FX Switching Noise/Ground Loops Questions

[aot]

Hi All,

So my question is about how to eliminate ground loops and noise in a guitar fx loop switching system.  Just so people know, I've got a good bit of experience with electronics, including EMC in RF circuits, this has me a bit stumped.  Please tell me off if I'm doing something stupid, but quite equally don't dumb down the technical things. Googling ground loops fx switching or relays tends to just give either adverts for hum elimination devices, or advice like 'use isolated power supplied', or 'lift the grounds':

I'm in the processing of building a relay based guitar fx pedal.  The idea is based on R.G. Keen's remote switching system (http://www.geofex.com/article_folders/rmtswtch/rmtsw.htm), using the relay switching modules and an arduino as the basis for the control circuitry.  To elaborate a bit more, I'm using 595 shift registers to give a lot of outputs for switching both FX loops and auxiliary devices (such as amplifier channels) , drive an LCD screen, save patches to an SD card in a plain text format (for easy PC editing), allow midi channel switching too for things like a digitech whammy.  To begin with, I'm making a  simple working prototype with just two relay loops, and all the switching circuitry on breadboard.

The relay switching module circuits are identical to R.G. Keen's, except that I am using a plastic stereo jack rather than two mono jacks (the circuit is normalised).  I've prototyped two modules on stripboard.  The guitar signal path is isolated from the switching signal path.  I've put these two modules into a steel box, along with two more (plastic) mono jacks for input/output (the box is similar to the standard marshall amplifier footswitch box - folded steel plate design).

So first question, what should I ground.  Originally I did not ground the guitar signal to the box, but the power supply for the switching was grounded to the case (metal power supply jack unfortunately, centre pin positive).  Now with no power connected I was getting a ground loop if I touched the metal casing.  Connecting the power supply resulted in a lot of ground loop hum.  To be clear I'm going into the high gain channel on my amp with the gain fairly high so I can diagnose things, with just the guitar and a decent cable there isn't any hum).

If I ground both the switching circuitry, and the guitar signal to the case then the hum is removed.  Seems like a solution, but I would rather keep my switching circuitry and the guitar signal paths isolated, and I have no idea what this will do once I start adding FX pedals with their own power supplies, especially once this thing is scaled up to 8 relay modules.

Another problem is that one of the pedals - a rocktron zombie which is a high gain distortion pedal oscillates (at 800-1000Hz) when put into the loop.  I have no idea why its doing this, perhaps there's a feedback loop set up, or the switcher causes the output to be capacitive which causes the pedal to oscillate.

Any help or advice is greatly appreciated!

Cheers,

Aaron

[R O Tiree]

First point is that your body attracts "hum" from all sorts of sources... mains AC, fluorescent strip lights, your computer monitor, any SMPS that happens to be lying around, etc. It's because you're mostly made of salty water and also your torso is lens-shaped (some people more than others...), so you tend to focus it. So, it's natural for this box of tricks to hum like a mad thing when you touch it if your signal and circuit grounds are separate. You hear/read about people saying that their guitars hum unless they touch the strings and they are thus grounding their guitars... RUBBISH! It's the guitar grounding THEM, not the other way around. So, when you touch the box, you are actually introducing more hum to the system.

You proved for yourself that, if you put all the grounds together, the hum disappears which, again, is perfectly logical.

Examine your entire ground path... if it goes A-B, B-C, C-D, ...Z-A again, then you do, indeed, have a ground loop. If the loop is not complete, then you don't. Copper traces have a finite resistance, though and a useful rule of thumb I saw recently on these boards (somewhere) for high-gain circuits is not to have more than 6 trace-inches from the point where your 0V wire enters the board to the furthest component that needs grounding. The component that is farthest away will have a tiny, but measurable, potential above ground and, the futher away it gets, the worse it gets. Plus/minus 6" leaves you plenty of flex, though.

As to the Zombie... it may be the separation between signal and circuit grounds that is doing it, or something else. Hard to tell. A quick search of the archives of these boards reveals a post from Mark Hammer to teddybear back in 2007, saying that just the IC alone in this pedal is set up for a gain of 235 (and that's not even considering the gains in the rest of the circuit). So, 1mV of hum will HOWL louder than your average single-coil equipped Strat at full bore! Quite where it's getting this 800Hz hum from, though, is a mystery.

[PRR]

In hi-gain audio you MUST(*) ground case to audio shielding-ground, as you learned.

If you are doing relay switching, you do NOT need to ground logic/relay power to audio ground.

> metal power supply jack unfortunately

So get a proper insulated jack.

(*)Alternatively, don't audio-ground the box (at ALL) but run totally enclosed shielding around each audio jack, shielded cable to *and around* each relay. I have seen work like this. It is incredibly tedious to build, impossible to debug/repair.

