[time-nuts] Driving clocks from 1pps

[Chuck Harris]

Murray Greenman wrote:
> In the light of the latest posts on driving clocks from 1pps, it sounds
> as though I'd better rethink what I was planning!
> 
> I am in the middle of the design of a micro which uses a 10MHz crystal
> to provide a digital clock, but the time is kept in line via GPS using a
> 1pps NCO, which is steered digitally, rather than altering the 10MHz
> oscillator in GPSDO fashion.
> 
> The plan was to provide two outputs (biphase) at 1pps to drive slave
> clocks, but in the light of the notes from Brooke and Chuck, I would be
> better off just providing a single output, and use a series cacacitor.
> 
> Chuck, I would expect that the 1pps would need to be about 50% duty
> cycle, or at least have a pulse width of 100ms or so. I can imagine a
> clock driven from 1pps with a low duty cycle would sound quite
> different.

Yes, for a simple circuit like that, you need a 50% 1/2pps duty cycle,
otherwise the clock will go: tictoc...................tictoc...............tictoc.....

The simple ubiquitous quartz clock that runs on a single AA cell has one
most important requirement and that is to keep the coil current off except
when switching seconds.  The motor has a permanent magnet, so it is a
latching stepper motor.

And the actual waveform used to drive the clock's coil is 1/2 Hz, not 1pps
as I showed.

If you have 1pps, you will need to use a toggle FF to cut the frequency in
1/2 to 1/2pps with a 50 percent duty cycle.

If you want to do something like divide down from 10MHz, most any modern
PIC can be used to make a one chip, zero extra part solution, as was
described by TVB earlier.

-Chuck Harris