Hum problem with transformer isolated guitar splitter

[rockhorst]

I'm working on a guitar splitter to drive three amps from a single guitar signal. I've had some requests for this from local guitarists. Some need transformer isolation, some don't, so I thought I'd design a board including transformers and a split rail design. The idea is basically similar to RG's Hum Free Splitter and various other projects (like the splitter on BYOC and some Jensen design notes).

Here's my schematic and pcb layout:





As per the various build notes, the output jacks are isolated from the enclosure. All channels pass signal, but the two transformer outputs are noisy as hell. It basically sounds like bad grounding. The noise goes away when grounding the output jacks to the enclosure via the common ground of the circuit, but that defeats the isolation. I've seen a post by RG noting that it should be a steel enclosure, not an aluminum one, but somehow I doubt that's the problem.

My current best guess is hum induced through the ground plane which runs underneath the jacks and transformers, which would be fixable with a new board. As these boards are quite big, they are relatively expensive to order, so I'd like to rule out other problems before getting a new batch.

Any suggestions on what might be causing the hum problems and how to kill it? Thanks in advance!

[rockhorst]

And here's a gutshot of the (humming) prototype for completeness

[antonis]

Sorry for the off-topic but...

What is the purpose (and the actual values) of R6 and R10..??

[rockhorst]

What is the purpose (and the actual values) of R6 and R10..??

They are jumper wires in a resistor package, so they are 0 Ohms as in the schematic. The purpose is to have the option to flip the phase of one of the signals if it were necessary for a particular amp setup. If you install them vertically instead of horizontally (see photo) the phase is flipped.

[antonis]

Probably my eyes need replacement..

(for sometime I thought that, with the vertical installation you sort-circuit the transformer's secondary...!!!)

[R.G.]

Try running a 100R to 1K resistor from one or the other or both of the "isolated" ground output jacks to the enclosure/signal ground on the input side.

Hum is cunning, stealthy, and persistent, as well as having a plethora of tricks and abilities. One if them is that a transformer isolated output can get hum from its input if the input is floating. The input ground has to be grounded somewhere, or it picks up airborne hum and converts it to a legal signal for passing to the transformer. Using a resistance to the isolated outputs can "ground" the input when all you're using is the isolated outputs.

Another slippery trick is that transformers make good pickups for hum in some circumstances. Try moving the whole package around and seeing if the hum varies with location. If it does, the trannies are picking it up. The closed aluminum box may or may not help with low frequency power line hum if it comes in on a magnetic field.

[rockhorst]

Try running a 100R to 1K resistor from one or the other or both of the "isolated" ground output jacks to the enclosure/signal ground on the input side.

I'm not entirely sure I'm reading your post correctly RG. I think you're suggesting to alter the output section in the following way:

Is that correct?

I tried this and the noise does go away (although my amp setup doesn't need transformers, it's a friend's setup that does and I can't test there right now). It does confuse me somewhat though. Doesn't this (at least in part) defeat the isolation? The jack that carries the input guitar signal (left in my photo) grounds the input cable and the pcb to the enclosure. With this mod the enclosure also grounds the 'isolated' outputs, although they're separated by a 100 Ohm resistor (higher value = more separation?). Could you elaborate on that a bit?

Also, I haven't seen any other reports of people having to do this mod when building for instance this BYOC kit or your circuit. It doesn't turn up in any schematic by yourself, BYOC or for instance Jensen.

[JFace]

You mentioned Jensen. Check out this ground lift implementation. It grounds audio frequencies to chassis, and could eliminate the hum. In cases where the amps share the same ground, the switch can be set to hard wired to chassis.

http://www.jensen-transformers.com/as/as066.pdf

[R.G.]

I'm not entirely sure I'm reading your post correctly RG. I think you're suggesting to alter the output section in the following way:
Is that correct?

Yep.

I tried this and the noise does go away... It does confuse me somewhat though. Doesn't this (at least in part) defeat the isolation? The jack that carries the input guitar signal (left in my photo) grounds the input cable and the pcb to the enclosure. With this mod the enclosure also grounds the 'isolated' outputs, although they're separated by a 100 Ohm resistor (higher value = more separation?). Could you elaborate on that a bit?

Sure. As I mentioned in passing, hum comes in a lot of forms, and they all sound similar to our ears. Often hum happens when there is a small voltage difference but low impedances so that large currents can flow. The hum we get, especially with coils involved, is magnetic pickup of the current flow. Another form is high impedance pickup of hum from the air by capacitive pickup.

For the current-flow pickup, you can stop or dramatically reduce the apparent hum by cutting the current flow to nil. The brute force approach is to completely break the conductive path that the hum current flows in. The more subtle approach is to introduce a (relatively!) high resistance that cuts the hum-frequency current flow down to below audibility with that particular set of coils, wires, and positioning. Ground lifts on audio equipment and amps are the brute-force method that have the disadvantage of opening you up to electrocution as this defeats the safety aspect of the grounding. Partial lifting with resistors, diodes, capacitors, and so on is a modestly safer way, and that's what most modern "ground lift" stuff does.

