Thanks for that suggestion. I am doing that in some version of my cards, but I didn't fully understand the last sentence. Before that, let me tell you how I have been doing it.

1. Shift the waveform up, by ADC fullscale/2 volts. Sensing voltage is attenuated enough to make sure that maximum input will not pull the waveform below zero.

2. Sample the waveform continuously till I get the peak value repeating. Peak to peak is my wave. Instantaneous reading - 2.5 volts is my actual value.

3. Now I can compute RMS and power factor.

4. I always distrust the above method for powerfactor calculation, even though for most methods, it seems to be working. Shift betwwen voltage peak and current peak is used for PF.

Please comment on point 4.

Now, in your post, you have asked me to implement a highpass filer. Highpass filter, I assume is to remove the DC offset that has been added. If that is the case, do I have to subtract from it again to get symmetric data? Won't this have more overhead than subtracting 2.5 volts directly from each adc value? (Haven't done any filter till now except LPF. I am using ATMEGA16). 'Floats the measured curve' was also not clear..