Ground loop isolation to get rid of hum?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Not sure where to put this thread, but the music area seems a good start. In my band, we have a completely self contained system in which all our sound sources get routed through our own mixer. We are very careful about ground loops to prevent excessive hum in performance situations. Ground lifts on virtually everything, a single point to allow grounding when needed, all 120 v equipment has its ground pins cut off, and are plugged into the same "Octopus", etc., etc. All we have to do is plug a pair of line level R/L RCA outputs into a sound system, and we're good to go.

But you know the story! If the main "house" soundboard is in a remote location, and conversion from ground referenced RCA style line to balanced outputs aren't provided, then all it takes is one little bit of leakage and you have a ground loop between our system and the main PA. The result of course is that all you can do is choose between one level of hum or another, based on whether or not your single point ground lift helps.

So what i want is to have our own "emergency" isolators. Something that would allow us to plug our line level RCA outputs into one side, and have completely ground isolated outputs, both RCA and XLR. maybe such a product already exists and is not too pricey, but my experience is that in the world of performance electronics, the simplest item is pricey enough to be able to easilly build 10 for the over the counter price of 1.

Any ideas?

(Moderator: Please feel free to move this post if there is a better area for it).
 
I'd go through jack, rather than try and stay in RCA. There are any number of good, passive DI boxes out there, giving you isolated, balanced outputs (passive because you don't need power supply, no batteries to go flat, and your mixer has a low impedance drive anyway). Optimum would be a stereo box with independent ground lift switches.

Of course, this being a DIY forum, perhaps you would like to build your own?

Even if an active transformerless output would do the job a little cheaper, I do recommend keeping the iron in there; it gives a much more total isolation. And you can carry your XLR-RCA adaptors with you to unbalance it again, without re-introducing ground problems. Here, however, you're probably going ti find you've lost considerable level (DI boxes are not one to one). Generally, though, an RCA input is minus ten dBs sensitivity, while the mixer can easily drive zero, and this ceases to be relevant.
 
@ chrispenycate Not sure what you mean by "go through jack", but I can tell you that mainly due to space considerations, our little Berringer (sp?) mixer has only RCA and 1/4 inch line level outputs (no balanced outputs). As far as the receiving end goes, if the room already has a DI box on stage, of course that solves everything. But sometimes all they have are extra XLR mic inputs), or as happened to us last Friday, actually had RCA jacks on the wall, wired into the sound system. As you can imagine, they were a problem in the making. The hum wasn't bad, but it was disturbing that I was powerless to do anything about it. And yes... I'm much prefer a DIY solution :)

So... I need to have a box that will cover both cases. And...

@ nightanole yes.. I agree... isolation transformers would probably solve everything.

But here's the bottom line.... these ground loops only seem to rear their ugly heads at performances. So if I am going to DIY this, I think I'd feel better implementing an existing design, in whole or in part. I wouldn't even know (for example) what transformers to buy for such a project, and which kid would leave me replacing ground loop noise with electricl or magnetic pickup noise. Any suggestions? :)
 
I guess if I used transformers (like the EDCOR Electronics Corporation. WSM6400) nightanole suggested, I'd almost have to add some op-amps to restore line level again. I can't think of any other way to maintain isolation, low impedance output, AND maintain line level. Of course if I make it a battery powered thing, it won't introduce any additional grounds or hum. And if I use two transformers back to back (10K->600ohm->600ohm->10K), then a a simple (gain=1) op amp circuit would be all thats needed to regain low impedance. Right?
 
You are miss reading, if you wire them like i said then they are just isolators and dont need opamps and stuff. What goes in is what goes out. so wired as i put its is 10k/10k no level shifting. You can also wire them for stereo to mono conversion, and 10k/600ohm adapters. The 10k means they will take down to a 10k load and be fine.
 
@nightanole: Ok, I'm a little confused how the same centertapped output secondary could be both 600 ohms and 10K. But assuming it is and can be wired as a 1-1 transformer, is it possible I might need a no gain (gain=1) driver amplifier on the output side? Consider that the mixer output probably does have an output amplifier to ensure that as a signal source, its impedance is very low. If I drive a 1/1 transformer that is basically 10K->10K impedance, is it possible I'd be creating a new avenue for noise pick up to get in? It just always seems to me that high impedance sources are greater noise makers than low Z outs. And line level is almost always pretty low Z out, right?
 
