[sdiy] Dirty Ground & Bypassing

[Larry]

Magnus

You mention bypass caps.  I've been out of it for a while, but I do 
remember from the good 'ol S-100 days that bypass caps were critical, and 
the preferred were tantalums.  But doesn't the frequency of the noise (i.e. 
clock speed) determine what cap values are best to use?  And, pray tell, 
what is the proper formula to pick a cap that will look like a short for a 
particular frequency?

You also mentioned good books on the matter.  Can you recommend some that 
are not TOO technical for my mathematically challenged mind?

Larry Troth

At 02:07 AM 9/27/01 +0200, Magnus Danielson wrote:
>From: "Jay Schwichtenberg" <schwich at qwest.net>
>Subject: RE: [sdiy] Dirty Ground & Bypassing
>Date: Tue, 25 Sep 2001 08:37:40 -0700
>
> > Antti,
>
>Antti, Jay,
>
> > It sounds like you have the right idea on doing the star ground. But!
> > Anything hooked on the digital side of things should go to the digital
> > ground.
>
>Indeed. Trouble arises as soon as digital meets analog.
>
>To which ground should one hook up?
>
>Whenever you change state on any of those digital lines, you will push
>current on that line but pull the same amount from either the ground
>or Vcc lines. To keep emissions low for instance, one has to ensure
>that the full loop of the current is small. If you have a signal
>passing over a ground floor (say digital) and then jump over to
>another ground floor (say analog), then the return current will pass
>over the most suitable point, and this is most probably quite far
>away. The bigger loop means bigger inductance, and this means more
>ground noise most of the time.
>
>One of the tricks being used is to build a narrow bridge between the
>grounds and one point. Over this bridge is the signal passed in the
>narrow path. They will "feel" a continous ground, but the quite or
>analogue ground will not feel the passage of many other stray
>currents.
>
>Another trick is to put small caps near the signal so that the
>returnpath is through the caps.
>
>All this assumes that one has done the propper bypass cap work. Here
>one should really spend some though to snap at the peaks in the right
>way.
>
>Inductance is really your enemy here. Well, most of the time at least.
>
>There are many good books to read on these matters.
>
>Cheers,
>Magnus