ProSoundWeb Community

Sound Reinforcement - Forums for Live Sound Professionals - Your Displayed Name Must Be Your Real Full Name To Post In The Live Sound Forums => AC Power and Grounding => Topic started by: Mike Sokol on June 05, 2014, 01:35:34 PM

Title: GFCI leakage testing
Post by: Mike Sokol on June 05, 2014, 01:35:34 PM
Eric says his gear works "everywhere" else-does that include other venues with GFCIs or not?  If the same gear, hooked up the same way trips one GFCI but not another I would suspect a bad GFCI-though the ba d one could be the one that does not trip!  On the other hand, if they have a bad GFCI. you'd think they would have other complaints and would hopefully have found the problem.

The first thing to realize is that ALL gear plugged into an electrical outlet has a certain amount of allowed leakage to its chassis and ultimately ground. For "ungrounded" appliances such as your iPhone charger, there can be a max of 0.75 mA of leakage and still meet UL standards. For "grounded" appliances such as your mixing console, backline gear and power amps, hot-to-chassis leakage can be as much as 3.5 mA and still meet UL approval.   

Secondly, these leakage currents are additive, so it's possible to plug two pieces of audio gear with 3 mA of leakage each into a single GFCI and cause it to trip. As previously mentioned, "Surge Strips" are a big source of MOV leakage, so plugging two of them into a common GFCI circuit can cause a trip, even of each MOV strip is technically in compliance. And old tube guitar amps develop a lot of power transformer leakage which can easily cause a GFCI to trip.

So what to do? Well this may be the time to begin measuring hot-to-chassis leakage currents of your own gear. Most of it will probably be around 1 mA or so, and that implies that 6 pieces on the same GFCI will cause a trip. Of course, anything with 4 or 5 mA leakage by itself is going to cause a second piece of gear to reach the threshold and trip.

For a primer on GFCI theory see here: http://www.noshockzone.org/rv-electrical-safety-part-viii-gfci/

And here's something I wrote about GFCI troubleshooting: http://www.noshockzone.org/rv-electrical-safety-part-x-%E2%80%93-gfci-troubleshooting/

As GFCIs show up at more venues, we're going to need to confirm our own gear leakage. If not, then we'll be stuck with random tripping that makes nobody happy.
 
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 05, 2014, 02:14:04 PM
FYI: I'm going to ask Klein Tools to send me a demo Klein CL2000 clamp meter for a ground leakage demonstration I'm planning. They also make a CL1000 clamp meter which costs around $100 and has 0.1 mA resolution (according to their literature). So I'll also ask them to send a couple CL1000 clamp meters I can send to a few of you on the forum willing to do some ground-leakage experiments with audio gear and report back to the group. Yes, you'll have to send them back to me eventually, but I'm trying to develop a simple field leakage test that will explain and help eliminate random GFCI tripping on stages. If I can find a few participants on the forum and Klein agrees to the loaner gear, then we'll all learn a little something about why these GFCI trips occur and what we can do about it.

PM me if you want to be part of this test. See http://www.kleintools.com/catalog/clamp-meters/400a-ac-clamp-meter for the meter specs.
Title: Re: GFCI leakage testing
Post by: Kevin Graf on June 05, 2014, 02:22:17 PM
Also remember that many so called 'power conditioners' dump the noise to the Safety Ground.
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 05, 2014, 02:38:00 PM
Also remember that many so called 'power conditioners' dump the noise to the Safety Ground.

Yup, they are not a friend to GFCIs.
Title: Re: GFCI leakage testing
Post by: Stephen Swaffer on June 05, 2014, 08:09:58 PM
Mike, not to rain on your parade, but you may be disappointed with the results.  I picked up a Klein CL2000 a few months ago, and would highly recommend it for anyone doing audio work  Very versatile, built in NCV the whole works-if I had to take only one of my meters to a job or gig this would be the one.

That said, experience has taught me to disregard fractional amps on most meters and  suspected it was a transducer limitation more than anything.  I put my CL2000 on a rock solid amp draw lighting circuit this afternoon, ,and I could consistently get a variation of 200 mA just by moving the wire in the jaws.

If you check out the Klein manual, my observation is in line with the 40 amp AC range specs.  The low amp range is much more accurate-but it also requires inserting the meter into the circuit via test leads.
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 05, 2014, 08:26:13 PM
Mike, not to rain on your parade, but you may be disappointed with the results.  I picked up a Klein CL2000 a few months ago, and would highly recommend it for anyone doing audio work  Very versatile, built in NCV the whole works-if I had to take only one of my meters to a job or gig this would be the one.

