GFCI to break ground loop

[rhollan]

This is more of a safety question, rather than a code question. In fact, I doubt the CEC (I'm currently in Canada), or NEC would apply because of the manner in which I solved my problem. Still, I'd appreciate comments regarding the safety of my solution.



I had a nasty ground loop between satellite dish ground and AC safety ground third pin grounding of a subwoofer with a plate amp. Unlike all other home entertainment devices, which are double insulated and thus do not require (nor have) a three pronged AC power cord, the subwoofer uses an "plate amp" with exposed metallic chassis back and a three-pronged AC plug.



In Ontario, unlike the rest of Canada and the U.S. there is no requirement to ground an antenna to a ground electrode, bonded to the rest of the building's grounding electrode system: Bonding to a cold-water entry pipe within the building within a certain distance of entry is sufficient. I think the reasoning is that the purpose of the ground is to discharge static rather than offer lightening safety, since even with a ground rod, a direct strike would still be devastating. I bonded mine to a point on such a pipe within two inches of where the house AC entry is also grounded (yes, code permits this as well here).



Alas, I still had a ground loop via the subwoofer AC ground. I considered breaking the loop by replacing the permanent AC outlet supplying power with a GFCI, and not grounding the GFCI, but this is only acceptable if there is no ground wire to begin with. Instead, I fashioned an adapter, with a 6 foot cord terminated in a two pronged AC plug, to a non-metalic single-gang box, where I installed a GFCI, ungrounded (obviously). I also applied one of the "not grounded" stickers supplied with the GFCI to this box. Using this "extention cord", I can break the ground loop to the subwoofer with no permanent wiring change. I figure that if a short to the plate amp chassis were to occur, the GFCI would trip.



Naturally, the AC cord for this extention is rated to the current capacity of any outlet in which it can be plugged (as is the GFCI, by design). The trip limit on the GFCI is 4-6 mA. With the two pronged plug and clear labelling of the GFCI as ungrounded, I can't envision ever confusing this as an extention that preserves a safety ground.



So, just how bad a boy have I been?

 

[jimwalker]

Re: GFCI to break ground loop

Is this self built adapter UL listed ?
I would be just as happy to have seen a gfci recepticle and you use one of them 3 to 2 prong adapters to plug the amp in.Neither one is really code compliant.

 

[hurk27]

Re: GFCI to break ground loop

Go to your room and don't come out until you do it the right way
All kidding aside if there is a ground loop and the equipment groung is still intact from the equipment to the GFCI the GFCI will still trip. And if you have removed the equipment ground from the equipment then you have introduce a problem as there is no grounding conductor to reference to yes it will still trip the GFCI if someone were to get in between the case of the equipment and something grounded but if there is a short in the equipment that places the case at 120 volt potential the GFCI will not trip as it can not see the case with out the grounding conductor. I do professional PA system and ground loops are a big problem but correcting for one in a safe manner is not hard in fact you can go to any auto sound place you know where they sell those big power amps that make those cars kids drive go boom down the road. there they have a little drvice already with RCA jacks on it that will isolate the input to the amp the reason you are hearing it from the sub amp is because of the frequency of the hum (60hz) it is coming from the hook up of the VCR or TV if connected by low level inputs I would say to break the loop at the first point the cable connects to the system then isolate between each component after that with one of these ground loop isolators this is what there made for. Some electronic supply house's will have a 75 ohm to 75 ohm isolator too and even your cable company might have them as this is a big problem when adding amps to a video system that is getting a signal from a grounded cable system. if you can get them the 75 ohm isolator is the way to go. It is never a good idea to remove any type of grounding.