If you use a 1-1 transformer, the impedance is reflected through, even if the current drive isn't. And the quarter inch jacks on the output of your Behringer; are you certain they're not three pole balanced? A lot of Behringer gear is wired like that (I can't remember if it's servo balanced or just shorts out one leg and loses 6dB when you plug in a guitar jack).

As regards ohmic separation for ground currents, just about any transformer will do; concentrate on getting a good frequency response. There should be minimal risk from capacitive coupling between windings, as the grounds should both be at the same voltage apart from minimal inductance currents.

If you want to drive high levels into a low impedance, the trabsformer is going to be big and heavy, Reduce the level, or increase the impedance (that pair of RCAs is probably at something like 50k, hi-fi values. Bleh.) and the transformer shrinks nicely. But it's a good idea, if you can manage, to have transformers capable of taking anything the desk can pump out:- +18dB into 300Ω, perhaps, screened. And that's chunky. Something you want to attach to the box, not a PC.

There should be no need for an active driver; I haven't got the specs for your Behringer on hand, but it should have no trouble driving any reasonable line. And a screened transformer (or a steel box, rather than a diecast) will eliminate all reasonable electromagnetic interference from there (although not from guitar pickups and leads, Fender Rhodeses, Mellotrons…

Check about those quarter inchers; it could be that a simple ground lift cable will solve - no, better to have switches. then doing comparisons is immediate.

What model number is the mixer? I could check up on its output wiring.
 
Thanks chrispenycate . Well there's a question that sent me hastily rushing to re-read the manual! But no, it plainly says the main outputs are unbalance Mono jacks. Its their Model XYNYX- 802. Oh wel... still an amazing mixer for under $100! VERY quiet Mic pre-amps, but thats another story. But OK.. thats good to know about the impedance being reflected through from one side to the other. Geesh... it has been a LONG time since I learned anything about that in school :-(

Well size and weight is a concern too. And I DO remember from tube amp days how the output transformer on a good amp was significantly bigger (and heavier) than the power transformer, for having to work with lower frequencies. Since I am pumping full range audio through this thing, it does make even more of a case for some kind of active amplifier, if only to allow a small transformer to do the job of a bigger one. Just let the transformer output drive a hi-Z amplifier input.

Andy yes, controling grounds has been very helpful! With virtually every piece of gear now running on a transformer, all my power feed cords except one have their ground pins cut off. The one that still has it is broken out with an alligator clip so I can decide to manually clip power system ground to an audio ground, for the slim number of case where that will actually help.

I'm just trying to have a bag of emergency tricks at this point. Unfortunately, its nearly impossible to predict in advance what you're going to run into when you jack your mains into a house system, and you often don't have a lot of time to remedy things. I'm always buying and building little "magic" solutions, just in case one will work in a pinch.

Hey speaking of which, here's a little ground isolation thingie by Crutchfield. It's only $15 so i ordered one. Wonder whats under the hood?...

PAC SNI-1 RCA Ground Loop Noise Isolator at Crutchfield.com
 
If you want a wideband solution with CMRR over 120dB an isolation amplifier will do the trick. Isolation available up to 3Kv or so, a DC to DC converter module is needed to power the amplifier on the isolated side.

For something slightly less demanding buy or build an in line instrumentation amplifier. It still puzzles me why audio equipment uses ground referenced signals given the amount of effort spent on removing ground loops, time much better spent on a few dollars of op amps to get a differential input or output.
 
Well thanks metalsculptor . I don't know if I need to go as far as an actual active amplifier just yet, though yes its crossed my mind. I did order that PAC SNI-1 I posted a link to as it was only a couple of bucks. and yeah it did have a little low end dropout. It got a little better after I hacked it and removed some input coupling capacitors, but the transformers are a bit tiny. I think with a little larger transformers I'll get decent wide band coupling with acceptable loss. I've got some "real" audio isolation transformers on order. We'll see how they work out.

