[sdiy] Analog & Digital Ground

[Harry Bissell Jr]

Gaarrrr Ian, you should know better than that :^P

"what we have here is a failure to communicate"

I understood the beads would be TWO beads, one in the
power and one in the ground, (like Fig 15) and I
disagreed with that.

I do agree with Fig 16, if they were talking about
both traces through the same bead.

I was thinking of the beads that have a single
conductor through them, like MOTM uses on their power
supplies.

I don't agree there should be a bead between the
analog
and digital grounds on the same board.

If the example you're thinking of is Fig 21... this
would be good for independant circuits in the same
rack but i think it would be more trouble when you
start to interconnect them.  A signal originating from
one board has to reference through two beads (extra
inductance) before returning to the source of the
current. I think it will increase the signal
distortion
(but that remains to be seen)

I'm not even sure I'd use the ferrite beads at all. I
usually don't. If the issue is to reduce radiated
EMI from the PIC, maybe they are needed to meet agency
approval. Since as a rule, I do not let the FCC enter
my home... moot point for me.

I think that careful PCB layout and routing of traces
to maximize capacitance and minimize inductance
(parallel power and ground traces) and proper
decoupling will do much more than the beads...

H^) harry



--- Ian Fritz <ijfritz at earthlink.net> wrote:

> I'd go by what the bead experts say.  This is a
> *really* good document:
> 
>
<http://www.steward.com/Steward/pdfs/emi/technical/Use%20of%20Ferrites%20in%20EMI.pdf#search='ferrites%20in%20EMI'>
> 
> (Sorry, Harry)
> 
>    Ian
> 
> 
> At 06:15 AM 4/11/06, Seb Francis wrote:
> >Thank you Ken and Harry for your replies. 
> Unfortunately they both say 
> >different things :-)
> >
> >Ken says yes isolate the digital parts via beads
> (including on the ground).
> >Harry says don't isolate the ground, only the power
> rail supply to the 
> >regulator for the digital parts.
> >
> >So, what to do ...
> >
> >As Harry points out, it's important not to share
> ground traces between the 
> >analog and digital stuff, but even if there are
> separate traces back to 
> >the PSU part of the board, is it not possible for
> high frequency glitches 
> >to be introduced from the digital ground into the
> analog ground?  After 
> >all, there are lots of low impedance power sources
> (0.1uF decoupling caps) 
> >all over the analog circuitry, so some current
> might flow from these as 
> >well as from the PSU components.
> >
> >I imagined that by putting a bead in between the 2
> grounds and having 
> >separate decoupling caps for the digital parts, the
> high frequency digital 
> >edge spikes would be mostly blocked from the analog
> parts of the board.
> >
> >I take Harry's point that you do not want
> inductance on the ground line, 
> >but I thought maybe a very small amount of
> inductance can be a good thing 
> >for isolation of very high frequencies.
> >
> >Any comments from you electronics gurus out there
> :) ... ??
> >
> >Seb
> >
> >
> >
> >harrybissell wrote:
> >>Hi Seb...
> >>
> >>I would not put a bead in the ground line. You do
> not want inductance
> >>there (imho). Better to just separate the ground
> traces so the PIC and
> >>opto
> >>current cannot cause voltage drops on the analog
> circuits.
> >>
> >>The bead would be a good idea in the positive
> supplt, where it will
> >>decouple
> >>the two supplies.
> >>
> >>H^) harry
> >>
> >>Seb Francis wrote:
> >>
> >>>Hi,
> >>>
> >>>I'm just laying out quite a large board which is
> almost all analog,
> >>>except for a PIC&opto for the MIDI interface.
> >>>
> >>>Normally I have totally separate analog and
> digital grounds which are
> >>>joined right the way back at the PSU (at the end
> of the power lead), but
> >>>in this case the PSU is on the same board.
> >>>
> >>>I am thinking that I should separate the PIC&opto
> ground from everything
> >>>else with a ferrite bead.  The power is already
> separate (it has its own
> >>>5V regulator).
> >>>
> >>>Does this sound sensible?  Also, should I hook
> the 5V regulator to this
> >>>digital ground as well?  And perhaps even connect
> the power input to the
> >>>5V regulator via another ferrite bead?
> >>>
> >>>Thanks,
> >>>Seb
> >>>
> >>
> >>
> >>
> 
>