As far as grounding of the cable,sat, or ant system It is required by the NEC and should always be made outside the structure NOT INSIDE it has to be connected to the main service grounding electrode. never invite lightning into a building as this is asking for a fire or worse. You say a system cant be protected from a lightning strike, Yes it can or be kept to a very minimum. I have five computers that are online 24/7 and I have a tower that is about 125' that has been struck countless times (I know when because of the flash bulbs going off as an indcator)never had any damage. so there is methods that can bring the damage down to a minimum but it takes a lot of knowledge and understanding of what lightning is looking for. Here is just one good place to start to get info on lightning protection there are others too:

Lightning protection Institute/com

 

[rhollan]

Re: GFCI to break ground loop

jimwalker: the GFCI is CSA listed (remember, I'm in Canada). It's probably UL as well, though CSA matters here. I thought of changing the recepticle in the wall and using a "widowmaker 3 prong to 2 prong adapter", but that would be less code-complient, since I'd be replacing a grounded recepticle with a GFCI. That's why I chose the "outboard" route.

hurk27: No. The plate amp chassis is bonded to neutral. If the hot lead shorted to it, the subwoofer fuse would blow and/or house breaker would trip. For the scenario you describe, the hot lead would have to come loose touching the chassis AND the neutral would have to be broken (exactly what a safety ground protects against).

In that case, yeah, the plate amp chassis will be hot until something (someone) provides a ground path, if the hot lead shorts to it, lacking an existing ground path. But, a ground path does exist there: the shield of the audio cables. The issue is whether the GFCI would trip (it's a class A device, designed to trip at 5 mA) before the current causes heating of the shield cables and cause a fire hazard. My guess, obviously, is "yes, the GFCI will trip before the place catches fire".

As for audio isolators, I admit cheapness: good ones are bloody expensive. Remember, we're taking about driving a subwoofer, and so need response down to about 10 Hz. Worse, the subwoofer is a feedthrough model, with integrated high/low pass filters, so response of any audio transformer should be flat to beyond 20 kHz, with little phase shift. Such transformers cost around $500.

I see two obvious alternatives: build a circuit to isolate the ground reference at the satellite multiswitch (which is tricky: unlike a cable feed, there's 14/18V phantom power for the LNB and 22 kHz signalling to the multiswitch); or track down the bloody ground loop like I should have

It's odd, because the house ground and satellite ground are bonded to the same point (a cold water pipe, legal here (?!?), within two inches of eachother. I can only assume that I'm seeing an inductive loop, with 60Hz induced on the ground conductors from satellite to receiver to subwoofer to house ground to water pipe to satellite.

 

[rhollan]

Re: GFCI to break ground loop

hurk27: Yes, I am aware of antenna grounding requirements per NEC and CEC: I had to redo an improper satellite antenna ground when living in Texas.

I found it incredible, but Ontario does not require an external ground rod, bonded to the grounding electrode system (#6 wire, yada yada yada, been there, done that). No licensed electician here will do it either, for liability reasons. You ARE required to bond to the cold water entry (for static discharge purposes). Yes, this is almost exactly opposite to what the NEC dictates. Yes, I think the NEC has it right. Yes, I'd get into trouble if I did that here.

With satellite antennae, the reasoning is as follows: they are generally mounted to the roof/side of the building, and any lightening strike isn't going to give a fig about a wimpy #10 or #6 wire going straight down: you WILL get a roof fire anyway. With antennae remote from the building, of course, it is a different story, and, as the only risk to the building is the strike entering on a leadin, it makes sense to ground well at the tower.

 

[hurk27]

Re: GFCI to break ground loop

rhollan: As for audio isolators, I admit cheapness: good ones are bloody expensive. Remember, we're taking about driving a subwoofer, and so need response down to about 10 Hz. Worse, the subwoofer is a feedthrough model, with integrated high/low pass filters, so response of any audio transformer should be flat to beyond 20 kHz, with little phase shift. Such transformers cost around $500.

I have used 1 to 1 isolation transformers in many PA system's and even ones with LF servo's that when down to 5 hz. and never had a problem with freq. responce. useing a spectrum analizer on it with white noise for signal it was flat I/O this was with the cheep one from radio shack but I was thinking the loop was from your cable as for a sat system you might be getting the loop because the main amp being powered from one phase and the sub amp being powered from the other phase check this you might be able to just move a breaker around in the panel to get them on the same phase. always try to power audio equipment from th esame circuit if possable. this would be true only for the sat reciver that is feeding audio for the amps as the others would be isolated by there LV power supplys. Boot leg grounds will do this also it is possable that there is a neutral/ground connection in one of the receptacles on this circuit. Just some things to check.

