[time-nuts] Simple AC mains zero-cross detector

Charles Steinmetz csteinmetz at yandex.com
Sat Dec 20 03:24:08 EST 2014

Gary <nuts at lazygranch.com> wrote:

>I try to minimize dangerous voltages. Anyway, the filtering reduces the
>slew, so you can't have it both ways.

Starting with 120v gives you 10x the slew rate that starting with 12v 
does, whatever filtering you use.

>If by post processing you are averaging, then you certainly have lost
>frequency variation data. Averaging is a filter.

You will not lose grid frequency variation data unless you average 
the 60 per second samples for *extremely* long periods of time, 
because the grid frequency is generated by rotating machinery 
weighing many tons that can only change frequency very, very 
slowly.  As I noted before, the simple system I described resolves 
frequency to better than 0.01 Hz in one cycle, so very little 
averaging is needed to achieve better resolution than anyone really 
cares about.  As long as the averaging function is more agile than 
the actual grid (and it will be under all practical conditions), all 
actual grid frequency variations will be preserved.

You proposed a method using steep hardware filtering, which 
presumably you do not think loses frequency variation data.  The 
system I described can easily duplicate whatever filter you propose, 
in post-processing.  So either your proposed system cannot track grid 
frequency variations, or my [built and tested] system can.  You can't 
have it both ways.

>If the event is due to noise, you resolved essentially garbage to a

No, you resolved a grid phenomenon in which grid-nuts are interested 
to within a microsecond.

>If you average, you have done filtering.

Yes, but for the reasons given above and in the other messages in 
this thread it is benign filtering that does not obscure any of the 
grid voltage features grid-nuts are trying to record.  Furthermore, 
it is done in post-processing so you can re-do it at will to resolve 
whatever you want to resolve.  You proposed a scheme with steep 
hardware filtering, which does not have this flexibility.

This is getting tedious.  If you are interested in grid logging, 
please build one of the simple systems I described and one of 
whatever system you think will work better, and present data to 
support your conjectures.  Until then, you really aren't contributing 
anything useful to the discussion.

Best regards,


More information about the time-nuts mailing list