Hero999 said:
If does need a rectifier as the signal consists of both positive and negitive components, stick an RC filter on the end and you'll loose everything as the positive and negitive sides will cancel.
You need a rectifier to get rid of the negitive side first before you can use a low pass filter. This is exactly the same as tuning into an AM radio station, to get the audio superimposed on the RF signal it needs to be rectified before you do anything else.
I suggested an amplifier because at such low voltages the input ofset error and linear region of a comparator might start to become issues, an amplifier will help make it more stable, although it isn't essential.
hi hero,
The comparator will remove the 'negative component', as its a threshold detector.
Set it for say, +80mV and everything below that value will NOT be output, input signals over +80mV will be above the threshold and will change the state of the comparator output.
The 'spike' component at the output of the compartor will be integrated out by the RC filter.
I dont see any reason to explain it in AM radio terms.
I must point out I have used compartors many times in my products and I have lots of experience in their use.
I still would like an answer from the OP regarding the period times I have asked about.
EDIT: hero,
For reference, I have tested a LM393 comparator using a signal of the same amplitude/polarity and 'noise' as the OP's drawing.
As I still dont know the freq/period of the OP's waveform I have assumed something about 100KHz.
The switching level Vref set at +80mV, with a 10M/10K positive feedback gave a switching threshold of +72/78mVolt[6mV diff]
with a well defined output signal of about +0.2/+4.8v, with a 3K3 pullup to +5V.