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 April 13, 2018, 12:03:47 PM

Title: Extended test leads on Kelvin Bridge
Post by: Mike Sokol on April 13, 2018, 12:03:47 PM
OK guys, so this is an interesting question that just stumped engineering tech support at Greenlee. I need to measure the actual low-ohm resistance of a 100+ ft cable buried cable run. Greenlee makes a portable Kelvin Bridge that will measure resistance down in the milli-ohm numbers. And the cool thing about a Kelvin Bridge is that the test lead resistance is nulled out of the measurement. That's because there's two isolated paths through the resistor being measured. Once is a low impedance current loop, while the second one if a very high impedance voltage measurement. And these two loops are kept separated in the test jaws, with one side of the same clamp. See Kelvin Bridge theory here: https://en.wikipedia.org/wiki/Kelvin_bridge

Here's the Greenlee CLM-1000 which is just a Kelvin Bridge Meter. And you can see that it has 4 test leads terminating in a pair of test clamps, just as I described above.  https://www.greenleestore.com/products/greenlee-clm-1000-cable-length-meter-for-awg-kcmii-wire-and-cable?utm_medium=cpc&utm_source=googlepla&variant=15452440451&gclid=EAIaIQobChMIg_aVucW32gIVC4nICh39_w_6EAYYAiABEgK2hfD_BwE

What's to stop me from extending one of the double test leads by 100 feet or so? From all my time using Kelvin Bridges to measure super-low ohm resistors down to ppm accuracy back in the 80's, the test lead resistance isn't actually inside of the test loop at all. So from a practical standpoint I could make this several hundred feet long if I wanted and it should still measure 0.01 ohms on the test resistance, or whatever.

Their engineer is sending me a free sample to experiment with as long as I share my test results. What do you all think? Can I extend one of the test lead pairs by 100 ft without goofing up the accuracy of a sub-ohm resistance test?
Title: Re: Extended test leads on Kelvin Bridge
Post by: Chris Hindle on April 13, 2018, 12:29:24 PM

Their engineer is sending me a free sample to experiment with as long as I share my test results. What do you all think? Can I extend one of the test lead pairs by 100 ft without goofing up the accuracy of a sub-ohm resistance test?

Mike, my guess is that's what the engineer wants YOU to tell him.
Uncharted territory, and all that.
Who knows, a future factory accessory may be the MS-100 Test Lead Extension Kit.
You sure do find interesting ways to push the limits of knowledge. - THANK YOU !!!
Chris.
Title: Re: Extended test leads on Kelvin Bridge
Post by: Mike Sokol on April 13, 2018, 12:35:32 PM
Mike, my guess is that's what the engineer wants YOU to tell him.
Uncharted territory, and all that.
Who knows, a future factory accessory may be the MS-100 Test Lead Extension Kit.
You sure do find interesting ways to push the limits of knowledge. - THANK YOU !!!
Chris.

Yes, that's exactly right. He said I only needed to promise to share my test data with him, which I think is a great deal. And of course I'll share it here as well.
Title: Re: Extended test leads on Kelvin Bridge
Post by: Jonathan Johnson on April 13, 2018, 03:06:21 PM
Yes, that's exactly right. He said I only needed to promise to share my test data with him, which I think is a great deal. And of course I'll share it here as well.

Demonstrating with test data is one thing, but you also have to prove it mathematically.
Title: Re: Extended test leads on Kelvin Bridge
Post by: John Roberts {JR} on April 13, 2018, 04:02:37 PM
Kelvin connections for measurement is a fairly mature technology so I suspect somebody has tried that.

In practice it may come down to the null range of bridge device.

You might be more symmetrical with two sets of 50' test leads with meter in the middle.

BTW subtracting out lead resistance is not a big deal, for a 100' buried conductor there may be larger stray ground currents and fields.

Have fun... but maybe check the library (IEEE et al).

JR
Title: Re: Extended test leads on Kelvin Bridge
Post by: Mike Sokol on April 13, 2018, 05:11:46 PM
Demonstrating with test data is one thing, but you also have to prove it mathematically.

