Optimizing PCB design for Audio Applications

[CranktPedalworks]

Hey gang, been referencing this site for quite a while and thought I'd use the chance to give something back.  Was looking into various aspects of PCB design and came across this well written pdf paper on optimizing the drawing of traces, component placement, etc in regards to audio PCB boards and thought I'd take a moment to share the find with the group.

http://www.ikonavs.com/OptimisedForAudio.pdf

 

[CranktPedalworks]

Did anyone else find it remarkable that signals can "bounce" off sharp turns on traces if the curve is not perfectly contoured and exact trace width maintained?  Any idea what effect this might have on the signal?

[Digital Larry]

I doubt it has much impact on audio rate signals.  This is mostly a concern for super fast rising edges as you might find in high speed digital.

[alanp]

I need to change my .sig, to "It's amazing what you can get away with." Especially in guitar pedals.

Worst that'll happen is that it'll distort, then you can just claim that it's *supposed* to do that!

[anotherjim]

It's all transmission wave physics. Wavelength of audio signal is too long to be bothered by it.
Any change in impedance causes a reflections of the wave. This is bad because it interferes with and messes with the timing and amplitude at the receiver. A sharp corner track has a diagonal longer than the width - there is more conductor there so less resistance. There are 2 changes close together. Entering the corner and leaving it.
Have a look at some of the funny routes the traces make on a PC motherboard.

If you have an internet connection over your copper phone wires, you will be truly lucky not to have met the consequence of a partial change of resistance in the connection. Voice calls work but the DSL won't connect. Now the Telco actual use wave reflection to test the wires from the exchange to you. A fast pulse (or a step in the line voltage) is sent and the reflection out and back is timed. If it's sooner than it should be for the length of wire, they know there is a line fault - and roughly how far away it is.
Only problem with that is if the fault is close to either end (it usually is at a cable junction), the test won't find it, because it's only some nanoseconds different. So they will insist the line is ok.

[R.G.]

The title of the paper is misleading.

It would be more clear if it was something like "Optimizing PCB Design for High Speed Digital Circuits Which Must Also Include An Analog Audio Section".

The design tips are good, but they're mostly tips about how to keep your high speed logic signals from spilling over into analog. They're mostly "don't care" items for pure-analog pedals.