[sdiy] Shielding -

[Magnus Danielson]

From: Glen <mclilith at charter.net>
Subject: [sdiy] Shielding - (Was: 100 MHz EMI, what can it be?)
Date: Mon, 14 Apr 2003 14:41:29 -0400

> At 10:29 AM 4/14/03 , Grant Richter wrote:
> 
> >In one interview with Wendy Carlos, she talked about building her studio
> >with aluminum backed drywall, and and connecting and grounding all of 
> >them to form a Faraday shield around the entire studio.
> 
> <SNIP>
> 
> >Of course, this is electrostatic shielding, magnetic shielding would be much
> >more expensive.
> 
> What options are available, if you want to provide magnetic shielding as
> well as electrostatic shielding?

Actually, what people have missed is that you dampen of magnetic fields with
metalic sheets too. The field will induce a current in the metal which will
cause a inverse field and the vector sum will decrease. Since the resistance
in the metal is more than zero, there is losses, so naturally will the effect
not be total, and due to the laws of physics it will be frequency dependent.

Well, you can get perfect cancelation if you use a super-conductor, but not for
arbitrary strengths of the field since the superconductor saturates in current
density which you do not want to do. I've personally use the reverse-induction
field in the form of the Meisner-effect to keep the superconductor to "fly" over
a magnet. Cool stuff. It was a YBaCuO-superconductor cooled with liquid
nitrogene.

<SNAP BACK>

Oh... those memories...

Anyway, the damping becomes lower with lower frequency, so it shields for
higher frequencies. In coax cables it is a combination of the shield directly
and symmetry and cancelation which causes the full effect. For shielded twisted
pair will the twisted pair become more effective with lower frequencies, so a
properly shielded twisted cable with proper diffrential drive and receiver is
actually a pretty good system.

A shielded room works well for most of the noise we receive. 50 Hz and the like
isn't well shielded, but FM radios, etc. should be really well blocked.

However, there are simple means to make sure that RF doesn't enter into devices.
You can do minor wonders with just capacitors and some planning.

> If the room is underground, would that help much?

Yes, some. The ground around the house will also attenuate signals in a similar
sense. However, it varies with the weather... dry ground probably isn't as good
as wet ground.

> I think it's a bit odd that electrostatic shielding of the room would help
> so much, if the equipment inside the room is built with good shielding to
> begin with. Maybe that was the problem--the gear inside the room wasn't
> really designed with excellent shielding?

Hey! Where talking audio-devices here... I *rarely* see anything which looks
like serious RF-protection measures in audio gear. I think building shielded
rooms for studios is greatly overdoing it. If equipment is sensitive to it, why
not fix it? All EMI/EMC protection done after a device have been designed always
becomes expensive. Doing a shielded room is overdoing it. For a DIY-er you can
do alot by fairly simple means if you want to.

* IEC-outlet integrated with a line filter with a real shield around it.

* Metalic case with good conduction between all surfaces and protective ground.
  Especially along the edges should conduction be good along the full edge.

* Ceramic caps to shield/case for all ingoing and outgoing electrical signals.
  This shorts common mode signal into shield.

* A cap between hot (+) and cold (-) of diffrential signals.
  This shorts diffrential mode RF signals.

* A series shunt-inductor will also help for signals. For diffrential signals
  a double-inductor (really a 1:1 transformer around a circular core turned
  90 degrees wrongly... the polarity mark of both coils in the same direction)
  is the best choice. You usually see clamps on monitor cables, PSU cables of
  portables etc., they are exactly this - put applied on the wrong side of the
  contact IMHO.

* For these caps, keep absolutely minimal lead-distance since they are to handle
  RF and not audio, so any series-induction will reduce the effect. I would
  start of with a 100 pF cap. Large caps actually make things worse most of the
  time contruary to popular beleif (the exception is in resonant curcuits where
  you probably want to increase the loss of that resonance to get started).

Some of these measures can be retrofitted into existing devices. However like
always, doing it after a box is designed makes it more expensive than
integrating it into the design.

A word of warning thought. Do not apply the caps to shield on the PCB end of
cables from the contact, then you inserted too much induction for them to be
effective, right at the contact is where they should be. If you contacts is
PCB mounter, just toss the caps on the solder-side, and trim down the leads to
become minimal - big loops isn't doing it the right way.

Cheers,
Magnus - with a history in doing EMC stuff...