[time-nuts] HP/Symmetricom 58535A and 58536A

Magnus Danielson magnus at rubidium.dyndns.org
Wed Nov 4 00:09:26 UTC 2009


Hi Ignacio!

> It reminds me when I was breadboarding a TVB divider for my Rb clock 
> project.  Once I removed the power to the PIC, but the 10 pps led 
> continued blinking at a very low intensity.  After minutes of  
> stupefaction and a search for a charged capacitor or other power source, 
> I realized that the beast continued working powered by the 10 Mhz signal 
> !!!

This is a classic CMOS "feature" as the input protection diodes will 
provide rectifiers to the supply - oriented in the propper directions.

The interesting thing is that devices can behave as almost properly 
functioning, except for rare occasions of not working, which is when 
none of the inputs provide the Vcc or GND reference needed.

Also, we had this problem with our early hot-swap boards. I had to teach 
our hardware guides about protection circuits that hooked the inputs to 
the powerplane and all the caps there, producing a high inrush current 
unless the drivers where properly isolated. Once the protection was 
designed in we saw no more of those errors. Have not heard about that 
kind of problem in the last 10 years now. Power-sequencing in the 
connectors is there for a reason in some hot-swap systems.

So, all connections to a circuit is potential power-supplies. Sometimes 
it has amusing effects and sometimes it has terrible effects.

Cheers,
Magnus



More information about the time-nuts mailing list