That said, experience has taught me to disregard fractional amps on most meters and  suspected it was a transducer limitation more than anything.  I put my CL2000 on a rock solid amp draw lighting circuit this afternoon, ,and I could consistently get a variation of 200 mA just by moving the wire in the jaws.

If you check out the Klein manual, my observation is in line with the 40 amp AC range specs.  The low amp range is much more accurate-but it also requires inserting the meter into the circuit via test leads.

It will be interesting to find out how repeatable the milli-amp range turns out to be on these inexpensive clamp meters. The Fluke meter and external ammeter clamp I pictured earlier is too expensive for field work, about $1,000 or so. Of course, a lot of digital voltmeters have a 10 amp current range, which may be what's required for this test. However, I really don't want to break the EGC on a test power cable if possible. It would be better if this was a completely benign test, but we shall see how it all works out. I'll talk to Klein engineering to see what they think.
Title: Re: GFCI leakage testing
Post by: Stephen Swaffer on June 05, 2014, 09:44:47 PM
Its my understanding that the mA range only functions as wired, not with the clamp-but I haven't had a chance to experiment.  It does sound like an interesting experiment-maybe I need t become a professor so I have time to experiment!

The other way to run a test would be to make a test jig with a low resistance precision resistor in the EGC, then use a mV range to measure the voltage across the resistor and derive the current from that measurement.  Really the same thing, but perhaps less chance of a variation in connection resistance form test to test.
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 05, 2014, 10:10:07 PM
It does sound like an interesting experiment-maybe I need to become a professor so I have time to experiment!

Hey, I've been running electrical experiments since I was 6 years old. I didn't become a professor for another 45 years or so.  ;D
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 05, 2014, 10:36:16 PM
The other way to run a test would be to make a test jig with a low resistance precision resistor in the EGC, then use a mV range to measure the voltage across the resistor and derive the current from that measurement. 

There's a much easier way. If you create multiple loops in the ground wire to clamp the meter through, you'll make a current multiplier. So just 10 loops would increase the resolution of a standard 10 mA minimum clamp meter to 1 mA. Cost would be next to nothing and this would allow everyone to use their existing clamp meters for the leakage test.

Any thoughts?
Title: Re: GFCI leakage testing
Post by: Kevin Graf on June 06, 2014, 09:34:42 AM
Any thoughts?

a] Many DMM have AC Current ranges.  My 30+ year old bench meter has a 200 uA full scale range, my Radio Shack DMM may go even lower.  So you loop the wire through the DMM and the clamp ohmmeter at the same time.

b] Trying to make accurate reading using the AC power line has it's frustrations.  First the line voltage bounces all around, second the high frequency noise may be read differently with different meters.  So instead of the AC line, use a big power amplifier.  Use a music player with a 60 Hz sine wave signal.  It will be safer and more accurate.
Title: Re: GFCI leakage testing
Post by: TJ (Tom) Cornish on June 06, 2014, 09:56:43 AM
Any thoughts?

a] Many DMM have AC Current ranges.  My 30+ year old bench meter has a 200 uA full scale range, my Radio Shack DMM may go even lower.  So you loop the wire through the DMM and the clamp ohmmeter at the same time.

b] Trying to make accurate reading using the AC power line has it's frustrations.  First the line voltage bounces all around, second the high frequency noise may be read differently with different meters.  So instead of the AC line, use a big power amplifier.  Use a music player with a 60 Hz sine wave signal.  It will be safer and more accurate.
All of these techniques have their place.  I have a .1 ohm 100w shunt resistor that is great for high-current applications, and being a 1% resistor, is plenty accurate enough:
http://www.digikey.com/product-search/en?pv1=1323&pv2=14&FV=fff40001%2Cfff80488&k=shunt+resistor&mnonly=0&newproducts=0&ColumnSort=0&page=1&quantity=0&ptm=0&fid=0&pageSize=25

Whether this is easier than looping the ground wire around a clamp meter probably depends on the circumstances.  The advantage of the shunt resistor is you can make an enclosure with the resistor and a receptacle with measurement test points via banana plugs, and you don't have to cut up a cord.  The disadvantage is that you potentially need more than one shunt resistor in your jig, depending on what you're trying to measure - one on the hot lead, one on ground, etc.  The hot lead shunt resistor will also be at line voltage potential, so care in measurement is necessary.

A clamp meter requires a cord with the wires separated, and can become a little bit messy looking if you're wrapping multiple turns (you can't over-under your turns or you cancel the current), and of course requires a clamp meter.