Actually, I can prove it mathematically. Some 40 years ago I designed and built Kelvin Bridges on the bench from loose parts, and we used them on a production line for Laser trimming resistors while they were spinning in a chuck. So we did a lot of measurements with a Kelvin Bridge that would have been impossible any other way. I'm in the upper 99% certainty that this will work, but I haven't had to do it in battle for a very long time. I'll run some basic tests next week, then go in and show my math on paper. Unless I'm thinking about this completely wrong it should work exactly as I describe. Let's see, so 2+2=5 correct?  :o
Title: Re: Extended test leads on Kelvin Bridge
Post by: Frank DeWitt on April 13, 2018, 06:13:38 PM
I agree with John.  On paper it works.  It will come down to the instrument having the range to deal with two very different probes.  My guess is that two 100 ft probes is better them one 2 ft and one 100 ft.
Title: Re: Extended test leads on Kelvin Bridge
Post by: Mike Sokol on April 13, 2018, 07:20:14 PM
I agree with John.  On paper it works.  It will come down to the instrument having the range to deal with two very different probes.  My guess is that two 100 ft probes is better them one 2 ft and one 100 ft.

My plan is to solder up 2 pairs of 100 ft extensions using 16 gauge twisted speaker wire. Now I know that won't have the uniform twist of a true twisted pair, but it should probably help reduce any stray hum pickup on the hi-z loop. I was even thinking of using mic cable and bonding the shield to the current/low-z side. But this tester is designed to work with up to 1,000 feet of unshielded wire, so I would think it has some pretty serious hum rejection from the get go. We shall see....
Title: Re: Extended test leads on Kelvin Bridge
Post by: Rob Spence on April 13, 2018, 10:23:00 PM


We waved as we passed Shenandoah this morning heading south




Sent from my iPad using Tapatalk Pro
Title: Re: Extended test leads on Kelvin Bridge
Post by: John Roberts {JR} on April 14, 2018, 10:00:15 AM
My plan is to solder up 2 pairs of 100 ft extensions using 16 gauge twisted speaker wire. Now I know that won't have the uniform twist of a true twisted pair, but it should probably help reduce any stray hum pickup on the hi-z loop. I was even thinking of using mic cable and bonding the shield to the current/low-z side. But this tester is designed to work with up to 1,000 feet of unshielded wire, so I would think it has some pretty serious hum rejection from the get go. We shall see....
My concern was also about stray fields/currents affecting the buried conductor. Typical VOM inject a reference current and measure voltage, so either could cause errors.

Maybe swapping leads and measuring both directions could reveal if my imaginary errors are present, or not.  ::)

JR
Title: Re: Extended test leads on Kelvin Bridge
Post by: Mike Sokol on April 15, 2018, 02:17:27 PM
My concern was also about stray fields/currents affecting the buried conductor. Typical VOM inject a reference current and measure voltage, so either could cause errors.

Maybe swapping leads and measuring both directions could reveal if my imaginary errors are present, or not.  ::)

JR

Yes, you're 100% correct. I don't know all the variable that can affect the readings, but I'm going to find out pretty quick if this will work. Please Stand By....
Title: Re: Extended test leads on Kelvin Bridge
Post by: Mike Sokol on April 23, 2018, 07:51:33 AM
Yes, you're 100% correct. I don't know all the variable that can affect the readings, but I'm going to find out pretty quick if this will work. Please Stand By....

First test with a single 75 ft run of 16-gauge extension on one of the Kelvin clamps seems to work perfectly. The meter has a calibrate function and test bar which nulls out to 0 ohms correctly, and I've used it to read several different lengths of wire correctly. I'll build a second 75 ft run for the other clamp today and take both of them to the test site on Wednesday. I'll report back how well it works under dirty conditions in a swamp. No kidding, I'm doing this in a swamp...
Title: Re: Extended test leads on Kelvin Bridge
Post by: Chris Hindle on April 23, 2018, 09:04:38 AM
First test with a single 75 ft run of 16-gauge extension on one of the Kelvin clamps seems to work perfectly. The meter has a calibrate function and test bar which nulls out to 0 ohms correctly, and I've used it to read several different lengths of wire correctly. I'll build a second 75 ft run for the other clamp today and take both of them to the test site on Wednesday. I'll report back how well it works under dirty conditions in a swamp. No kidding, I'm doing this in a swamp...
Don't let the alligators bite............
Chris.
Title: Re: Extended test leads on Kelvin Bridge
Post by: Mike Sokol on April 23, 2018, 01:51:28 PM
Don't let the alligators bite............
Chris.

