[time-nuts] Why are 1PPS signals so skinny?

[SAIDJACK at aol.com]

Not really, your setup requires all inputs except the very last one to  be 
high-impedance to work, and to have a trigger point of 1.25V as well to work 
 properly (when used with a proper 50 Ohms source). So no difference there.
 
So it doesn't make any difference, since the same exact inputs will work  
just as well with the open-ended circuit if the trigger point is set to 1.25V 
on  those inputs. Same requirement for both setups.
 
Daisy-chaining is a bad idea because you are getting the propagation delay  
between the different hubs as mentioned. Using a simple T with open-ended 
cables  would make all inputs switch simultaneously if the cable lengths 
after the T are  the same.
 
Daisy chaining also creates stubs along the way due to the capacitive  
loading on the cable at the inputs, and the small amount of additional wiring at 
 the stub, and these stubs will cause reflections going wild due to 
impedance  mismatch at the stub, and run amok between stubs and between the ends of 
the  cable.

BTW: when setting the threshold to 2.5V and tapping-off somewhere in  the 
cable, this is called reflected wave switching, as opposed to incident wave  
switching, which is what happens when you set the threshold to 1.25V.
 
Also, the Thunderbolt has less than 5 Ohms output impedance, so you  get a 
reflection going back from the 50 Ohms end-termination anyway because the  
impedance is mismatched!
 
The signal won't stop at 2.5V, it will go all the way up to 5V  in  static 
conditions just the same as in my scenario. This can be seen in the plots  
that Didier had sent earlier.
 
But worst of all: if your 50 Ohms end-termination falls off, or goes away  
because you turn-off that piece of equipment providing that termination, 
then  almost sudden all of your inputs can see 10 Volts on the line, and could 
blow up  due to overload. Having a 1.25V threshold input seeing a 10V signal 
is not a  good idea..
 
You get all the drawbacks and more, and no real advantage. At least when  
connecting to e.g. a Thunderbolt output that has << 50 Ohms series  impedance.
 
bye,
Said
 
 
In a message dated 5/15/2012 13:44:42 Pacific Daylight Time,  
dave.martindale at gmail.com writes:

It is  worth noting that skipping the end termination is probably a bad idea
when  daisy-chaining a signal from one output to more than one device input.
The  input at the end of the cable will see a clean rise from zero to 5 V
(or  whatever the driver's open-circuit voltage is), but the other inputs
along  the length of the cable will not.  They will see an initial rise  
from
0 to 2.5 V as the series termination at the driver and the cable  impedance
act as a voltage divider while the cable is being charged. Later,  they will
see another step change from 2.5 V to 5 V as the reflection  returns from
the open-circuit far end of the cable.  If the input  threshold is
automatically set at half the input voltage swing, the input  could trigger
on the outbound or the reflected pulse, or even somewhere in  between.

This is in contrast to having a 50 ohm termination at the end  of the cable
(plus the 50 ohm series termination at the source), where all  inputs along
the length of the cable see a single edge transition from 0 to  2.5 V.  They
will each see the edge at a different time due to  propagation delay, but
all will see a clean edge.

Dave