Using the current range of your multi-meter (which is really a voltage measurement over a current shunt) is potentially convenient, but there are limits to the range of current that can be measured - most meters are 10A on the high range, and a couple hundred mA on the low range.  Since most meters are banana plugs, some kind of test adapter must be made, and it can be easy to damage the meter with overcurrent and/or plugging into the wrong banana jacks.
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 06, 2014, 10:08:14 AM
a] Many DMM have AC Current ranges.  My 30+ year old bench meter has a 200 uA full scale range, my Radio Shack DMM may go even lower.  So you loop the wire through the DMM and the clamp ohmmeter at the same time.

Not a bad idea. I've thought about using one or the other, but certainly could use both for this test. I worry about the legal aspects of opening up the EGC to insert a DMM during a field test, but perhaps I'm being over cautious. I would need to make a simple break-out box with banana jacks to connect the meter inline with the EGC. There could be a shorting strap for when you don't have a DMM in the circuit and a 10-loop coil for the clamp meter to boost sensitivity of milliamp range. 

Quote
b] Trying to make accurate reading using the AC power line has it's frustrations.  First the line voltage bounces all around, second the high frequency noise may be read differently with different meters.  So instead of the AC line, use a big power amplifier.  Use a music player with a 60 Hz sine wave signal.  It will be safer and more accurate.

True for in the lab, but not practical for field testing. I want something you can throw on a backline quickly to troubleshoot GFCI tripping issues. This needs to be a quick and dirty test that doesn't take too long, but gives you a pretty definitive go/no-go answer. 
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 06, 2014, 10:19:29 AM
All of these techniques have their place.  I have a .1 ohm 100w shunt resistor that is great for high-current applications, and being a 1% resistor, is plenty accurate enough:
http://www.digikey.com/product-search/en?pv1=1323&pv2=14&FV=fff40001%2Cfff80488&k=shunt+resistor&mnonly=0&newproducts=0&ColumnSort=0&page=1&quantity=0&ptm=0&fid=0&pageSize=25

That could eliminate my worry about opening up the ground accidentally. I'll have to a few calculations to figure out the best resistor value to use that will produce sufficient voltage drop at 1 mA to show up on most meter AC voltage scales. Then it just needs a couple of binding posts for the meter leads, which will only carry voltage but no real current. This is only one-half of a Kelvin Bridge, so it won't give an accurate leakage reading at any incoming voltage other than 120-volts, but it's accuracy will only vary by the percentage that the line voltage deviates from 120-volts. Again, if I want PPM accuracy I'll do it on a lab bench. But for real world applications this could be a pretty good fit.   
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 06, 2014, 11:19:39 AM
I have a .1 ohm 100w shunt resistor that is great for high-current applications, and being a 1% resistor, is plenty accurate enough:
http://www.digikey.com/product-search/en?pv1=1323&pv2=14&FV=fff40001%2Cfff80488&k=shunt+resistor&mnonly=0&newproducts=0&ColumnSort=0&page=1&quantity=0&ptm=0&fid=0&pageSize=25

Inside my head calculations suggest this should be a 1 ohm resistor since a standard DMM goes down to 1 mV AC and 1 mV across a 1 ohm load equals 1 mA. So a standard accuracy DMM could read 1 mA of hot-to-chassis leakage, which is plenty close enough for this test. A better meter could measure sub-milliamp currents as well. Wattage for the resistor is a little more tricky since it could be possible to pass 20 amps (circuit breaker limit) and 120 volts through this resistor, which would result in 2,400 watts of heat. Of course, the meter would be off-scale at that point which is a pretty good indicator that something is terribly wrong. I'll look for a 100-watt/1-ohm resistor for this test. Most of the time there would be fractions of a watt resistor heating with normal leakage currents. It would only get interesting (read hot) if there was a line-to-chassis short happening during the test procedure. 
Title: Re: GFCI leakage testing
Post by: TJ (Tom) Cornish on June 06, 2014, 11:29:07 AM
Inside my head calculations suggest this should be a 1 ohm resistor since a standard DMM goes down to 1 mV AC and 1 mV across a 1 ohm load equals 1 mA. So a standard accuracy DMM could read 1 mA of hot-to-chassis leakage, which is plenty close enough for this test. A better meter could measure sub-milliamp currents as well. Wattage for the resistor is a little more tricky since it could be possible to pass 20 amps (circuit breaker limit) and 120 volts through this resistor, which would result in 2,400 watts of heat. Of course, the meter would be off-scale at that point which is a pretty good indicator that something is terribly wrong. I'll look for a 100-watt/1-ohm resistor for this test. Most of the time there would be fractions of a watt resistor heating with normal leakage currents. It would only get interesting (read hot) if there was a line-to-chassis short happening during the test procedure.
The 100 watt resistors are not available in anything higher than .1ohm - there is probably a smaller one.  Keep in mind that the resistor won't stand alone - at the very least, there is resistance in the wiring loop; and other than in a catastrophic short, this resistor is part of a normal circuit.  If we calculate from a 20A loop current, that gives a 2v drop over the .1ohm resistor, so the resistor should never dissipate more than 40 watts - if the loop current is truly 20A. 

