[time-nuts] HP Z3801A info needed - 1PPS interface

[ed breya]

Thank you Francis, for the info on the 1PPS interface. I like that TI 
65ELT23 pECL to TTL converter - it looks ideal for the situation. For 
now though, I'll be going with the LM311, then upgrade to the better 
solution next time I do another round of wrapping things up.

I studied the Z3801A manual some more, and I now see that the 1PPS 
signal should be easier to deal with than I thought. When I started this 
part, I was concerned about the power demand of the 1PPS signal load. I 
had planned on handling any duty factor, but I see the pulse width 
during "on" (+5 V output) is 50 uSec maximum, so the DF and average 
power are quite small. I made it to be able to deliver 100 mA 
continuously, but now I can scale down the PS. This did turn out to be 
handy though, since it allowed for DC testing of the driver section. It 
works beautifully with the 74HCT541.

So, the next step will be to cut the shunt regulator demand from well 
over 100 mA, to something more reasonable like 20 mA, to run the 
circuit. This section is almost ready to go. I'll put in the LM311 
circuit and wrap up a few other things.

The plan is to bring the 1PPS PECL signals in from the DB-25 connector 
via either shielded twisted pair, or twin coax lines, with the shield 
grounded at one end only (don't want an extra internal ground loop).

One question I have, is if anyone (especially Francis, if you're 
listening - you probably know the answer), happens to know how the PECL 
outputs are terminated or sourced inside. I assume the idea is to be 
around 50 ohms, or 100 ohms differential, for driving cables. I'm not 
too concerned about exact matching, since it's less than a two foot 
trip, but just wondering if the outputs are intended to be terminated at 
the load end, or are reverse terminated inside, and don't need it. My 
plan is to have some DC load at the comparator inputs, like maybe a few 
k ohms, and asymmetrically DC biased to force the "off" state (0 V out 
from the 1PPS BNC) if the PECL signals are disconnected. When they are 
connected, they will override the small offset currents and run normally.

Ed