[sdiy] reducing crosstalk

[Peter Keller]

On Fri, Jan 25, 2008 at 09:01:26PM +0100, Derek Holzer wrote:
> Where should I start with this? Is the crosstalk coming from a bad 
> grounding? The ICs? The connection wires? I'd be happy to upload a photo 
> of the circuit and a drawing of the schemo if that would help. In 
> general, where would such crosstalk come from? Do I need to etch a board 
> with a good ground plain? Since I want to use the output as either audio 
> or CV, high-pass caps for DC removal isn't an option. Or is it?

I'm kinda interested in this problem. Can you post a picture of it?

I've run into serious crosstalk a few times, and usually it is a result of:

A) Since you have an unregulated supply it might variably droop too much,
which would induce signals into other parts of the circuit.

B) If you run lines parallel to each other for any moderate (like 5 inches
or more) length of distance with high currents (10-15mA) in them, they
will easily capacitively and magnetically induct between themselves. Also
be aware about parallel analog/digital lines--especially clock lines. For
your photodiodes, I'd have a plus/gnd pair of wires for each one and
twist the signal wires to quell the magnetic interference. You can use
shielded 2-pair conductors for this, and connect the shield on the
receiver/board side to ground.

C) You should use bypass capacitors for all of your IC Vdd pins, have them
be 0.01uF and keep the leads as short as you can.

D) Check this page out which seems to do a good job explaining grounding.
http://ai.kaist.ac.kr/~suh/DIY/ground

I'm sure there are more ways crosstalk can show up, but these are ways I've
had personal experience with.

Later,
-pete