I did this exercise when I built my test jig, and it seemed to me that I wouldn't have any trouble getting an accurate voltage measurement across a .1ohm resistor.

Disclaimer - I wrote this post in about 90 seconds - apologies if the math and/or logic is wrong.
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 06, 2014, 01:12:42 PM
I did this exercise when I built my test jig, and it seemed to me that I wouldn't have any trouble getting an accurate voltage measurement across a .1ohm resistor.

Disclaimer - I wrote this post in about 90 seconds - apologies if the math and/or logic is wrong.

While you can get an accurate reading of BIG current with a 0.1 ohm resistor, remember this needs to have readable voltage drop with 1 mA of current, yet survive a complete hot-to-chassis short and just enough series wire to allow 20 amps of fault current. Yes, I have to sit down, draw this out, and do the math, but I'm pretty sure a 1 ohm resistor is what's needed here. However, I've been wrong by an order of magnitude before and will be wrong again, so I'll do the math first.

I did a quick eBay search on Chinese resistors and found a 1 ohm, 100 watt resistor for less than $5 with free shipping. Not sure how accurate it is, but I have a serious Kelvin Bridge from my old missile guidance system building days that will measure down to 5 ppm accuracy at low ohms. I'll bet it's close enough for this field tester and the price is right. http://www.ebay.com/itm/1-Ohm-1R-100W-Watt-Power-Metal-Shell-Case-Wirewound-Resistor-/370550984569?pt=LH_DefaultDomain_0&hash=item56468f4f79
Title: Re: GFCI leakage testing
Post by: Kevin Graf on June 06, 2014, 01:40:51 PM
This is exactly what you need!

(http://www.electronicsurplus.com/ItemImages/149886-1.jpg)

Shunt 150 Amp 100Mv. | EMPRO HA150100
http://www.electronicsurplus.com/Item/149886/EMPRO%20-%20Shunt%20150%20Amp%20100Mv_%20-%20HA150100/

Electronic Surplus was in downtown Cleveland for decades.  EE's and other would stop by on lunch break.  But they moved to a far east suburb. Now you have to pack a lunch to go shopping.
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 06, 2014, 02:24:08 PM
This is exactly what you need!

Electronic Surplus was in downtown Cleveland for decades.  EE's and other would stop by on lunch break.  But they moved to a far east suburb. Now you have to pack a lunch to go shopping.

Wow, that is cool...! Might be a bit of overkill, but this looks like something you could use to bring Frankenstein's monster to life. I do love surplus stores. 
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 06, 2014, 02:46:55 PM
Here's an idea on how to keep the resistor from blowing up if something goes wrong during the test. What about a pair of husky diodes in reverse parallel across the sensing resistor? Once the voltage drop across the sensing resistor hits 1/2 of a volt, it will short out (essentially) and prevent any more current through the shunt resistor (which I still contend needs to be 1 ohm). So this would measure up to 500 mV across the shunt resistor which works out to 500 mA of measured ground leakage current. In that case, since the resistor could only ever see 500 mA of current times 500 mV of voltage, the max wattage it needs to dissipate would be 1/4 of a watt. Or I could use 5 volt Zener diodes instead which would allow a max linear reading of 5 amps leakage current producing 5 volts of sensing voltage, which works out to a 25 watt resistor required. I'm sort of liking that since there's no way it could overheat my 100-watt/1-ohm resistor, and it would allow more than sufficient leakage current test levels.   
Title: Re: GFCI leakage testing
Post by: Lyle Williams on June 06, 2014, 04:26:11 PM
My cheap earth current rig is just an AC milliammeter inserted directly in the earth line.  It is bypassed by two pairs to diodes that limit the voltage across the meter movement.

