[time-nuts] PPS sync

[Hal Murray]

jerry at hanler.com said:
> It’s interesting how it jumps around from PPS to PPS.  

You can work out what the offset will look like.

Assume your clock is roughly 10 MHz.  So you divide by 10,000,000 to get a 
PPS.  If your clock is 2.001 HZ fast, then you need to divide by 10,000,002.  
That gets close, but it will drift by 0.001 HZ or 1 ms per second.  So every 
1000 seconds you will have to skip an extra cycle.

If you plot the offset, you get a ramp going down.  If your clock is 
10,000,001.999 you get a ramp going up.  (I hope I got the sign right.)

It gets more complicated if your clock is not near an integral number of 
cycles/second.  Suppose it is 10,000,002.501.  Now you have to skip an extra 
cycle every other second.  You get 2 lines offset by a half-second.

That all assumes you have a good GPS antenna and such so that there isn't 
much noise/jitter on the calculated time.

If the temperature on your crystal is changing, you can get hanging bridges.

Neat graphs here:
  http://www.leapsecond.com/pages/m12/sawtooth.htm


-- 
These are my opinions.  I hate spam.