4-wire Bench DMM questions

[zargnut]

I just ordered a rigol 3058 after a good experience with their little oscilloscope. I got it primarily because space and depth of shelving is very limited, and can't accomodate a used HP/Fluke. Does anyone have any experience with the DMM? Dave, think Rigol will give you a DMM to eval?   Maybe they forgot about the mods...

regarding fabricating a set of 4 wire leads. I have a bunch of beldon 1192 quad mic cable laying around and was wondering if I could use this to wire up a 4-wire set that terminates at two micro-clips.

wire specs:

http://www.belden.com/techdatas/english/1192A.pdf

thanks for any advice.
dan

[alm]

I don't think cables for 4W resistance measurements are very critical, the whole point of 4W measurements is to eliminate lead/contact resistance. Be sure to use proper clips, with the only connection between the two wires through the wire you clipped on, regular crocodile clips will short the two wires, so contact resistance between the jaws and the wire (the largest variable) is still an issue. You can actually do good resistance measurements with just for wires with four clips, you just need a larger area to clip on to, and it's more hassle.

You could connect the shield to the guard terminal if your meter has one, see the manual for proper connections in that case. Note that you'll probably only use two of the four conductors, since you typically want two separate cables. I would connect two in parallel, since the commercial leads tends to have a larger wire gauge than 24 awg.

[mikeselectricstuff]

Clips for 4-terminal measurements are called kelvin clips :
http://uk.farnell.com/silvertronic/502001/insulated-kelvin-clip/dp/1851741

[zargnut]

awsome, thanks.

regarding the wire gauge.. with the low impedance wire and short length, does the gauge really matter for 4wire resistance measurements?

Do the sensor and current leads share a ground? IE could I wire the copper sheath to both grounds and double up the insulated conductors for each + sense/current wires? Or am I missing the whole isolated circuit point here?

[mikeselectricstuff]

with 4-wire measurement, resistance of the leads is cancelled out, so pretty much any wire will work. The 4 meter terminals are connected with 4 seperate wires to the clips.

[alm]

with 4-wire measurement, resistance of the leads is cancelled out, so pretty much any wire will work. The 4 meter terminals are connected with 4 seperate wires to the clips.

To be more accurate, something like 99.9% is canceled, it's not an electrometer or differential voltmeter. There is still some current flowing through the sense wires, but much less than the 1mA or so (depending on the range) of a 2 wire measurement, especially if the meter has >10Gohm input impedance at that range. I don't think it's any significant issue in this case, but if you have four wires available, and only need two, why not connect them in parallel instead of leaving them floating?

Do the sensor and current leads share a ground? IE could I wire the copper sheath to both grounds and double up the insulated conductors for each + sense/current wires? Or am I missing the whole isolated circuit point here?

The manual for your DMM should tell you how to connect any shielding, at least the manuals I've seen do. Connecting shield to the current lo terminal should be fine I believe, the inputs of a DMM aren't really differential, the capacitance between lo and ground tends to be much higher than between hi and ground. If the meter has a separate guard terminal, you should use that, but that seems to have fallen out of favor in recent DMM designs. I don't think most commercial Kelvin leads are shielded, that would make them harder to connect (they usually just have dual banana jacks).

[EEVblog]

I just ordered a rigol 3058 after a good experience with their little oscilloscope. I got it primarily because space and depth of shelving is very limited, and can't accomodate a used HP/Fluke. Does anyone have any experience with the DMM? Dave, think Rigol will give you a DMM to eval?

I doubt Rigol like me any more!

Dave.

[saturation]

You can find 2 common varieties of Kelvin clips for milliohm measurements, one with a separate guard clip and one without it.  DMM kelvin's maybe different for that used by a  high end LCR meter. it depends on the units, which produces a different voltage on the guard to reduce parasitic LC.

If your DMM does not have a guard terminal, you can just build the non-guard version.  To make the Agilent style clip you take your coax cable and solder a connection to all the cable shields then connect it to a single alligator, this is the guard lead.  For the non guarded version, you needn't connect it.

The most costly part of the 4 wire setup are the clips; if you use ordinary alligators or mini-clips, they may not provide consistently low resistance to reduce errors when measuring precision to milliohms, which is what you want the 4 wire for anyway.  Good kelvin clips are gold plated and run about $20-30 each.  There are a few silver plated versions for a little less money, but don't know if they are as good as gold.  'Brand name' kelvin leads cost about $150-200.  There are no-name Chinese brands for $20 complete, just check them to insure they really work, particularly if its really gold plated, not paint





The guard terminal reduces the common mode voltage to the DMM inputs; however since modern DMMs have very high CMRR, EMI or noise pickup maybe far below the stated milliohm resolution of the DMM's to make it practically unneeded.  For example if the CMRR is -100dB, then it will only start to be a problem if you can resolve 0.01 milliohm.