TWO diodes each way to get a high enough voltage limit across the meter.  EACH WAY to ensure that it can conduct on both sides of an AC cycle.  200A diodes in the hope that they won't blow before the fuse/breaker should a fault develop in the gear under test. The diodes were scrapped/surplus items.

Of course I can't use it to test anything that is bolted in a rack or has an alternate ground.  You need a differential current meter for that and they are a couple of grand from memory.

Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 06, 2014, 04:46:02 PM
TWO diodes each way to get a high enough voltage limit across the meter.  EACH WAY to ensure that it can conduct on both sides of an AC cycle.  200A diodes in the hope that they won't blow before the fuse/breaker should a fault develop in the gear under test. The diodes were scrapped/surplus items.

I think that a big old bridge rectifier with the DC outputs bonded together would give me a bi-directional 1-volt clamp. It's an unorthodox usage since the PIV rating would never come into play. Once there's around 1 V peak (or 3/4 of a volt RMS) voltage across the resistor, then it would shunt. So my top leakage current limit would be around 750 mA before the diode shunt kicks in.  I would trust something like this 100-amp 1600 PIV unit. http://www.ebay.com/itm/MDQ100-16-Single-Phase-Diode-Bridge-Rectifier-100A-Amp-1600V-LJN-/151294539123?pt=LH_DefaultDomain_2&hash=item2339db7573

or even this 50-amp/100-volt bridge rectifier could work. Remember, the reversed junction will never see more than the forward bias of the stacked P-N junctions, maybe 1 volt or so. http://www.ebay.com/itm/50-Amp-100-Volt-Bridge-Rectifier-MP15010-50-Amp-Full-Wave-Diode-Rectifier-/130905566376?pt=LH_DefaultDomain_0&hash=item1e7a946ca8

Also, something rated for 50 or 100 amps steady state should be able to absorb the limited energy of a shorted 20-amp circuit breaker on a typical branch circuit. I'm not hooking directly into cam-locks with this thing (yikes).
Title: Re: GFCI leakage testing
Post by: Lyle Williams on June 06, 2014, 05:03:17 PM
My cheap earth current rig is just an AC milliammeter inserted directly in the earth line.  It is bypassed by two pairs to diodes that limit the voltage across the meter movement.

TWO diodes each way to get a high enough voltage limit across the meter.  EACH WAY to ensure that it can conduct on both sides of an AC cycle.  200A diodes in the hope that they won't blow before the fuse/breaker should a fault develop in the gear under test. The diodes were scrapped/surplus items.

Of course I can't use it to test anything that is bolted in a rack or has an alternate ground.  You need a differential current meter for that and they are a couple of grand from memory.

Diodes less than 200A would probably work.  I just couldn't find any good data on the momentary current handling of rectifier diodes.  You need to take a momentary current possibly 20x greater than the breaker value.  230vac/10a is the common household circuit here, so finding surplus diodes with a their continuous = my required peak allowed a very conservative choice in ratings.

If you test in a way that assumes the device under test has a live chassis, then just directly inserting a meter into the ground line is ok.  That is actually one of the options in the portable appliance test standard I work to.


I also have a separate test jig that simply has a 100 ohm resistor in the earth line.  An AC voltmeter put across the resistor will show 0.1V per mA of earth leakage.  That jig has a RCD/GFCI plug on it where it connects to mains power.
Title: Re: GFCI leakage testing
Post by: Lyle Williams on June 06, 2014, 05:18:54 PM
I also have a separate test jig that simply has a 100 ohm resistor in the earth line.  An AC voltmeter put across the resistor will show 0.1V per mA of earth leakage.  That jig has a RCD/GFCI plug on it where it connects to mains power.
Title: Re: GFCI leakage testing
Post by: Stephen Swaffer on June 06, 2014, 09:15:53 PM
Unless you plan to routinely hammer your resistor with over voltage a simple 20 A fast blow fuse would give you reliable protection-if you are really concerned go with Buss's Lo peak fuses-they are designed to limit arc flash (I know OT) so they blow really quick-I would hazard a guess as to how fast but probably have to admit I was wrong.
Title: Re: GFCI leakage testing
Post by: Lyle Williams on June 06, 2014, 09:40:40 PM
Unless you plan to routinely hammer your resistor with over voltage a simple 20 A fast blow fuse would give you reliable protection-if you are really concerned go with Buss's Lo peak fuses-they are designed to limit arc flash (I know OT) so they blow really quick-I would hazard a guess as to how fast but probably have to admit I was wrong.

