Grounding (Was Power supply musings)

[Martin Czech]

> I have attached a small diagram just to confirm I am clear on some terms, sorry for the
> poor quality.
> 
> If I understand correctly, the consensus is to connect the mains ground to the chassis
> ground only, and to connect the jack grounds to the power supply ground only?

I think that this is *the* correct setup, at least for +-15V supply.
It may be neccesary to have an hf path to chassis ground via high
voltage cap.


Of course only with a safe two chamber isolated transformer.  The
transformer iron as well as shield windings should get some protective
earth , too.  J.H.'s proposal of parallel diode clamping would allow
for chassis ground= signal ground, if you're not so sure with your trafo.

As stated in "The Art of Electronics": Use a IEC power jack. These come
with fuse and hf filters.  This also ensures proper
mechanical grounding, no matter how strong you pull at the cord.
Some money, but saves headache!

I recently burned a motor servo control to death, this motor was used to
move the "gouilliotine" door (light proof IC probe chamber door).  It was
a self construction from our lab tek, one cap was bad and it exploded.
The guy had ommited a fuse in the apparatus, the bad thing was that the
lab magnetic fuse automat trips @ 16A and the servo supply burned with
only a few amps, a little nice fire in the lab!!

So , necer ommit a primary slow fuse, about 1.5x - 2x the maximum expected 
primary current. DC supplys with caps lead to transient charging currents,
the fuse should be able to deal with that. Also, larger supplys draw a lot of
current when switched on. This is due to the large cap charging current and
there is no magnetic flux in the core, especially toroid trafos are a good
short when starting up.

For a live show I once had borrowed a big amp, but couldn't switch it on,
since the fuse automat did always trip. Took me quite a while...

Better is to use some kind of starter circuit, a series R limiting the
current to safe values and a relay to short this R after a few seconds.
This circuit should drop imediately if the power is switched off,
because otherwise a on-off-on sequence will blow the fuse again!
I have used a little extra 12V trafo (3 EURO) for the relay,
with rectifier and cap. A resistor throtels the relay, so it needs some 
time to close the relay. Of course, in case of undervoltage the relay might
never close thus burning the limiting resistor of 100Ohm 1Watt...

I think there was a better circuit for the last Elektor Gigant amp,
2000 Watts. It is not so easy to make a good startup circuit, because
at the beginning you start with nothing....



And I learned to have a snubber network in parallel to the trafo primary,
say 100 Ohm in series with high voltage C of 100nf (?) or so. This
prevents switching arcing.

Another good idea is to have a transient suppresion after the fuse,
some varistor that will kill those evil 1kV line transients.


> 
> Two pieces of equipment wired this way may be connected together via a patch cable
> safely and without noise?

Under normal conditions: yes.  But if you have severe earth currents
(unknown earth, building to building connection) or severe hf dirt the
only way to go is balanced, in extreme cases trafo balanced.

> 
> What if the two pieces of equipment used Serge style banana patch cables which have no
> shield ground?  It seems there would be no common  ground reference between the two.
> Does this matter?
> 
Yes, I would say two floating devices with no common ground should give hum load as hel!

m.c.