[sdiy] Tri-state outputs and CMOS inputs.

[Florian Teply]

Am Freitag 20 Februar 2009 16:56:22 schrieb Dave Kendall:
> Hi all.
>
> I can't find a definite answer anywhere to this one - Do CMOS inputs
> need pull-up/down resistors if connected directly to the outputs of
> ,say, a 3-state latch - which is not connected to anything else)
> In my experiments, it *seems* to work OK, but there are too many
> variables in the rest of the circuit to be sure....
>
> My gut feeling is that CMOS inputs *should* need resistors, but it'd be
> good to know for definite.
>
Short answer: it depends... ;-)

Longer story: as long as the 3-state output connected to the cmos inputs never 
goes to its third state (Hi-Z), everything is fine as it is, just with the 
added expense for the unneeded 3-state output.
But if the output enters hi-Z mode, the input is quasi-floating, so virtually 
everything could happen: Input acting as if a logic high was applied, acting 
as if logic low was applied (maybe even dependent on the level seen before 
output went to hi-Z, same-as-before and opposite).
Basically, the CMOS input would be self-biasing to somewhat around half its 
supply voltage, for which behaviour is undefined. Worst case would be highly 
exothermic destruction as both input MOSFETs would be conducting and thus 
shorting your supply.

HTH,
Florian