[ January 04, 2004, 09:16 PM: Message edited by: hurk27 ]

 

[rhollan]

Re: GFCI to break ground loop

hurk27: On the audio 1:1 transformer, the specs for the Radio Shack ones are terrible. I suppose they're adequate for P/A systems, but am surprised that you see decent response down to 5 Hz.

With regard to the ground loop, yes, I've tried all the obvious things, like powering all the audio and video gear from a common outlet. Lifting the satellite antenna ground OR the power ground on the subwoofer plate amp (temporarily, with a "widowmaker" 3 prong to 2 prong adapter) made the problem go away.

Unlike a Cable TV feed, satellite antennas are essentially low noise block (LNB) amplifiers and RF downconverters (10 GHz to 1.0 Ghz with 1.0 Ghz bandwidth) coupled to an RF cavity at the focal point of a parabolic reflecter. This amp requires power, of course, and phantom power is supplied via the same coax that receives the downconverted signal. Switching between left and right polarizations is accomplished by changing the supply voltage from 14V to 18V as appropriate. The phantom power is supplied either from the multiswitch, or satellite receiver (multiswitches are either powered or unpowered, and the unpowered ones steal some of the LNB power). Because of the need for phantom power on the coax, the solution that works for a Cable TV feed (an RF isolation transformer) will not work here. To complicate things, there are two means of selecting between satellites in a multisatellite system (like mine): 22 KHz (or not) superimposed on the LNB power, or 22 KHz modulated with a digital code (DiSeq). This is interpreted by the multiswitch and can also be passed on the the LNB in the case of DiSeq: DiSeq also allows for rotor rotation and is a bidirectional protocol. So, RF isolation transformers are out. It would have been my preferred solution.

Another alternative is to use TWO multiswitches, and 2.2 Ghz-rated RF splitters ahead of them and the LNBs. Such splitters are available that support power passing on one port only. This allows one multiswitch to act as a master, selecting the appropriately polarized signal from each LNB, and the other as a slave. Since the slave does not need to power the LNB, RF isolation transformers could be employed. Unfortunately, such transformers generally roll off around 1.0 GHz and are designed for Cable TV use. An alternative would be a pair of high-quality RF capacitors in series with signal and shileld lines between the splitters and slave multiswitch. The only problem with this approach is the 3.5db loss the splitters introduce.

 

[hurk27]

Re: GFCI to break ground loop

I just cant see this coming from a 16 volt DC phantom power as the output phantom power to the coax is isolated from the 120 supply. and the fact that it is DC? and lifting the ground on the sub amp does remove the hum then it has to be a loop between the sub amp and the main amp or you would see the hum in the main amp wouldn't you? I wondering how did you run your speaker wires not that there subject to picking up this hum from power conductors but is there a shield on them and is it bonded to the sub amp? I had a case where someone did this at a roller skating rink one time and they bonded both ends of the shield which produce the loop to the amps ground. shields only need to be bonded at the input location.

The plate amp chassis is bonded to neutral

I just saw this and this would cause this ground loop as if the chassis is bonded to the neutral and the EGC in the cord you have created a ground loop just like a boot leg connection in a receptacle these two can not be connected on the load end is this the way it was manufactured? if it was it is in error. You call this a plate amp. are you referring to it because it has tubes or FET output? I have never heard a amp called this except in tech language. When cable came out this was a problem with older sets that used the chassis as part of the X-Ray shielding if someone unknowingly replaced the cord set in reverse it would place 120 volts on the cable and or cause a fire. or at least a bad shock hazard. since then there has never been allowed to bond the neutral to equipment chassis and the fact that the supply cord has a grounding conductor too would violate all kinds of codes. and cause the ground loop your looking for.