Roger that.... And I just tried the second 75-ft extension and it calibrates and seems to work correctly. This is getting interesting.
Title: Re: Extended test leads on Kelvin Bridge
Post by: Chris Hindle on April 23, 2018, 06:42:21 PM
Roger that.... And I just tried the second 75-ft extension and it calibrates and seems to work correctly. This is getting interesting.
Kinda looks like that as long as you can get calibration, limits are up to you.
Now, here's a question. Is the bridge "good enough" to make a valid measurement in, say, a 20 foot D-U-T line, using 300 feet of leads ?
I would imagine that as long as the leads don't drastically outsize the D-U-T, you should be good.
Chris.
Title: Re: Extended test leads on Kelvin Bridge
Post by: Stephen Kirby on April 24, 2018, 12:28:30 AM
If you can manage to stretch things out, anchoring one end and twisting with a drill can result in a pretty even wind.  A shop I was at made cables for the Bradleys.  We'd lay out the wires across several work benches, clamp one end and use a cordless drill to twist it.  You have to hold a pretty good tension on it so make sure the anchored end is solid.  Then we'd have one person shooting an air gun up one end of shrink sleeving while someone shoved the cables up the other end.  Then run them through turned down SMT reflow ovens to shrink the sleeving.  Fortunately we got to charge Mil-Spec prices for all this labor.
Title: Re: Extended test leads on Kelvin Bridge
Post by: Mike Sokol on April 24, 2018, 09:36:00 AM
Kinda looks like that as long as you can get calibration, limits are up to you.
Now, here's a question. Is the bridge "good enough" to make a valid measurement in, say, a 20 foot D-U-T line, using 300 feet of leads ?
I would imagine that as long as the leads don't drastically outsize the D-U-T, you should be good.
Chris.

Back when I did this seriously in the 80's we used a battery source for the current and a DC null meter. The comparison source was a precision decade resistance box calibrated annually at the National Bureau of Standards. With that simple setup we could match resistors down to 5ppm and sometimes as low as 2ppm. When I get some time I'll have to write a paper on the theory of Kelvin bridges, but in theory you can add hundreds of feet of cable to the clamps and still read sub-ohm resistances. It doesn't work at all like a standard 2-wire meter.   
Title: Re: Extended test leads on Kelvin Bridge
Post by: John Roberts {JR} on April 24, 2018, 09:49:54 AM
Back when I did this seriously in the 80's we used a battery source for the current and a DC null meter. The comparison source was a precision decade resistance box calibrated annually at the National Bureau of Standards. With that simple setup we could match resistors down to 5ppm and sometimes as low as 2ppm. When I get some time I'll have to write a paper on the theory of Kelvin bridges, but in theory you can add hundreds of feet of cable to the clamps and still read sub-ohm resistances. It doesn't work at all like a standard 2-wire meter.
The Kelvin connection for precision testing is known to engineers, maybe not so well known around sound reinforcement forums.

This technique has other applications audio, like using sense leads on speakers to ignore wire losses. Again not common for sound reinforcement.

JR
Title: Re: Extended test leads on Kelvin Bridge
Post by: Chris Hindle on April 24, 2018, 12:29:46 PM
Back when I did this seriously in the 80's we used a battery source for the current and a DC null meter. The comparison source was a precision decade resistance box calibrated annually at the National Bureau of Standards. With that simple setup we could match resistors down to 5ppm and sometimes as low as 2ppm. When I get some time I'll have to write a paper on the theory of Kelvin bridges, but in theory you can add hundreds of feet of cable to the clamps and still read sub-ohm resistances. It doesn't work at all like a standard 2-wire meter.
Thanks Mike and J.R.
Very enlightening.....
I am well versed in "standard" instrumentation, but this is new to me.
I've never had a need for this kind of tool, at this kind of precision.
Chris.
Title: Re: Extended test leads on Kelvin Bridge
Post by: Stephen Kirby on April 24, 2018, 04:25:52 PM
The Kelvin connection for precision testing is known to engineers, maybe not so well known around sound reinforcement forums.

This technique has other applications audio, like using sense leads on speakers to ignore wire losses. Again not common for sound reinforcement.

JR
Going around with some folks at work about how close the sense leads have to be to the resistive element.  There's this bright idea that if you partition the board pad for an SMT resistor the current sense is more accurate than if you just run two traces off the one pad.  I tried explaining that the component's internal parasitics from the resistive element down to the board pad would swamp any differences.  Much as the leads on a conventional axial resistor does.  Maybe not as bad, but there are comparatively high resistance metalizations going from the element on top around the sides of the ceramic body to where it's connected to the board.
Title: Re: Extended test leads on Kelvin Bridge
Post by: TJ (Tom) Cornish on April 24, 2018, 04:40:34 PM
The Kelvin connection for precision testing is known to engineers, maybe not so well known around sound reinforcement forums.