If you completely open the ground path, it's  - well, open. If you open all the ground paths out to your guitar, the guitar acts much like it does with an open shield - it picks up capacitive mode hum as well as the normal M-field pickup the coils in the pickups do. In this box, the input is "grounded" to the case, but the outputs are isolated from case, so you can get situations where the guitar waves around in the air, gets capacitive pickup, the opamps convert this to a nice, solid signal and put that on the transformers to be transferred to the amps. In this case, the transformers are doing what they're told, but due to the isolated nature of the input, they're being told to send hum. The cure is to tie the "isolated" grounds to something that is really tied to a power line ground with a resistance (i.e., not frequency selective) that is small compared to capacitive pickup but large compared to the value that lets small-voltage offsets cause M-field pickup. So medium-value resistors "ground" the setup for capacitive pickup but "lift" it for circulating currents.

You can still get hum by setting the box with the transformers in in on an amp that has a magnetically leaky transformer and getting coil pickup of the magnetic field, but that's yet another way. Neither isolation of grounds nor soft pull-downs will fix that one. Only high-mu shielding or better yet distance will help that one a lot.

Also, I haven't seen any other reports of people having to do this mod when building for instance ... or your circuit. It doesn't turn up in any schematic by yourself, BYOC or for instance Jensen.

One of my favorite aphorisms about the net is that the internet is very wide, but it isn't very deep. The net presents itself as a vast source of wisdom. It's not. It's a vast source of random information, much of which is pure cr... er, fertilizer. Go look up the Dunning-Kruger effect for an understanding of some of this. It's often true that the more places you look the more confused you can get.

Not seeing reports is just that - not seeing reports. With a phenomena like hum that has so very many sources and causes, the reports on the net could either result from the people that run into problems not reporting them for various reasons, or from the reporting people happening not to hit the problem. You now personally know that this version of hum induction is possible, right? And that this is at least one possible cure for it? So other people didn't report it. Shrug. You have now reported it and the internet is better - if the next guy through happens to read this. If not, well, he hasn't seen reports of this.

Also - and this is not a slam at BYOC, just an observation - BYOC is in business to make money from kits more than being an in-depth technical resource. We see a lot of people coming here looking for fixes for BYOC builds that had problems that they didn't get solutions for at BYOC. No criticism implied, they undoubtedly do a good job of what they do; but they have to stay focused to stay in business.

Jensen is very technically adept - and their business is to sell very high quality transformers at a profit to keep the company running. Their guitar isolator was the inspiration for the transformer isolator at geofex. It needed some work to achieve better frequency response with more guitar situations and also needed (in my opinion) some work to achieve good results with much cheaper transformers. That use of the amplifiers to extend low frequency response for cheap trannies is the real value in the geofex design. That's the most important bit I added.

As a side note, I find it amusing when stuff I put on the net early on at geofex has been taken and run with by others. The BYOC project is remarkably similar to the primal project on the topic at geofex - which is OK, as I have no legal right to restrict others from doing that. Those things on geofex are simply design work by, as the patent offices says, one skilled in the art. Other notable ones include the many-winding isolated power supply and a funny one in the battery simulator where the impedance of a battery is simulated by a variable voltage regulator and some resistance. Someone submitted the patent application on that one about six months after it went up at geofex and was actually awarded a patent. That one surprised me, and made me think less of the patent office at the same time. 

It is also entirely possible that these and others are just cases of converging designs too. But I don't put up as much stuff at geofex as I used to. 

[rockhorst]

A thousand thanks R.G. Very illuminating.
I encounter the shallowness of the internet and it's readers regularly when teaching physics (last year a student insisted I had Archimedes' principle wrong, because the internet said so). It's funny to see myself falling for the same mistake.

@JFace: thanks for that application note, I had missed it.
Does the 10n cap in series with the 51 Ohm resistor make it frequency selective?

R.G. basically stated the useful range of resistors to try r(100 Ohm to 1KOhm), so I may be asking more than I can chew: what are typical numbers for capacitive pickup and M-field pickup?

[R.G.]

A thousand thanks R.G. Very illuminating.
I encounter the shallowness of the internet and it's readers regularly when teaching physics (last year a student insisted I had Archimedes' principle wrong, because the internet said so). It's funny to see myself falling for the same mistake.

The thing that makes this hard to deal with is that sometimes the internet is right. It once again puts the premium on on to go LEARN so one can separate the wheat from the cr... er, chaff.

@JFace: thanks for that application note, I had missed it.
Does the 10n cap in series with the 51 Ohm resistor make it frequency selective?

51 ohms does the same as the 100 ohms I mentioned. This value is somewhat arbitrary. The 10n does make it frequency selective. The 10n is a "short circuit" at RF, so it helps tame RF picked up.

R.G. basically stated the useful range of resistors to try r(100 Ohm to 1KOhm),

That was a raw guess. I've seen as low at 10 ohms and as high as 10K.

so I may be asking more than I can chew: what are typical numbers for capacitive pickup and M-field pickup?

Different mechanisms, maybe different cures.

[bool]

It's long since I last worked with trafos, but here's an "empirical approach" suggestion:

I gather that you don't have much tech info on your trafos? Here's what I would do. I would install 1K trimmers as ground-lift resistors and then work over all possible/intended connection scenarios and set the trimmers to the most "optimal" value that works in most cases. Then replace with fixed resistors. Also imho you don't need to rework your PCB because it's entirely possible to fit these resistors on a ground-lift switch, PTP way.. ymmv

[rockhorst]

Great suggestion Bool! I was going to just modify the PCBs on hand, but there's more then enough room to encorporate it into the design for later use.