[sdiy] Mains interference

[rburnett at richieburnett.co.uk]

> Regarding silicon junctions shifting RF into the audio band. Dirty
> contacts can do that too...

Which is why I advised Tony to check that his earth bonding in the house 
was up to scratch.  It's good advice to check for poor or corroded audio 
interconnects too though :-)  A poor screen connection at one end of an 
audio cable vastly increases susceptibility to RF interference.

> Would an overdose of ferrites and ferrite sleeves be a good general
> advice for studio connections and circuit design?

Ferrite common-mode chokes just outside the metal enclosure work wonders 
for quick-fixing because they stop nasties travelling down the screen of 
the cable from getting inside the unit.  You often see them on things 
like VGA/DVI cables and USB cables.

There's lots of other things you can do at the design stage.  For 
instance chassis grounded connectors are better than PCB mounted ones 
because any RF current flowing on the screen will be conducted through 
the metal case rather than flowing through wires and across PCB traces 
where it can wreak havoc with your circuit.  Where you don't want to 
connect the shield of an audio connector to the metal case for hum-loop 
reasons, you can use a low-valued ceramic cap so that it is connected at 
MHz frequencies but still open at 50/60Hz.  Ground-planes on the PCB are 
another good idea as they conduct RF currents much better than thin 
ground traces.  Little surface mount ferrites on the PCB can be 
effective if they are right up against where the connection comes into 
the box.  Ideally you don't want the RF getting inside the box though, 
because it can then capacitively couple to other things.  Finally, 
small-value ceramic capacitors across input and output cables act like a 
short circuit at MHz frequencies.  They will take RF currents that 
arrive at your unit on one conductor and send them packing in the 
opposite direction on the other conductor.

EMC textbooks have loads of tips like this for building in good 
performance at the design stage.  Unfortunately, when the products 
themselves weren't designed as immune as they should be, that leaves us 
having to play EMC engineer clipping ferrites on the external cables, 
fitting external mains filters, etc.

-Richie,