MIKE ROSS wrote in a message to Roy J. Tellason:

 RJT> MIKE ROSS wrote in a message to ROBERT SAYRE:

 MR> The problem is that the capacitance of a diode increases as its
 MR> reverse bias voltage becomes smaller.

 RJT> Why is diode capacitance a problem?

 MR> Because of the Miller multiplication effect in for example the
 MR> basic common emitter transistor configuration. Where a portion of
 MR> the output is fed back through the collector-base capacitor to the
 MR> input. That current is a direct function of stage voltage gain.
 MR> Thus an inverting stage with a capacitance of 5pF in the feedback
 MR> path and a gain of 100 (re: Av+1) will in effect see a virtual
 MR> capacitance of 500pF at the input due to the Miller multiplication
 MR> effect. This virtual capacitor in parallel with the input will slow
 MR> down a pulse edge as it charges through a non-zero source 
 MR> resistance. (The Miller effect isn't exclusive to transistors, any
 MR> inverting voltage amplifier will have it.)

I've heard the name before,  but forgot about it because that sort of thing
doesn't usually enter into what I'm dealing with.  I wonder if that's
perhaps responsible for the otherwise odd design of TTL inputs?  I'll bet
that characteristic makes it harder to design RF amplifier stages,  too, 
something I've never really had much of a grasp of.

But that being the case with amplifier stages,  why is it a problem for diodes?

 MR> The diode capacitance actually becomes a maximum just as it starts
 MR> to conduct.

 RJT> Just before,  actually.  The junction still needs to be slightly
 RJT> reverse-biased.  And the minimum reverse-bias gives the same effect as
 RJT> having those capacitor plates *real* close together.

 MR> So how do you suppose the capacitance varies between zero and the
 MR> threshold of conduction at around 0.5 volts for a Si diode? (which
 MR> is fully conducting at say 0.7 volts, ie what happens between 0 and
 MR> 0.5v?) 

Good question.  I'd have to review some theory on that,  and don't even
know if I have the books where I read about this stuff in the first place. 
(Probably the "GE Transistor Manual".)  It's apparent to me that
with some reverse bias,  you can have charge carriers "pushed"
(by the bias potential) away from the junction,  thereby giving the effect
of an insulating layer on either side of it.  I'm not real clear at the
moment as to what sort of effects come into play when you start to forward
bias a junction,  though.  It's been *years* since I reviewed any of this
stuff...

 MR> The choice of Schottky diodes is because they have a low forward
 MR> voltage and can be made very tiny with very little capacitance.

 RJT> Tiny in terms of the area of the junction?  Or something else?

 MR> Yes, because the metal to oxide junction can be made as a parasitic
 MR> diode of the point-contact type. This type of metal-semiconductor
 MR> point contact diode has very little capacitance because it is
 MR> basically a single capacitor plate with a needle tip real close and
 MR> the area under the needle tip can be very tiny. OTOH as you 
 MR> probably know a diffused junction has a much larger area by 
 MR> comparision.

Makes sense.

I remember when articles about stuff used to go into all sorts of aspects
of how semiconductors were fabricated,  both discrete devices and chips. 
And back then I used to really wonder what the heck difference it made to
most of us,  though I can see in retrospect that there are a number of
characteristics of the devices that will vary,  based on fabrication
techniques.

---