 

[rhollan]

Re: GFCI to break ground loop

hurk27: The pantom power for the LNB has nothing to do with the ground loop, except that it means that I can't put an RF isolation transformer inline.

The term "plate amp" in subwoofer parlance means an amplifier built on a rectangular, open frame, metal chassis. This is mounted over a rectangular hole cut in the subwoofer, with the electronics end inside. The "bottom" of the plate contains ports for inputs, power, switches, and indicator lights. It is called a "plate amp" because it lookes like a piece of "plate" metal. All solid state, No tubes.

The ground loop is not present if I take the subwoofer out of the loop and drive the main speaker amp directly (the subwoofer has line level inputs and outputs). The problem is definately 60Hz riding the signal ground input to the subwoofer.

As this is consumer equipment, audio lines are unbalanced: signal and ground, and not balanced: differential signal and ground. Thus, the signal ground IS the interconnect cable shield. I agree that a balanced audio signal would permit me to lift the shield at one end of the interconnect.

I'll have to double check about neutral being grounded in the subwoofer plate amp, but I am fairly sure this is the case. Even if not, a path to ground would exist through the audio signal interconnect shield or a body sufficient to trip the GFCI if hot came into contact with the case. Yes, this means that the case would become energised if there was no alternate path to ground until some unfortunate person provided one, but this is no worse that just plain GFCI protected equipment.

 

[dereckbc]

Re: GFCI to break ground loop

Originally posted by rhollan:
No. The plate amp chassis is bonded to neutral. If the hot lead shorted to it, the subwoofer fuse would blow and/or house breaker would trip.

If the neutral is shorted to the chassis you have a code problem, and you probable found you noise problem

 

[hbiss]

Re: GFCI to break ground loop

The term "plate amp" in subwoofer parlance means an amplifier built on a rectangular, open frame, metal chassis. This is mounted over a rectangular hole cut in the subwoofer, with the electronics end inside. The "bottom" of the plate contains ports for inputs, power, switches, and indicator lights.

In real world "parlance" that would be a powered sub woofer. Can't tell you how much I hate audiophile BS and hype.

 

[rhollan]

Re: GFCI to break ground loop

hbiss wrote: In real world "parlance" that would be a powered sub woofer.

Well, yes, except what matters in this contrext is the active electronics, and not the fact that it is integrated into a subwoofer, and amplifiers designed to be integrated into subwoofers are called "plate amplifiers".

 

[hurk27]

Re: GFCI to break ground loop

By rhollan:
but this is no worse that just plain GFCI protected equipment.

Except one thing. A GFCI is only there for emergency and is not to be depended on to protect every one as it wont. a person touching this amp will still receive a shock and can be killed. Look on any GFCI receptacle box and there will be a warning that this device will not protect every one there are people that might have a bad heart or other medical problems that could cause death even with the smallest shock. Now since you have installed a non code compliant wiring you can be held accountable. It was not long ago that an electrician miss wired a heater and burned down a house killing a 12 year old boy and he got 5 years in prison for it. Now how expensive is that isolation transformer? is it worth risking any ones life including your's? This is why I was trying to help you find the loop.
This is why I would never lift a ground in a professional PA system as if anyone ever got hurt not only would it be my butt but I would not be able to live with the fact that what I did caused someone harm.

 

[hbiss]

Re: GFCI to break ground loop

Well, yes, except what matters in this contrext is the active electronics, and not the fact that it is integrated into a subwoofer, and amplifiers designed to be integrated into subwoofers are called "plate amplifiers".

Yeah, OK. I just did a little reasearch to "enlighten" myself. Your statement implies that the amp is somehow acoustically or electrically coupled to the the driver other than just the normal voice coil connection. I see no evidence of that. The amp just hangs out on the back of the box. True, that would keep the voice coil leads as short as possible and help damping, but that's about it. I don't even see the amp being specifically designed for a specific subwoofer, they are available as a DIY retrofit for use with any speaker. Like I said, it makes a powered speaker although I suppose when you call it a plate amplifier you can get something like $745.00 for it! :roll: .

