10 linear power supplies on one board: shared ground? or isolated ground planes?

[xefned]

I'm still debating the idea of making a small batch of PCBs for the Weber stompbox transformer with 9 secondary windings.

If I were smart, would I share the ground plane among all 10 PSUs, or give each PSU a separate plane?



I found a debate where both methods were criticized, at least hypothetically, with no real conclusion.
My hunch is that fully floating grounds will eliminate the possibility of hum through shared grounds. This is how I've always run mine. Any other thoughts out there?

RG Keen's Spyder design seems to imply floating is better. Scott Swartz seems to do it this way too at his GGG project. Same situation with the layout at Geofex. Still, I can't help but worry that this is somehow problematic. I don't own a commercial unit to open and see how the pros do it.

 

[xefned]

OK, ISOLATED it is! I read a bunch of posts by R.G. Keen over at DIYSB and the answer seems to be obvious: "Of course you isolate the grounds, you just isolated everything else along the way, why share ground hum between channels now?" I will now let this thread sink to the bottom of the ocean floor.

FWIW, I've been using this design with floating grounds for almost 10 years with no problems so I'm gonna fret over one dude's concern.

This project is the opposite of economical. Each channel has an adjustable regulator, 3 smoothing electrolytics: 1000µF, 100µF, 10µF, and of course, separate bridge rectifiers and transformer windings (thanks to the Weber) for every channel. It's really the gonzo way to do it.

Sensible people will just buy the Walrus Aetos; it has separate transformer secondaries for each channel. And at $170, it's less than the cost of a boutique pedal. No brainer. Incidentally, I contacted Walrus Audio support and they confirmed each 9v output has a dedicated secondary. I believe Voodoo Labs does it this way too. Some other brands appear to use the word "isolated" even when they're just regulating the same AC source. I guess "isolated" can mean whatever you want.

But I already have one of these Weber transformers laying around. They've been selling them since 2004 so a few other people out there could probably use this. Years later, they will find this thread thanks to search keywords like: WPDLXFMR-1 Transformer, DIY stompbox power supply.

 

[Stickman393]

I believe you came to the correct conclusion.

The thing that comes to mind for me here is that, regardless of if the power supplies are floating or not, many guitar pedal circuits will share one element: they will bond the negative end of the power supply to instrument ground.

So, once the power supply is connected to the FX, and then to the instrument, they're no longer floating independently. They all share a common ground point: that is, instrument ground.

It would seem to me that good practice would be to keep these grounds isolated from each other until they reach the circuit in which power is being used. If one was to, say, bond all those grounds together at the PS, and then have all the pedals that are being powered share a common ground via the instrument cable, you would have created a circular path in which electrons could flow with little resistance.

I think I read about as far as "ground dirt" and then my eyes glazed over a bit. I suppose maybe this dude is referring to noise inherent in the mains power supply? Though that's got nothing to do with ground. Ground is ground, ground is dirt, ground is simply what we tend to use in the states as a center reference for 230v single phase utility power. We bond our neutral wire's potential to ground, making 115v difference between ground (or neutral) and L1 or L2.

We choose ground cause...well...why not. It's a large reservoir of atoms and electrons and its not real easy to push those electrons around. It provides consistency, and with circuit breakers it keeps us safe if conductors break loose and energize something that has low resistance to ground. Current follows that path and the circuit breaker trips, instead of a floating setup where...ah...you might touch something and just happen to present yourself as a path.

Granted...I'm relatively uneducated in electrical theory. They taught us the bare minimum in the apprenticeship, a lot of what I've put together has been independent study. So I can, and often do, get things wrong.

 

[fig]

We choose ground cause...well...why not.

I've always thought the electrons did the choosing...interesting stuff! That is part of the whole electron hole in my edyoukashun.

 

[Stickman393]

Electrons...they just wanna hop to the next atom with an inviting, empty spot in it's orbit. But for current to continually flow through a conductor...instead of just hopping from one spot to another, something has to continually maintain potential on either side of that conductor.

Tie one end of that voltage generator to ground, and you essentially force that line to SIT DOWN, STOP MOVING, jeezus, kids these days...

While line 2 continues acting a fool, jumping back and forth around the other kid. Who is also line 1, but he's grounded.

