David,
Telecom ("Bellcore") usage is to keep signal, surge and power currents off
chassis and safety grounds. This is understandable. It is due not only to
audio sensitivity, but to the need to protect equipment from substantial
peak (hundred of amps) surges at fairly high (thousands of volts) peak
voltage.
Practically speaking, if you isolate a ground, you must isolate all the
signal and return conductors referenced to it as well. It requires good
balance in signal and return conductors, and measures to sure that balance
is not disturbed. This takes meticulous design, and some expense in
manufacturing. I have seen this done with signals in the hundreds of MHz,
but care had to be taken in layout, connector and cable selection and
construction.
Surge protector current must be routed away from signal return current. If
the chassis of a device is connected to the surge return path, a surge on
metallic conductors will be present on the chassis of the protected device.
This may be of little consequence to the device, where all grounds and Vcc
rise to the surge potential at once. But if it connects to another
equipment whose chassis ground is different, a sizable potential can appear
between signal conductors from the surged device and client equipment not
sharing its ground. That is not a good thing!
At sufficiently high frequencies there is no "ground" at all, really, only
return paths of various shapes, lengths and impedance. We can then make the
argument that - as long as we don't have to deal with surge currents -- our
chassis and digital grounds be tied together. This simplifies shielding.
But we still have to deal with surge somehow, and now, I suggest, we can
say that the SURGE return should be isolated, not the digital ground.
Higher frequency ESD transients are more problematic. Grounding
circuit-pack faceplates only at the bottom is NOT helpful! Connecting them
to the board ground is BAD. Before being laid off, I was slowly converting
people to the idea that ESD currents need to be treated as UHF radio
energy, and shields constructed accordingly; faceplates with 360 degree
grounds to the chassis. In some cases, it is enough to provide paths to the
chassis before such currents can flow onto the board, but induced fields
are often troublesome. For sure, ESD must NOT be allowed into circuit pack
Vcc and ground. An ESD trace or guard band is often used, but this can be
unnecessary -- why parallel a metal panel with a trace? -- or insufficient;
a guard band NOT connected to chassis, as when a card is being inserted,
may actually couple an ESD event to places where it can do harm. Anyway,
when it IS inserted, you can't reach it, so why have it? And a trace
leading ESD to the backplane puts it exactly -- among all the nice clean
signals -- where you do NOT want it! (I've also seen an encircling trace
resonate and cause EMC problems.)
My approach is very simple: Where will current go, and what will it do?
If you use this approach, you will find others coming over to your point of
view, especially when you are once in a while spectacularly right.
Cortland
-------------------------------------------
This message is from the IEEE EMC Society Product Safety
Technical Committee emc-pstc discussion list.
Visit our web site at: http://www.ewh.ieee.org/soc/emcs/pstc/
To cancel your subscription, send mail to:
[email protected]
with the single line:
unsubscribe emc-pstc
For help, send mail to the list administrators:
Ron Pickard: [email protected]
Dave Heald: [email protected]
For policy questions, send mail to:
Richard Nute: [email protected]
Jim Bacher: [email protected]
All emc-pstc postings are archived and searchable on the web at:
http://ieeepstc.mindcruiser.com/
Click on "browse" and then "emc-pstc mailing list"