> high gain distortion pedal oscillates

If the output of a hi-gain amp is brought near its input, it will oscillate. (There's precise theory on the necessary conditions, but Murphy's Law makes it highly likely.) That's what you done by bringing in and out to the same relay, pins 0.1" apart. Shielding bewteen pins may help. It might be necessary to do the loop in two relays several inches apart.

[aot]

Thanks for both of your posts.  Isolating the logic/relay ground from the box was next on my list of things to try, so I'll give that a go, along with other suggestions here and report back.  Regarding the oscillation, I've actually had this issue before using standard switches and this pedal - I tried to make a single switchable fx loop with a switchable insert using a 4PST footswitch and two DPDT toggle switches - the zombie was placed as an insert using a stereo jack.  So its either the stereo jack or the close proximity of the relay pins.  In any case I'll do some testing to figure out what's causing it.  Perhaps an appropriate resistor on the output could dampen these oscillations

[R O Tiree]

Paul's right - changing the power jack for an insulated one would make things a lot easier and robust and would, indeed, allow you to separate logic/switching from audio. I'd clocked that the jack was a metal-cased, pin-positive one, which is why I suggested bringing all the grounds together.

I suppose another reason would be that the de facto industry standard for (certainly 9V FX) power supply jacks is the plastic cased "Boss-style" 2.1mm pin, pin negative, barrel positive. It would only be a matter of time before you plugged a "standard" jack into it by mistake, with possibly disastrous results for your Arduino, etc.

As to Paul's suggestion about the Zombie, I think he's probably hit the nail on the head. I once built a Fuzz Factory on veroboard for a friend and I had the input and output on adjacent tracks. Howled like a banshee most of the time. I should have remembered that one. Sorry. Even after we got it sorted out, neither of us liked it at all, so I de-soldered the precious Ge trannies and threw the rest in the bin. Anyway, try two separate mono jacks, two separate patch leads and two separate relays (one to switch the input, one to switch the output). Belt and Braces solution

[aot]

I had a go at isolating the power jack from the metal box - using some electrical table around where the jack makes contact with the case as I didn't have a plastic socket,  and I checked with a multimeter to ensure there was no continuity between the case and the power socket ground.  I still got a lot of buzzing, so it looks like this didn't work! Unless I'm doing something wrong?

I ended up trying different value resistors to connect the logic ground to the signal ground (keeping the case grounded to the signal ground), and settled on a 2.2k resistor as it resulted in inaudible amounts of hum going into a high gain channel. I didn't do an exhaustive search of all the resistor values though, so this could probably be raised a bit

[fzzynvts{2}]

First point is that your body attracts "hum" from all sorts of sources... mains AC, fluorescent strip lights, your computer monitor, any SMPS that happens to be lying around, etc. It's because you're mostly made of salty water and also your torso is lens-shaped (some people more than others...), so you tend to focus it. So, it's natural for this box of tricks to hum like a mad thing when you touch it if your signal and circuit grounds are separate. You hear/read about people saying that their guitars hum unless they touch the strings and they are thus grounding their guitars... RUBBISH! It's the guitar grounding THEM, not the other way around. So, when you touch the box, you are actually introducing more hum to the system.                    oved for yourself that, if you put all the grounds together, the hum disappears which, again, is perfectly logical.

Examine your entire ground path... if it goes A-B, B-C, C-D, ...Z-A again, then you do, indeed, have a ground loop. If the loop is not complete, then you don't. Copper traces have a finite resistance, though and a useful rule of thumb I saw recently on these boards (somewhere) for high-gain circuits is not to have more than 6 trace-inches from the point where your 0V wire enters the board to the furthest component that needs grounding. The component that is farthest away will have a tiny, but measurable, potential above ground and, the futher away it gets, the worse it gets. Plus/minus 6" leaves you plenty of flex, though.

As to the Zombie... it may be the separation between signal and circuit grounds that is doing it, or something else. Hard to tell. A quick search of the archives of these boards reveals a post from Mark Hammer to teddybear back in 2007, saying that just the IC alone in this pedal is set up for a gain of 235 (and that's not even considering the gains in the rest of the circuit). So, 1mV of hum will HOWL louder than your average single-coil equipped Strat at full bore! Quite where it's getting this 800Hz hum from, though, is a mystery.

What if I, in equal parts, love the act through which I currently lay blame for my lensish shape as much as rockin excellent, buzz free tone, but labor under an unnatural phobia for sheilded cords? Mars relocation? Aquire appreciation for acoustic inst. and small crowds? My lens shape keeps the groupies at bay. (My first and last* smartas post. Thanks in adv. for your tolerance and restraint. Happy new year!) *--probably. We try.