For anyone who's designed PC audio boards...

[markk]

Does it matter if you use a ground plane layer in a typical audio PC board? Any reasons not to. I'm just finishing up a prototype PC board for my active eq and was wondering if I should use a GP layer?

[johnk]

> Does it matter if you use a ground plane
> layer in a typical audio PC board? Any
> reasons not to. I'm just finishing up a
> prototype PC board for my active eq and was
> wondering if I should use a GP layer?

It probable won't make a differences either way. I've done both with no apparent difference.

[Pete Schumacher ®]

> Does it matter if you use a ground plane
> layer in a typical audio PC board? Any
> reasons not to. I'm just finishing up a
> prototype PC board for my active eq and was
> wondering if I should use a GP layer?

Having all that copper is certainly an advantage from an impedance standpoint, and it's a good shield for tougher EMI environments.

But as long as you have plenty of copper for your ground traces and you're not talking high frequency current spikes, an entire plane isn't always necessary.

[markk]

Thanks guys,

It's just a prototype board. I started with all the horizontal traces on one side and vertical on the other, but it's not really that complex or crowded and it was easy enough to move everything but the ground to the bottom layer. Basic boards are cheap enough that it's a reasonable gamble to prototype this way. (No soldermask or silkscreen).

I cut the GP out of the drawing below.

[Pete Schumacher ®]

> Thanks guys,

> It's just a prototype board. I started with
> all the horizontal traces on one side and
> vertical on the other, but it's not really
> that complex or crowded and it was easy
> enough to move everything but the ground to
> the bottom layer. Basic boards are cheap
> enough that it's a reasonable gamble to
> prototype this way. (No soldermask or
> silkscreen).

> I cut the GP out of the drawing below.

If those are op-amps, I'd surely want to add some good decoupling caps to the 15V rails, near each amp. .1uF ought to be good, with a 10uF at the board input.

[markk]

> If those are op-amps, I'd surely want to add
> some good decoupling caps to the 15V rails,
> near each amp. .1uF ought to be good, with a
> 10uF at the board input.

I've read that, and it's easy enough to place small .1uF pairs at each opa2134. What's the rationale for the 10uF's? They're a little hard to squeeze in this prototype board. I'm sure I can squeeze it tighter, but I've only got 2.5x3.8" to work with.

[Pete Schumacher ®]

> I've read that, and it's easy enough to
> place small .1uF pairs at each opa2134.
> What's the rationale for the 10uF's? They're
> a little hard to squeeze in this prototype
> board. I'm sure I can squeeze it tighter,
> but I've only got 2.5x3.8" to work
> with.

I guess it all depends on how far from the voltage source you are, but the .1uF will probably do the trick. There's always the chance to induce ringing in the power rails, and having local decoupling takes care of that.

I figured a cheap BPE wouldn't be that large, but then again, you're the one having to squeeze it on. ;-)

[johnk]

> I guess it all depends on how far from the
> voltage source you are, but the .1uF will
> probably do the trick. There's always the
> chance to induce ringing in the power rails,
> and having local decoupling takes care of
> that.

> I figured a cheap BPE wouldn't be that
> large, but then again, you're the one having
> to squeeze it on. ;-)

On boards of this size (about 2.5 x 3.75) one 0.1uf from the rail to ground should be all that is needed. There is no need to place one at each opamp.

[unknownuser]

With the power rails so close together, surface mount across the traces would be a good way to go.


(Originally posted by: Jimbo)

[markk]

> With the power rails so close together,
> surface mount across the traces would be a
> good way to go.

Granted, this needs to be done in a pc board operating at Ghz frequencies. I'll even accept Mhz speeds.

When you're talking audio frequencies with a well behaved op amp like the opa2134, this sounds like overkill.

While I understand the rationale for bypass caps, using surface mount or using 8 bypass caps, with 2 larger ones on a small board next to a halfway decent regulated power supply, again, feeding a robust op amp (psrr for the opa series averages ~106dB) just seems overkill.

Part of design elegance is achieving the result with the minimum.

[Pete Schumacher ®]

> Granted, this needs to be done in a pc board
> operating at Ghz frequencies. I'll even
> accept Mhz speeds.

> When you're talking audio frequencies with a
> well behaved op amp like the opa2134, this
> sounds like overkill.

> While I understand the rationale for bypass
> caps, using surface mount or using 8 bypass
> caps, with 2 larger ones on a small board
> next to a halfway decent regulated power
> supply, again, feeding a robust op amp (psrr
> for the opa series averages ~106dB) just
> seems overkill.

> Part of design elegance is achieving the
> result with the minimum.

if the power runs are short to the source, then a local .1uF from each rail to ground is probably adequate.