[time-nuts] Clocking a PIC16F628A from a Rubidium Standard

[Hal Murray]

> One thing I did notice is that the oscillator seemed to power the chip!  If
> I were doing it again I'd probably use some kind of powered buffer on the
> oscillator input so that when I power down the circuit it actually stops.

That's a reasonably common problem.  Occasionally it's a feature.

Most digital inputs have protection diodes from the input pin to VCC and GND. 
 They come for free in most semiconductor technologies.  They damp out 
glitches/reflections.

If VCC is open and the circuit isn't drawing a lot of power, you can get 
enough power from an input signal through the diode to power the whole 
circuit.  It can be nasty to debug if you have something like VCC open to a 
single chip.  The outputs of that chip will be a diode drop lower than the 
rest of the circuit.

I thought about setting up a PIC to make 32 KHz from 10 MHz using power from 
the 10 MHz input signal but never got around to actually building it.

----------

It's tricky to safely connect to external signals that may be active when the 
local power is off.

I'm not sure a powered buffer will solve the problem, at least without some 
more info about what type of buffer you plan to use.

In the simple case, you can just insert a big-enough resistor.  It needs to 
be big enough to limit the input current yet small enough not to distort the 
input signal too much.

There are some families of chips made without the protection diodes.  The 
usual application is letting chips running on 3.3V listen to signals from 
chips running on 5V.  (I don't have part numbers handy.)



-- 
These are my opinions, not necessarily my employer's.  I hate spam.