From: "Christopher R. Carlen"
Subject: Re: Ground loops?
Date: Fri, 13 Sep 2002 14:25:22 -0700
Organization: Sandia National Laboratories, Albuquerque, NM USA
NNTP-Posting-Date: Fri, 13 Sep 2002 16:44:44 +0000 (UTC)
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.0) Gecko/20020529
X-Accept-Language: en-us, en
> Say I had three printed circuit boards. Each board requires a +5V supply and
> ground, and in addition the boards are inter-connected via ribbon cables
> using separate cables. Let's say the environment is a little noisy from an
> electrical point of view, so the ribbon cables use the every second
> conductor is grounded type construction.
> First, what is the best way to connect power? My instinctive reation to this
> is star ground where each card's ground goes to a common point. Similarly
> for the +5V.
> Second, should the grounds on the ribbon cable connectors be grounded on
> both ends? Again my reaction is grounding at both ends creates ground loops
> that could pick up noise from the environment. My feeling is that only one
> end of the connectors chould be grounded and the other ends left floating.
> Any views?
Ground at both ends. That way, the return currents flow adjacent to the
conductors. That way you both radiate less noise and also, the loop
area that picks up induced noise from ambient EM fields is minimized.
If you have conductors carrying signal and the grounds are connected at
only one end, this may help reduce crosstalk some (grounding both ends
should be better here too), but the return currents have to flow
somewhere, that means somewhere you have connected the grounds of both
boards together. If this is on a different cable than the ribbon that
we are talking about, then your ground wire and signal wires form a much
larger area to intersect with ambient fields.
See what I mean?
This is one aspect of the problem, that of ambient noise field pickup.
Another issue is the sort of "ground loop" formed by large currents
flowing through ground connections, causing a voltage drop across the
ground bus. Then circuits at one end of the bus incur offset errors
relative to the other end of the bus, or AC components dropping across
the ground bus can get injected into low level circuits. It is for this
issue that the star grounding topology might be appropriate.
Are you strictly digital here or analog, or mixed?
Christopher R. Carlen
Principal Laser/Optical Technologist
Sandia National Laboratories CA USA