[time-nuts] Hints on PPS Buffer design...

Clay Autery cautery at montac.com
Sat Jun 17 09:14:20 EDT 2017


On 6/17/2017 4:01 AM, Hal Murray wrote:
>> Q3: It's only a 1Hz frequency, but is low inductance a desired trait of the
>> chosen resistors? 
> It's a 1 Hz repetition rate, but the bandwidth depends upon the rise time.  
> If the rise time is ballpark of 1 ns, the bandwidth will be ballpark of 1 
> GHz.  So, yes, you want low inductance.  That includes the power to the chip 
> as well as the resistors.  Surface mount is your friend.  So are ground/power 
> planes.
>
> How good is your scope?
Not nearly good enough, as I don't have one.  Have to borrow.  Thank you
for pointing me in the right direction on bandwidth...


>> Q:  Why does everyone pick FIVE x 100 Ohm resistors?  That's 20 Ohm out, not
>> counting the gate impedance on the hex inverter... 
> The FIVE is 6-1.  The one is for isolation.  The 5 is the rest of the 
> package. You might as well use them all as drivers.  You don't want to use 
> them for another signal (even if it is supposed to be identical) or you will 
> get minor crosstalk when you do things like plug or unplug a cable.
Yes, I got this part... At least I generally get something.  ;-)
> I don't know why you are saying "gate impedance".  That's over on the input 
> side.  I would have said "output impedance" or "driver impedance" of the chip 
> or section.
Apologize, was parroting the term used in the archived Time Nuts post I
found referencing the need to address the "driver impedance" in
selecting series resistors when trying to set the output impedance.

> I'm not sure why they picked 100 ohms.  Assume the net source impedance is 25 
> ohms.  Suppose the far end is terminated with 50 ohms.  There won't be any 
> reflections so the source value doesn't matter.  25 ohms will provide a 
> higher voltage at the far end than 50.  If you have a CMOS driver and a CMOS 
> receiver, 1/2 the voltage at the receiver is nasty.  It might be OK if you 
> have HT type receivers.
OK... I'll have to try and study this some more, because for a simpleton
like me, and impedance mismatch between the PPS out port on the Nortel
and the device to which it is connected of 25:50 ohms, reduces to 1:2,
flipped around is a 2:1 VSWR, which means a fairly significant power
loss...  Maybe it doesn't matter much with a short cable.. but the cable
is also tiny/lossy.

> I would suggest a bit of lab work.  What are you going to use on the far end?
> Lots of gear has 1000 ohms rather than 50 so a 50 ohm source impedance takes 
> care of the reflections and leaves (almost) the full voltage at the receiver.
Yes, I intend to do extensive testing.  I was just trying to trim away
some of the waste up front... time, materials, et al.  :-)
>> thus I can only include that I need to use something slightly more than 250
>> Ohms on a 5 gate parallel setup)
> More than 250 divided by 5 and rounded up a bit for the output impedance of 
> the chip will be more than 50 ohms.  I'd do some experiments.

Yep... gonna need some more instruments asap...  I may just have to wing
it and accept "good enough" until I can round up the rest of the
required instrumentation.

Thanks for your assistance and not brutalizing me for my lack of
knowledge.  I'm trying to learn.

 - C. Autery


More information about the time-nuts mailing list