This technique has other applications audio, like using sense leads on speakers to ignore wire losses. Again not common for sound reinforcement.

JR
Not common, but not unheard of.  For a long time Nexo used sense lines from the output of the amp for feedback and limiting.  Now that virtually all vendors have gone to branded amps and proprietary processing, they don't need to see on the other side of the black box via sense lines; they just absorbed the amp into their own black box.  :)
Title: Re: Extended test leads on Kelvin Bridge
Post by: John Roberts {JR} on April 24, 2018, 04:42:04 PM
Going around with some folks at work about how close the sense leads have to be to the resistive element.  There's this bright idea that if you partition the board pad for an SMT resistor the current sense is more accurate than if you just run two traces off the one pad.  I tried explaining that the component's internal parasitics from the resistive element down to the board pad would swamp any differences.  Much as the leads on a conventional axial resistor does.  Maybe not as bad, but there are comparatively high resistance metalizations going from the element on top around the sides of the ceramic body to where it's connected to the board.
Easy enough to prove (or not) empirically.... just do it both ways and measure the difference (maybe with another bridge.  ;D

JR
Title: Re: Extended test leads on Kelvin Bridge
Post by: brian maddox on April 24, 2018, 06:19:35 PM
Not common, but not unheard of.  For a long time Nexo used sense lines from the output of the amp for feedback and limiting.  Now that virtually all vendors have gone to branded amps and proprietary processing, they don't need to see on the other side of the black box via sense lines; they just absorbed the amp into their own black box.  :)

Our friend John Meyer did the same thing before they decided to just put the amp in the speaker and be done with it.

I still come across UM-1s with the wrong amping/processing and no sense line on occasion.  Needless to say, they don't sound good.  :)
Title: Re: Extended test leads on Kelvin Bridge
Post by: Mike Sokol on April 25, 2018, 09:22:58 PM
I just used my extended kelvin bridge today in a swamp and it worked perfectly. And the measurements with the 75 ft extensions match what it measured without them. I may have a winner here, but I need more verification.
Title: Re: Extended test leads on Kelvin Bridge
Post by: Chris Hindle on April 26, 2018, 01:36:52 PM
I just used my extended kelvin bridge today in the swamps of Baton Rouge and it worked perfectly. And the measurements with the 75 ft extensions match what it measured without them. I may have a winners here, but I need more verification.
Always nice when a plan comes together.
Still got all your fingers and toes, i presume?  ::)
Chris.
Title: Re: Extended test leads on Kelvin Bridge
Post by: Mike Sokol on April 26, 2018, 09:27:45 PM
Always nice when a plan comes together.
Still got all your fingers and toes, i presume?  ::)
Chris.

Yes, but there was a lot of mud everywhere. There were several other EE's on the site monitoring what I was doing and they all want to buy a Kelvin Bridge now once I explained operation to them. For standard field testing like you need for audio production the accuracy is simply overkill, but for diagnostics of sub-ohm resistances it's really handy. I'll share more later once I'm able to.
Title: Re: Extended test leads on Kelvin Bridge
Post by: Scott Holtzman on April 26, 2018, 10:17:12 PM
I went to look up Wheatstone bridges and I am trying to understand why the Kelvin bridge offers better low resistance resolution than the Wheatstone.

Any comments?
Title: Re: Extended test leads on Kelvin Bridge
Post by: John Roberts {JR} on April 27, 2018, 09:31:35 AM
I went to look up Wheatstone bridges and I am trying to understand why the Kelvin bridge offers better low resistance resolution than the Wheatstone.

Any comments?
The magic about Kelvin measurements is that the reference current injected to develop a voltage drop in the DUT (device under test) to impute resistance, flows through separate leads that are not part of the voltage measurement. In a conventional meter the test lead resistance would introduce errors. With a 4 wire connection, there is no current flowing in the voltage sense leads so no IxR errors.

Obviously this is more of a concern for low impedance measurements where the test lead resistance can be significant.

JR
Quote from: wiki sez
The operation of the Kelvin bridge is very similar to the Wheatstone bridge, but uses two additional resistors.
https://en.wikipedia.org/wiki/Kelvin_bridge (https://en.wikipedia.org/wiki/Kelvin_bridge)
Title: Re: Extended test leads on Kelvin Bridge
Post by: Mike Sokol on April 29, 2018, 09:29:55 AM
Obviously this is more of a concern for low impedance measurements where the test lead resistance can be significant.