 

[rhollan]

Re: GFCI to break ground loop

hbiss wrote: "Your statement implies that the amp is somehow acoustically or electrically coupled to the the driver other than just the normal voice coil connection."

It does? That wasn't the intention. Amps designed to be used for subwoofers are commonly called "plate amps" because of their shape. They aren't particularly high-fidelity devices either, since they only have to be flat from about 10 to 200 Hz. Indeed, there is a whole controversial school of thought about multiamplification being better than wideband (if 20 Hz to 20 KHz could be considered wide)single amplification in stereo systems. Mine cost around US$200 and the whole custom-built sub was less than US$1000. Hardly rarefied HiFi. Beats the pants off a (IMHO, junky) Sunfire at a similar price and a (half-decent) Velodyne at double the price, though. Serious "audio nuts" don't even really dabble in separate subwoofers. (My old two-way speakers crossed ribbon drivers to integrated 12" open-frame woofers -- three a side -- took much power to drive them but I had nice bass extention to 18 Hz).

But this is digressing from the safety issue.

hurk27: your concern is noted. I intentionally designed the GFCI workaround as a separate "extention" cord and DID NOT rewire the permanent AC outlet with an ungrounded GFCI.

Also, I "oopsed" when I reported the chassis of the plate amp was bonded to AC neutral -- it shouldn't be and isn't. So, if there is no path to ground (i.e. no line-level signals connected), and the hot lead shorts to the chassis, one can receive a shock until and if the GFCI kicks in if their body has a path to ground. Also, if the GFCI does not kick in, there is a potential fire hazard in the return path through the signal lead shield. Given a 5 ma GFCI trip point, this is unlikely, though fusing them might be a good idea.

Finally, the subwoofer is located in a place without the means to readily ground a body: the chassis of other equipment is doubly insulated and ungrounded, and the location is a living room, not bathroom or exterior location where GFCIs are usually called for. So, even if the chassis became electrified, the risk of a path to ground via a body is remote.

I would still like to find the source of the ground loop though, and rid myself of this hack, but all attempts have been futile.

hbiss: "I see no evidence of that."

Er, perhaps because there is none...?

 

[hurk27]

Re: GFCI to break ground loop

rhollan
1. Does the main amp also have a hum when you connect the sub or does the hum only come out of the sub?
2. If it only is herd through the sub amp then there is AC current on the shielding between these two amps, which can be verified with a volt meter. Check the power cord wide blade (neutral) to the chassis ground or the shield connection on the main amp to see if there is any shield leakage to it and this is also a common problem if a RFI filter cap has shorted out in the main amp.
3. Do you loose the hum if you disconnect all the audio/video cables from any video source(TV,VCR,DVD) that also has a 75 ohm connection to the sat system?

Any component in this daisy chain could be the one causing the loop and each one has to be tested one at a time till you get to the one that does cause the hum but if it is coming in through the sat system or just the cable it self then any component connected to this cable will produce this hum.

As of now I'm betting that the problem is in the sub amp as it is it's ground disconnection that removes the loop. so I would do a continuity test between the shield on the audio input and the neutral prong. if this is not the case then I would suspect a neutral/ground connection in the circuit that this amp is plugged into. again this would require a continuity test after lifting the neutral and ground at the panel (with the circuit turned off of course) What causes the hum is AC current on the shielding between two pieces of audio equipment. Why we call it a ground loop when it is cause by the neutral/ground connection is beyoned me . and the only thing that can cause this current is a ground/neutral connection ether in the equipment itself or in the circuits feeding them. The reason this happens to cable TV is because the cable is bonded at the pole and at the house and it is in parallel with the neutral drop to the house. this will cause some of the neutral current to be on the cable and when we hook up video equipment that has this cable connection to any audio equipment that has a chassie ground in the power cord, you now have your loop. And the amplifier will just do what it is supposed to do amplify it. and they do it real good at 60 Hz. specially when you have S181 MT series EV's loaded with EVX's 18's speakers, powered by a crown micro tech 3600 watt amp

[ January 06, 2004, 12:36 AM: Message edited by: hurk27 ]

 

[rhollan]

Re: GFCI to break ground loop

hurk27: I believe I went through all the tests you suggest and am 99.9% certain the loop is between the satellite feed and sub ground pin (lifiting either makes the hum disappear).