That would be fine for protecting the low ohm shunts suggested earlier.  My 100 ohm shunt would not trip a breaker/fuse if a chassis became live.  The 10w shunt resistor would then burn out under 500+w of load, leaving the chassis live.  Hence my RCD/GFCI.
Title: Re: GFCI leakage testing
Post by: John Roberts {JR} on June 06, 2014, 10:40:30 PM
Clamp diodes will work fine as you intended and you don't have to size them that much oversize. Diodes tend to fail as short circuits when they overheat and melt so the shorted diode will make even less heat than before, A few amp diode bridge will probably be more than adequate. You just want enough thermal mass that it doesn't vaporize from the initial surge.

I tested this back in the '70s when I was looking into alternate grounding schemes, and UL was receptive to accepting diodes for safety grounds if i paid them the $$ to set up a file and certify them (I didn't). 

JR
Title: Re: GFCI leakage testing
Post by: Lyle Williams on June 07, 2014, 05:37:56 PM
Thanks for the info JR.

Engineering has been defined as "doing for $1 what any fool can do for $2".    :-)

As I only wanted one test jig I took the $2 (well, probably $200, had the parts been new) path.

Title: Re: GFCI leakage testing
Post by: Stephen Swaffer on June 07, 2014, 07:16:56 PM
That would be fine for protecting the low ohm shunts suggested earlier.  My 100 ohm shunt would not trip a breaker/fuse if a chassis became live.  The 10w shunt resistor would then burn out under 500+w of load, leaving the chassis live.  Hence my RCD/GFCI.

Just keep in mind that if you place diodes across a 100 ohm resistor limiting the voltage to 1 volt your test range will max out at 10 mA.  Really not a problem as long as you keep in mind that any 10 mA reading might be much higher.

Also, a 100 ohm load will only draw 1.2 amps at 120 V, so it will not trip a breaker-but it will also create "only" 144 watts of heat-still too much for a 10 watt resistor though.  Not trying to be picky-just don't want someone to be confused if they are trying to do the math.
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 07, 2014, 08:59:40 PM
Just keep in mind that if you place diodes across a 100 ohm resistor limiting the voltage to 1 volt your test range will max out at 10 mA.  Really not a problem as long as you keep in mind that any 10 mA reading might be much higher.

Also, a 100 ohm load will only draw 1.2 amps at 120 V, so it will not trip a breaker-but it will also create "only" 144 watts of heat-still too much for a 10 watt resistor though.  Not trying to be picky-just don't want someone to be confused if they are trying to do the math.

Steve, I'm proposing a 1 ohm resistor which will output 1 mV per 1 mA of current. So the max current could be 120-amps under a dead short to the chassis (of course, the breaker would trip before that's reached steady-state, but a peak current around that is possible if the input line is stiff enough). Remember, I can get a 100-watt/1-ohm resistor for less than $6 with shipping, so I only have to clamp it around 5 volts or so for a max heating of 25 watts. With that setup I could measure up to 5 amps of leakage current (that's a lot more than required) which would output 5 Volt AC to my meter. With a 1-volt clamp, this would measure up to 1 amp leakage current with 1 volt output. Again, I'm looking for perhaps 10 or 20 mA, so even 1 amp (1,000 mA) is WAY over the expected values. I'm just trying to make this idiot proof in case something goes wrong (and yes, I admit to being an idiot sometimes).
Title: Re: GFCI leakage testing
Post by: Lyle Williams on June 08, 2014, 04:20:01 AM
Just keep in mind that if you place diodes across a 100 ohm resistor limiting the voltage to 1 volt your test range will max out at 10 mA.  Really not a problem as long as you keep in mind that any 10 mA reading might be much higher.

Also, a 100 ohm load will only draw 1.2 amps at 120 V, so it will not trip a breaker-but it will also create "only" 144 watts of heat-still too much for a 10 watt resistor though.  Not trying to be picky-just don't want someone to be confused if they are trying to do the math.

Sorry, I should have typed using my twangy 230v Australian accent.  :-) 

Under the standard I am required to test to, 5ma is a fail under all conditions and 1ma a fail under most.  Actually most stuff just gets meggered at >1Mohm to pass.  Gear without a mechanical on/off switch (gear that needs power connected for the power switch to work) needs a direct (cheap jig and multimeter) or differential (expensive gear) earth current measurement.

De-racking gear for testing with the cheap jig would be a pain, but testing earth leakage a whole roadcase at a time avoids the hassle.