Good explanation and you're quite right about this only really needed for low-resistance measurements. In my case I'm measuring EGC bonding resistance/impedance down to a few tens of milli-ohms, and the two test endpoints can be 100 feet apart or more. Even the short test leads in a standard ohmmeter can introduce a few hundred milli-ohms of series resistance, which contaminates these measurements.

What I like about the Klein tester is that it's a simple handheld meter with a quick calibration procedure and built-in temp gauge that provides the proper correction factor if you want to calculate the actual length of a spool of wire. That's because this unit is being marketed as a way to measure how much wire is left on a spool. So you set if for 10 gauge or 2 gauge wire or whatever, then use the Kelvin clips on both ends of the wire, and will tell you if you've got 100 or 200 or 300 feet of wire on the spool, down to a few percent. I don't care about that since I'm interested in EGC bonding. So Klein is interested in my experiment since they've never considered my application and it might be a new market for them.
Title: Re: Extended test leads on Kelvin Bridge
Post by: Mike Sokol on April 29, 2018, 09:33:14 AM
Not common, but not unheard of.  For a long time Nexo used sense lines from the output of the amp for feedback and limiting. 

IIRC this was also used to improve damping factor on long cable runs. Pretty simple concept since the sense lines just become part of the negative feedback loop in the power amplifier. In theory you can make a long run of 18 gauge behave like really heavy wire, except for voltage drop and reduced headroom (of course).
Title: Re: Extended test leads on Kelvin Bridge
Post by: John Roberts {JR} on April 29, 2018, 09:50:15 AM
IIRC this was also used to improve damping factor on long cable runs. Pretty simple concept since the sense lines just become part of the negative feedback loop in the power amplifier. In theory you can make a long run of 18 gauge behave like really heavy wire, except for voltage drop and reduced headroom (of course).
To continue down this music veer, when designing power amplifiers being able to publish a crazy high damping factor is consider a meritorious specification, while in the real world an amplifier with DF of thousands will not realize much benefit with even a short length of speaker wire attached. ::) When chasing high amplifier DF numbers even the short wires inside the amplifier going to the output terminals matter. 

Last century while product manager for power amps at Peavey, I was given a power amp with near infinite damping factor due to a clever feedback arrangement (Jack Sondermeyer RIP got a patent for it). In fact he could tweak some terms and create a power amp with negative output impedance so it could compensate for the external speaker wires too. We even momentarily considered adding a control to the power amps allowing you to adjust compensation for wire length (resistance).

But nah... we quickly realized this would just create more confusion with Peavey customers than provide any practical benefit, but it was nice to publish DF with extra zeros for the customers who believed it was important.  8)

FWIW amplifiers using sense lines was useful but impractical for live sound reinforcement where systems were set up and torn down repeatedly. An open sense line would lead to bad behavior from the amps. In design for mass markets we have to ASSume customers will do the worst, not the best.

JR
Title: Re: Extended test leads on Kelvin Bridge
Post by: Mike Sokol on April 29, 2018, 12:41:06 PM
Back to Kelvin Bridge theory. Here's a really good explanation of how it works: https://www.allaboutcircuits.com/textbook/direct-current/chpt-8/kelvin-resistance-measurement/ and as you can see, you can insert a lot of resistance in the test leads without adding to the resistance being measured. Because there's two different loops, one low-z path for current and one high-z path for voltage, the lead lengths are automatically cancelled out of the measurement. Here's a picture of the basic setup, the most import part being the Kelvin Clamps which have each half of the clamp isolated from each other.
Title: Re: Extended test leads on Kelvin Bridge
Post by: Mike Sokol on April 29, 2018, 02:07:01 PM
FYI: Here's exactly what I used to use back in the 80's, and which I still have on my shelf of vintage (old) test gear. https://www.ebay.com/itm/ESI-Portametric-PVB-300-Potentitometric-Voltmeter-Bridge/272452748026?hash=item3f6f7302fa:g:GewAAOSwHMJYLVuD
It should still work if I make a replacement holder to substitute a battery stack for the original mercury battery, which of course is no longer available.

I was originally going to take this on my gig, and it would have worked. But it is a bit cumbersome to null-out and explain to other engineers, while the Greenlee CLM-1000 meter is really quick and easy to use.