But, since you've taken the time to suggest the obvious, I WILL go and check all these excellent points AGAIN - I am not beyond believing that I may have missed something and 99.9% certainty is not 100%.

I will verify later today, but here's what I know:

1) Only ground points are sat antenna feed (bonded to RF shield at multiswitch and subwoofer ground prong.

2) Signal path is sat LNB to sat multiswitch to sat receiver to TV to (via TC monitor out) audio receiver AUX to subwoofer to main amp to main speakers.

3) Removing the sub from the signal chain and driving the amp from the audio receiver does not result in hum -- adding the sub and tuning the passthrough low will result in hum from the sub and mains speakers.

4) Signal shield is bonded to ground at the sub.

 

[hbiss]

Re: GFCI to break ground loop

Ok, rant aside, let me see if I can make some sense here. I also have considerable experience in commercial sound and this really should not be as bad as you think.
Just the facts:
1) Everything except the sub woofer amp has 2 wire cordsets.
2) The sat dish is bonded to ground at the same point (basically) on the water line as the service.
3) Everything is interconnected by either coax or unbalanced shielded cable.
4)Lift the ground pin on the subwoofer amp or disconnect the sat dish ground and your problem goes away.

Sounds to me like there is a voltage difference between the neutral bar in the panel (I will assume that this is a main service panel, not a sub panel) that feeds the receptacle that powers the subwoofer and the point on the water pipe where you attached your sat ground bond. Your equipment is essentially in parallel with the grounding electrode conductor that runs between the neutral bar in the panel and the water pipe. There's your ground loop. It is not unusual to have a few amps flowing through the grounding electrode conductor and with your system audio wiring in parallel with it the difference in potential will be impressed on the audio.

What to do? Since this is "consumer audio" with unbalanced in and outs you can't lift the shields at one end going to the subwoofer. The only other thing you can do is eliminate the ground potential by grounding the subwoofer amp line cord ground pin at the same place as the sat ground bond. What I would do is run a dedicated circuit for the subwoofer amp, I suppose this could also power any of the other equipment too as long as they are 2- wire cords. I would make this an IG (isolated ground) circuit with IG receptacles. At the panel I would bring the IG out and bond it with the sat ground on the water pipe.

Anybody have a problem with this?

 

[dereckbc]

Re: GFCI to break ground loop

Originally posted by hbiss:
Anybody have a problem with this?

Not really, other than what you proposed doesn't sound like it meets code, and just rather expensive. Also if this is a wood frame home, every circuit is already IGR, so why pay again for it? The poster stated "the neutral is bonded to the chassis of the sub-woofer amp". With the ground pin connected, load current will flow in the EGC creating a voltage drop that is picked up and amplified. This is not in compliance with UL or NEC, and makes since why there would be noise.

I would take two or three actions.

1. Open the neutral-to-ground bond in the sub-woofer amp.

2. Re-establish EGC circuit in sub-woofer.

3. If 1 & 2 didn't correct the noise. I would then use a surge reference equalizer device (surge arrestor available by many manufactures) to supply power to all the A/V equipment that is equipped with I/O ports (coax, etc). That would eliminate ground potential problems between equipment, rather than run a new IGR. Or in other words create a pseudo "single point ground" That would eliminate any difference in ground potentials between sat/receiver and sub-woofer amps

[ January 06, 2004, 04:05 PM: Message edited by: dereckbc ]

 

[hbiss]

Re: GFCI to break ground loop

The poster stated "the neutral is bonded to the chassis of the sub-woofer amp".

I seriously doubt that he is correct in that assumption.

...I would then use a surge reference equalizer device

You got me there. You are not talking about those isolated power systems that use a center tap on an isolation transformer 120v secondary to provide a ground are you? If not, I'm not familiar with what you are talking about.