Title: Re: GFCI leakage testing
Post by: Stephen Swaffer on June 08, 2014, 03:42:36 PM

Sorry, I should have typed using my twangy 230v Australian accent.  :-) 


I wondered if that were the case-but rest of the thread was talking about 120 V systems.  I run into to many people that work enough with electricity that they should know better, but they don't understand basic relationships.  Ohm's Law and power calculations are very simple, basic physics - we might as well encourage that understanding with a discussion about a real world test.

Also, I know no one else would do this, but it is very easy to overlook a side effect of a solution like clamping diodes.  Although it gives plenty of range for this test-next month I might try to measure something entirely different and grab my handy tester-only to get confusing results!

I agree that a 1 ohm resistor seems right for this test. 
Title: Re: GFCI leakage testing
Post by: Lyle Williams on June 08, 2014, 06:41:03 PM
For my no-shunt jig with the diodes, there is a side effect of the diodes worth mentioning. They can clip the ac waveform peaks.  The measurement limit with two diodes is less than 10ma.  Current measurements are only linear until the diodes start to clip the peaks.

A one ohm resistor is good as long as your voltmeter reads millivolts (or even microvolts) well.  My multimeter jumps around quite a bit at low levels.
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 09, 2014, 07:18:43 PM
A one ohm resistor is good as long as your voltmeter reads millivolts (or even microvolts) well.  My multimeter jumps around quite a bit at low levels.

I was messing with GFCI leakage measurment ideas today and revisited my idea of looping a wire multiple times through a clamp meters jaws to increase the resolution. Looking for a plastic spool for the loops I found this, a spool of 12-gauge magnet wire already on a plastic spool. I could simply drill a big hole in the center for the clamp jaws to fit through, and rewind it for a 10X or 100X multiplier depending on how many wraps you want to use. Solder on some green stranded THHN (I think) wire to run between the ground screws in a pair of male and female Edison plugs, and now you have a fractional milli-amp tester from any cheap clamp meter with 10 mA resolution you have laying around. What do you all think? 
Title: Re: GFCI leakage testing
Post by: Stephen Swaffer on June 09, 2014, 11:31:55 PM
How well the wire/loops is centered in the jaws affects accuracy.  I would suggest drilling a hole in a way that centers the coils as much as possible, and also limits movement to reduce variability. Then you could test for accuracy with a low voltage AC supply, a potentiometer, and an accurate mA wired in series with the loop.

Of course, the more coils you have the more inductive impedance you introduce to the circuit. Without some serious calculations, I have no idea if this even approaches the 1 ohm resistive impedance we had been discussing-that may very well more of a theoritical consideration than a practical one.
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 09, 2014, 11:48:27 PM
How well the wire/loops is centered in the jaws affects accuracy.  I would suggest drilling a hole in a way that centers the coils as much as possible, and also limits movement to reduce variability. Then you could test for accuracy with a low voltage AC supply, a potentiometer, and an accurate mA wired in series with the loop.

Of course, the more coils you have the more inductive impedance you introduce to the circuit. Without some serious calculations, I have no idea if this even approaches the 1 ohm resistive impedance we had been discussing-that may very well more of a theoritical consideration than a practical one.

My plan is to build both the resistive tester as well as the coil multiplier and compare field accuracy. I'm fairly certain that centering the clamp in the coil isn't required for accuracy, but perhaps I'm wrong. That's easy enough to test once this is all set up on the bench. And yes, I'll check overall calibration against my fancy Fluke meter and external clamp that's goes to 4 decimal place accuracy and has a factory calibration sticker on the side. It reads down to 1/10 of an mA in the jaw clamp, but unfortunately costs around $1,000.

I just want a simple and affordable system that anyone can use to read ground leakage currents to 1 mA sensitivity. I think that's the best way to troubleshoot GFCI tripping, not only for ProSound users, but for RV owners as well who are always complaining about GFCI random tripping.
Title: Re: GFCI leakage testing
Post by: Jeff Carter on June 10, 2014, 12:20:01 AM
Of course, the more coils you have the more inductive impedance you introduce to the circuit. Without some serious calculations, I have no idea if this even approaches the 1 ohm resistive impedance we had been discussing-that may very well more of a theoritical consideration than a practical one.

One example of a coil inductance calculator here. At some level this is an approximation but gives a bit of a feel for magnitude and scaling:
http://www.66pacific.com/calculators/coil_calc.aspx (http://www.66pacific.com/calculators/coil_calc.aspx)

Inductance goes up as the square of the number of turns so it's most worrisome for a large number of turns. Depending on the geometry inductance scales somewhere between linearly (short coil) and quadratically (long solenoid) with radius, so a small coil will have lower inductance than a large one for the same number of turns.

