[time-nuts] Question about an HP5370B TIC

John Miles jmiles at pop.net
Fri Sep 30 07:56:01 UTC 2011

> > If I put the 5370B in the +/-TI mode I expected to see both + and -
> > time differences when the two signals are closely aligned - and  mainly
> > do.  Typically the difference is + or - a usec or less.  Occasionally
> > however the 5370B displays very near to a +1 or -1 second  (999.999 xxx
> > or -1.000 000 xxx s for example) reading.  Needless to  say this plays
> > with trying to measure a mean value of the  alignment.  The 5370B is
> > measuring the "outside" between the  pulse pair when I want it measure
> > "inside" (or short time interval)  between the pulse pair.
> How are you collecting the data?  If each sample gets to a PC, you can fix
> up in software.
>     If X > 0.5: X = X -1.0
>     If X < -0.5: X = X+1.0
> One disadvantage of that is that those samples eat 2 pairs of pulses so
> are only getting half as as many samples per hour.
> I like John's suggestion of triggering one side off the other edge of the
> pulse.
> Is the pulse from the OCXO supposed to be synchronized with the GPS
> If not, just zap the counter to move the pulse around within the one
> window.  (I'm thinking of shorting out one of the counter bits with a
> screwdriver or clip lead for a fraction of a second.)
> My first thought was to insert a delay line, but that will take 1000 feet
> coax.  Maybe more.  I don't have that handy.  Frys says $80 or $90 for 500
> feet.   I'm too cheap to buy it just for one test like this.  (But I might
> use this as an excuse if I see some at a good price.)

The easy way to 'cheat' may be to lie to your GPS clock's control software
and tell it you have several hundred feet of antenna cable.  Most commercial
clocks will try to maintain accurate time vis-a-vis the antenna site, so
they can delay the pulse to compensate.

Or you can just let it wrap, and compensate in software...

-- john, KE5FX

More information about the time-nuts mailing list