Eh? Eh?

 

[fig]

Mr. Van de Graff would approve that message

 

[xefned]

2½ years later, I finally got the nerve to send for a prototype. Wish me luck!

It's the product nobody needs, which seems like a fine reason to proceed.

 

[andrewsrea]

If each AC transformer winding has its own isolated ground, then keep the rectifier-regulator-filter circuit isolated. Your ground of your pedal-to-pedal instrument cables will serve as the 'star grounding system.'

I utilize a simpler method of star-ground in the supply, with the positive voltage having a 10 ohm 1/2w isolation resistor, only send ground to one pedal and positive to all pedals. The instrument cable serves as a ground and my noise floor is inaudible. I use a rectified 1.5A 18VDC switching supply as the brick's power source and use a LM7809 1.5A regulator, allowing both 18V and 9V ports.

The advantage of my method is not to have an EMF from the magnetics near the rectification, regulation and distribution. The disadvantage, you need to use good cables and keep the pedal jacks clean and rust-free. I perform an annual check and my pedal boards have been operating reliably for years.

 

[xefned]

Great info, thanks! What a unique concept. So I could lift the ground on 8 of the 9 DC jacks, and it would still work. That never occurred to me. Great idea. I will experiment with it.

 

[andrewsrea]

Great info, thanks! What a unique concept. So I could lift the ground on 8 of the 9 DC jacks, and it would still work. That never occurred to me. Great idea. I will experiment with it.

Correct. I typically connect the ground 'feed' to the pedal with the highest power requirement (typically a digital delay or reverb).

 

[xefned]

I once built a power supply that called for a 10Ω resistor from the supply ground to the chassis ground. I didn't fully understand what that does. But you use a 10Ω resistor on the positive rail? to ground?

I was advised to "add pads" because "pads are free" so I've added a pad to the ground plane of each and a separate rail.

A wise and powerful legend in the world of pro audio advised me thusly:

I would put in pads for a 100 ohm resistor to link the grounds and maybe even a .1U cap, just so you have them, having them lightly tied together might be good for safety, hum, and popping when plugging them in. They can be open, the R, a jumper, a cap or R and C

I probably should have added twice as many pads for both a resistor and a capacitor to chassis ground on each supply, but space was tight on the board. And I suspect I'll run all 9 supplies with floating grounds.

 

[andrewsrea]

I once built a power supply that called for a 10Ω resistor from the supply ground to the chassis ground. I didn't fully understand what that does. But you use a 10Ω resistor on the positive rail? to ground?

I was advised to "add pads" because "pads are free" so I've added a pad to the ground plane of each and a separate rail.

A wise and powerful legend in the world of pro audio advised me thusly:



I probably should have added twice as many pads for both a resistor and a capacitor to chassis ground on each supply, but space was tight on the board. And I suspect I'll run all 9 supplies with floating grounds.

To mitigate eddy currents keep the signal circuit ground in a 'star' (emanating from one point), then connect that to 'earth' (chassis ground) via a very low ohms resistor (10 ohms works, I typically use a 5 ohm). It keeps the junction of the dissimilar metals of the enclosure (typically aluminum) and the circuit ground (typically copper), from inducing eddy currents into the ground system or signal path.

I do this for tube amps and can see the value of doing this for a power supply, but for pedals I simply keep the connection to chassis using the output jack and use isolated Cliff jacks for the input jack.

In as such, I would recommend connecting each isolated circuit to a common point using a 10-ohm resistor for each and wiring that junction to chassis.

 

[xefned]

To mitigate eddy currents keep the signal circuit ground in a 'star' (emanating from one point), then connect that to 'earth' (chassis ground) via a very low ohms resistor (10 ohms works, I typically use a 5 ohm). It keeps the junction of the dissimilar metals of the enclosure (typically aluminum) and the circuit ground (typically copper), from inducing eddy currents into the ground system or signal path.

I do this for tube amps and can see the value of doing this for a power supply, but for pedals I simply keep the connection to chassis using the output jack and use isolated Cliff jacks for the input jack.

In as such, I would recommend connecting each isolated circuit to a common point using a 10-ohm resistor for each and wiring that junction to chassis.

Thanks! I will give that a try.