To answer your question though, it's kind of just something audio has gotten boxed into. On the one hand, grounding is primarily a safety issue, with standards created and set fourth by electric companies. So we have all this methodology that really is ill suited for audio. Add to that the simple fact that balanced signal transmission adds the cost of 3 conductor cables and additional components (like transformers), and what we have is a case of every piece of equipment made with an eye to "pushing" the ground isolation issue off to someone else to handle.

It should be getting easier, because at least for me, 100% of the equipment I use in performance works off little wall transformers and DC converters that isolate pretty well. But anytime you have a stage 100 feet from the sound board, there are always surprises. I'm convinced Murphy's first name must be Groundloop. So all I can do is keep adding to my bag of tricks, and hope to get one up on him more often then not. In fact I have such a performance coming up tonight. Wish me luck! :)
 
Good luck, avoiding a ground loop in a performance setting with long cable runs and multiple power sources will need a bit of luck. In the industrial instrumentation field we either locally sample and digitize then send the data over ethernet or use shielded twisted pair to differential inputs. Anyway going to a better quality signal isolation transformer would probably improve the low end, it is not easy to make a transformer give a flat response over 3 decades of frequency.
 
Metal: Well last night we ended up doing pretty well. It was a good test, but then again the sound guy was well equipped with the essential ingredient, DI boxes! Essentially they allow you to convert one or more of the low impedance balanced microphone input to the sound board, and interface to a line level out using an impedance matching transformer, and (I believe) some added passive electronics. Those boxes even have ground lift switches which surprisingly gave pristine quite results in the GROUNDED position! The sound was great, and our tracks do include some very high quality drum samples and string bass. I was impressed, and I think it would be wise for me to discover exactly what components and circuitry are inside one of those boxes.

Fenris: you're right... sorry. :) Bad eyes!
 
DI boxes come in several flavours; passive (which I prefer for low z sources such as synthesizers and mixers as, with no power requirements, they can give total ground isolation should it be needed) transformerless active, and transformer coupled active (better for pickups and really high impedance sources.

The former consist of merely a transformer, maybe a pad switch so you can take power amp outputs, a capacitor to remove any DC offset on the incoming signal, and a grounding switch, either on/off, or with added "resistive" link.

The hum rejection and audio quality depend entirely on the quality of the transformer (as does, unfortunately the price. No-one seems eager to make a really good cheap transformer). I have in the past used mic input transformers recuperated from old gear, with very acceptable results. They almost always step down the signal, assuming you want highish impedance input, but in your case this is not essential, and if you found a good pair if one to ones – but they will be bigger and heavier than stepdowns.

Ruddy Behringer techspecs; they give output drive and impedance for main outs, control room outs, headphone outs but nothing for the recording out. It is quite possible (likely, even) that it's lower level and higher impedance than the main outs, to match better to the domestic apparatus for which they are designed. In which case taking out of the jacks through a reasonable DI could end up with as much signal (and a better low end response; transformers work better driven from a lower impedance) as a one to one on the RCA connectors.
 
@ chrispenycate Thanks! I have some (hopefully) good 1:1s on order, and I am hoping that "passive" will be sufficient. It will be driven from my mixer output which puts out a clean signal up to +10db, so I'm sure I'll be able to compensate for any loss, but we'll see. As far as low end response, I had already tried a "cheapo" product that had some much smaller transformers, though really they were no smaller then I've found in microphones in the past, and I thought the low end dropout was significant.

I think I may have made an error in my testing though. My mixer is one of those that has a feature that drives me nuts... it has separate MAIN and MONITOR output level controls, but the MAIN control affects the monitor too. I wanted to change that behavior, but rater than do it by hacking the mixer itself, I made a separate passive level control box to pass the MAIN output through. It contains a simple dual 10K pot, and it allows me to change the MAIN level if I need to, without affecting the monitor mix. Anyway, when I tried that inexpensive dual transformer isolator, I connected it between the output side of my passive MAIN volume control, and the amplifier. Since that control was set somewhat lower than MAX, I was driving my transformer with what was likely a much higher Z source than if I'd fed it directly from the mixer. And since you're saying transformers work better driven from a low Z source, that may account for a lot of the low end loss I was hearing!

