage or any present RF, e.g. from a mobile.

 

[puviarasu]

reducing power line noise

instead of using single ended AMP use a differential amp it will eliminate the effect of HUM or common mode noise

 

[Audioguru]

powerline noise in phone lines

My old record player (remember vinyl records?) was extremely sensitive but did not pickup hum.
It used shielded audio cables to eliminate the hum from nearby power lines.

 

[halls]

power line noise suppression techniques

At last I could get rid of the noise using some software controlled switching system. In the charging period of the capacitors I close a switch that links the big panels to the sensing circuit, and when I want to get the measure, just open the switch so no interference comes from the panels...

Problem solved

Thanks everyone for the replies!