In terms of actual numbers, I wound a few magnetic field coils in my PhD work. Inductances for the largest ones I did (square, about 12" a side, 64 turns) were on the order of 1 mH give or take, or a reactance of about 0.3 ohms at 60 Hz. As such I think a 10x multiplier with the dimensions of the spool should definitely be OK, and even 100x is probably OK given that the reactance of the coil adds in quadrature with the 1 ohm sense resistance to give the total impedance of the circuit (I assume you're worried about the coil impedance and sense resistor forming a voltage divider?)
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 10, 2014, 12:28:45 AM
In terms of actual numbers, I wound a few magnetic field coils in my PhD work. Inductances for the largest ones I did (square, about 12" a side, 64 turns) were on the order of 1 mH give or take, or a reactance of about 0.3 ohms at 60 Hz. As such I think a 10x multiplier with the dimensions of the spool should definitely be OK, and even 100x is probably OK given that the reactance of the coil adds in quadrature with the 1 ohm sense resistance to give the total impedance of the circuit (I assume you're worried about the coil impedance and sense resistor forming a voltage divider?)

I love it when the eggheads crawl out of the woodwork.  ;D

I would only do one current measuring method at a time, so too much series impedance at 60 Hz would reduce the actual current leakage and mess with the measurment. But my gut feel from messing with coils this size for speaker crossover systems is that any inductive effect would be quite small and not significantly change the measurement. I'll run the numbers and gather some empirical data to compare them against once I get the experiment on the bench.

Too much fun...  8)
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 10, 2014, 06:35:21 AM
For my no-shunt jig with the diodes, there is a side effect of the diodes worth mentioning. They can clip the ac waveform peaks.  The measurement limit with two diodes is less than 10ma.  Current measurements are only linear until the diodes start to clip the peaks.

A one ohm resistor is good as long as your voltmeter reads millivolts (or even microvolts) well.  My multimeter jumps around quite a bit at low levels.

Yes, diodes are a double-edged sword... they offer protection but also the possibility of non-linear readings. My plan was going to be to mark the unit with a max number it will measure before being suspect. Of course, clipping diodes plus a 100-watt dissipation resistor is kind of a belt and suspenders approach. But I've learned a long time ago that as soon as you think something can't happen, worlds collide. Besides, as mentioned earlier, this is a great mind experiment for all of us to understand the design and calibration of test gear. This will help when we want to develop actual procedures to troubleshoot GFCI tripping scenarios.
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 10, 2014, 06:54:35 AM
As such I think a 10x multiplier with the dimensions of the spool should definitely be OK, and even 100x is probably OK...

Case in point... you can buy a commercial product called an Amprobe ELS2A that splits out the hot wire of an Edison outlet for clamp meter measurements. It includes one section marked X1 and a second section marked X10 which obviously has 10 loops internally. I just want something similar that measures the EGC (Ground Wire) current, and maybe has a 100X multiplier loop since the ground leakage currents should be really small for properly functioning gear.

No, this isn't going to be laboratory grade gear with traceable calibration and certification. I want something simple that any ProSound or RV technician can build or purchase in order to better troubleshoot GFCI issues. I just hate flying blind when troubleshooting, and think that GFCI tripping problems will only get more common. 
Title: Re: GFCI leakage testing
Post by: Stephen Swaffer on June 10, 2014, 08:17:34 AM
The only reason I mention centering the coils on the clamp is that Klein mentions centering the wire in the clamps for max accuracy.  I know you aren't supposed to read instruction manuals, but its amazing how many times people asked me, "How did you know that?" when I fixed something in the factory!   In any case, I was reading to understand the variation I commonly notice when trying to check the amp draw on the motor.
Title: Re: GFCI leakage testing
Post by: Mike Sokol on June 10, 2014, 08:40:08 AM
The only reason I mention centering the coils on the clamp is that Klein mentions centering the wire in the clamps for max accuracy.  I know you aren't supposed to read instruction manuals, but its amazing how many times people asked me, "How did you know that?" when I fixed something in the factory!   In any case, I was reading to understand the variation I commonly notice when trying to check the amp draw on the motor.

As much as I preach about reading manuals, maybe I should read the manual. I'll try out during testing and see how much it changes the readings.
Title: Re: GFCI leakage testing
Post by: Stephen Swaffer on June 10, 2014, 09:17:46 AM
Might also look into using an off the shelf CT, if you can find anything surplus.  A 500 amp CT would be 100:1 and a 50 amp would be 10:1.  Just use them backwards from their intended hook up.