So I'll have to re-try my experiment. But even so, the transformers I have on order have at least 4X the iron mass of those in this cheap unit I bought, so I'm sure using them will be a wiser choice. If I do continue to use my external passive level control, I'll put it on the output side (though I'm seriously considering saying the hell with the warrantee and modifying the mixer!).

You mentioned isolation capacitors to filter any DC in the output, and I'm very curious what might be the best way to do that, or if its even necessary. Again, referring to the "cheap" unit I originally experimented with, it did have perhaps 10uF in series with the input side of each transformer, but I've always been skeptical of using a single polarized electrolytic capacitor in this way. On the other hand, the only place I've seen non-polarized electrolytic used has been in tweeter circuits in loudspeakers. So I would like to know what is the best capacitor type there, and how I'd choose a value. And bottom line, I measure less than 1mV dc offset on the output of either channel on my mixer, and seriously wonder whether I should even bother with a capacitor. What do you think?
 
Last edited:
I would check out the transformer specifications for any permissible DC, while grain oriented silicon steel might be able to soak up 0.1% DC some of the nickel alloy cores would have trouble with this.

If the DC still worries you, put a resistor in series with the transformer and look at the unloaded current drawn by the transformer with a oscilloscope when driven by a low frequency sine wave. If it is asymetrical compared with the output voltage voltage then DC might be an issue
 
how do I choose a sensible value?

You need to know the load impedance and the desired lower roll off frequency. A rough indication of the load impedance is the input impedance of the mixer times the transformer turns ratio squared. The capacitor should have an impedance of say 1/2 or less of this value at the lower roll off frequency. The formula for the impedance of a capacitor is 1/2piFC where C is the capacitance in farads and F the frequency.

If this capacitance turns out to be 4.7uF or less use a film capacitor else aim for an impedance of 1/5 of the input impedance and use a bipolar electrolytic capacitor. The rationale here is that a low impedance behaves like a short circuit and even a non linear short circuit is still a short circuit. I don't believe that electrolytic capacitors are as bad as people claim but they do dry out over time.
 
Metal: Yeah this whole impedance thing has been both the reason I was slow in getting transformers for isolation, and now, the reason I've no idea how to choose a capacitor value for DC isolation. Now I've tested my circuit with no capacitor with the mixer I'm intending to use to drive it, and I think I'm safe without it. But really, this whole impedance things is turning into a real mystery for me.

For one thing, one person in this thread replied that with a 1:1 turns ration, impedance is "reflected" through a transformer. Well I know the point of all this is to interface a line level output from my (known) mixer, to the line level input of another (unknown) mixer. I chose some fairly hefty transformers (1/4 watt rating), which happen to be 10K, and 1:1. (OK, they are center tapped, so I could theoretically make them 5K 1:1). But what is the output impedance of my mixer? It doesn't say in the specs. And as to the input Z of the mixer I'll plug into, remember the point of all this is ground loop isolation in a performance situation, which means I'll never have a clue what I'm interfacing to.

I do know that for signal level (not power) audio equipment, the output Z is supposed to be very low, and input Z is usually high. But who knows... The sound man may often pull out DI boxes when they see our setup, just to have a convenient way of connecting our line level outputs to low Z microphone inputs, and those DI boxes have their own transformers with (once again unknown) impedance.

Bottom line, I've made my isolator box with switches to put the transformers in or out of the circuit, and I just have to hope for the best. I've tried it with the output side connected to a variety of amplifiers accepting line level input, and they all seem to sound the same with or without the transformer. So that's a good result. Did I make a wrong choice buying 10K 1:1 transformers? Should I worry about a capacitor with less than 1mV DC offset? I don't know anymore. It all seems to sound good and nothing gets warm :). I WOULD like to better understand how to have better planned on the correct impedance, but it seems like there are too many unknowns in a situation like this, and so I'll just have to cross my fingers and hope this works for most situations I run into.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.