Isolating ABY problem solved - but why?

[Andi]

Hi chaps - I was wondering if anyone could shed any light on this for me.

One of the things I make is an isolating (op-amp driving transformer) ABY pedal. Previously I've put them together on stripboard, but recently I had some PCBs made up - exact same circuit.

Built one up on PCB, same components as always, and it popped with any switch.

Built one up on stripboard, same components, put it in place of the PCB and no pops.

I've checked and rechecked the PCB and the circuit is exactly the same; could it be the physical layout that's causing the problem? Even pull down resistors at every decoupling cap couldn't fix the PCB one.

I am most puzzled and would greatly appreciate any wisdom - thanks in advance.

[R.G.]

It's likely that Mother Nature is trying to teach you something subtle. Could it be layout? Sure. Probably is. But WHAT???

And that can't be determined without looking at the layout of both the stripboard and PCB stuff.

I personally would bet that the signal ground on the buffer side shares a ground trace with the indicator leds, or possibly that you're using stereo jack switching and forcing the LED switch transient into the same wire that carries your signal ground. But I have no facts to back that. Those are just click causing mistakes that would be common in converting to PCB.

More help will need pictures.

[puretube]

http://diystompboxes.com/sboxforum/viewtopic.php?t=32618

(sorry...)

[Andi]

Thanks chaps.

One thing that's occured to me is that the stripboard layout has a separate buffer board and isolation board (I buffer the signal at the input to prevent loading in the Y mode), so that also means that there are 2 op-amps for that version, only one (using both sides) for the PCB version. I should be more careful about my "same circuit" claims.

This is the PCB layout - if you can point out anything obviously wrong I'd be most grateful. I don't have one for the stripboard version, but I'll see if I can do one.



Each ground needed (on the switches, for the LEDs and so on) is hooked up to a single point. I do use a stereo jack to switch power in both PCB and stripboard versions. The drain resistors for the decoupling caps aren't shown on this - I originally built without, and the stripboard version works without.

[puretube]

your output current, which wants to flow from the negative side of C3 to the minuspole of C1, can do so however it wants: down south underneath the xfmr, but also to the left, taking with it the flow of the input return at R4`s footpoint, where it`s mingling...
the part that flows through the right side, hovers on the ground return of R2.

The "one" ground point is not the battery`s minus pole, but the minus of C1.

"Where" you ground the pulldown resistors, matters too.

Actually, the PCB-layout is one (small-scale) ground-loop.
Try this: cut the groundplane through right "north" of R1. (mill with a drill).

You might still stay in trouble, if all 4 mounting holes are connected to the metall chassis with metal stand-offs:
isolate 3 of them from ground, and let only 1 of them make contact to the box.

[Andi]

Ah!

So it would be more sensible to "properly" route the ground traces rather than using a ground plane? I really should have realised that, but didn't - thanks. I think I'm stuck in the basic idea that grounds are "perfect" - ie sinking infinite current with no flow around them. I'll try cutting the ground plane as well and see if that helps.

I use all plastic standoffs, so no worries there. The pulldown resistors were positioned physically close to the caps, grounding straight onto the plane.

Thanks again for the help - I will redo the PCB layout and actually think about the ground a bit more!

[puretube]

no written guarantee from me, though;
but good points to start searching...

wish you success!

[Andi]

Fixed it - you chaps were essentially right. The left hand side of the PCB has its half of the op-amp wired as a buffer, with its own input and output. I ended up isolating it entirely from the right hand side transformer driver, and giving it its own ground connection. All the pops are now gone.

Thanks for putting me on the right track and getting me thinking about the right things.

[Phorhas]

Is it generally a good idea to give each "function"  block of a CRKT it's ground plane, then conecting each ground to the metal box?

[puretube]

to the chassis, or connect together on the PCB itself (and then run to the chassis- but:
at 1 physical location only!

Andi:  

[R.G.]

Ground planes come up now and then when we're talking about layouts.

In general, for audio you do NOT want ground planes. It's much better to have a specific conductor carrying ground current where you want it instead of an "everywhere" ground. For RF, the situation is reversed because you want inductance grounds more than you want specific control.

Where you mix analog and digital parts, you face the dillemma: plane or not. In general, use a plane if you have a continuously clocked part like a microprocessor and put the plane under the digital stuff only. Give the digital plane its own drain wire back to wherever you're calling "zero volts", and connect the analog grounds to the digital plane with one and only one wire.

Think of grounding as a sewer system. Keep the used-electron sewage from one section from mixing with that from another section because mixed backups are even worse than individual backups.

[Andi]

Thanks, RG. I will bear that in mind for future PCB layouts - and I should probably go back through the ones I have to redo them with "proper" ground traces.

[Phorhas]

so it's ok for the power block and signal blocks to share grounds on the PCB?

[puretube]

I love audio ground planes: not neccessarily as a proper ground-"bus" carrying the current,
but as a "meadow of peace & silence" amongst all the signal radiating trace-antennas and components and leads;

btw, nice analogy:
imagine a patchcable with only one side of the shield soldered to the groundlug of one of the plugs
- like being done as a remedy against hum-loops in multiple FX scenarios -
isn`t that a perfect 3-dimensional groundplane?