The guard clip is mostly connected to low, or ground, whichever reduces noise and creates a stable value.  In some old series Agilent DMMs, the guard terminal has a switch on the meter than shifts it between low and ground.

[alm]

The guard terminal reduces the common mode voltage to the DMM inputs; however since modern DMMs have very high CMRR, EMI or noise pickup maybe far below the stated milliohm resolution of the DMM's to make it practically unneeded.  For example if the CMRR is -100dB, then it will only start to be a problem if you can resolve 0.01 milliohm.

That's nice in theory (I haven't checked the specs), but I remember trying to measure a ~3-4mOhm shunt once with some 6.5 digit DMM (either HP 34401A or Keithley 2000), and I was unable to get meaningful results (inconsistent values that were much higher than expected), presumably because of either 50Hz pickup or Seebeck effect (thermocouple). Note that with a test current of 1mA, the voltage drop over 1mOhm is only 1µV. I didn't actually verify if guarding made a difference, I just used a lab supply to shove a few amps through it (it was a beefy shunt anyway), and measure voltage and current. But either way, it doesn't take a lot of current to induce 1µV in 10Gohm. I know that to accurately characterize a shunt, you need to measure it at the rated current, but I didn't have a large enough current source available.

The guard clip is mostly connected to low, or ground, whichever reduces noise and creates a stable value.  In some old series Agilent DMMs, the guard terminal has a switch on the meter than shifts it between low and ground.

I'm not sure if the other position is ground, the HP 3456A manual (the only one I have on hand at the moment) appears to imply something different, but is not very explicit. But don't quote me on that, I've never used it. Most newer DMM's except the very accurate ones don't have it anymore for some reason, they claim that they're not needed anymore.

[saturation]

I agree with you completely, I prefer to have the guard terminal there, as an additional insurance against erratic measurements; it still comes in the 3458a DMM, its a tested design, but without internal changes since 1989.

You could of course, split the low input jack, so you can sneak the guarded leads in there, and see what happens.

I think a while back, people where miffed about the absence of the guard and it was made an FAQ.

http://www.home.agilent.com/agilent/editorial.jspx?cc=US&lc=eng&ckey=1000000883:epsg:faq&nid=-536902435.0.00&id=1000000883:epsg:faq

Schematic, with Z1,2,3 referring to impedance between the input connectors, see photo.  I think this works for low ohms measurements, but with megaohms with milliohms resolution we get more issues of CMRR, electrostatics, and EMI.

NI is one of the few to recommend this type of connection, I've never had to face this issue to date, but if improves megaohm stability, please feedback:



http://zone.ni.com/devzone/cda/tut/p/id/3981#toc5

The guard terminal reduces the common mode voltage to the DMM inputs; however since modern DMMs have very high CMRR, EMI or noise pickup maybe far below the stated milliohm resolution of the DMM's to make it practically unneeded.  For example if the CMRR is -100dB, then it will only start to be a problem if you can resolve 0.01 milliohm.

That's nice in theory (I haven't checked the specs), but I remember trying to measure a ~3-4mOhm shunt once with some 6.5 digit DMM (either HP 34401A or Keithley 2000), and I was unable to get meaningful results (inconsistent values that were much higher than expected), presumably because of either 50Hz pickup or Seebeck effect (thermocouple). Note that with a test current of 1mA, the voltage drop over 1mOhm is only 1µV. I didn't actually verify if guarding made a difference, I just used a lab supply to shove a few amps through it (it was a beefy shunt anyway), and measure voltage and current. But either way, it doesn't take a lot of current to induce 1µV in 10Gohm. I know that to accurately characterize a shunt, you need to measure it at the rated current, but I didn't have a large enough current source available.

The guard clip is mostly connected to low, or ground, whichever reduces noise and creates a stable value.  In some old series Agilent DMMs, the guard terminal has a switch on the meter than shifts it between low and ground.

I'm not sure if the other position is ground, the HP 3456A manual (the only one I have on hand at the moment) appears to imply something different, but is not very explicit. But don't quote me on that, I've never used it. Most newer DMM's except the very accurate ones don't have it anymore for some reason, they claim that they're not needed anymore.

[Eliminateur]

I doubt Rigol like me any more!

Dave.

What happened?, wasn't rigol honchos in contact with you and all?
btw, what happened to the Extech killer-tweezer, did they send you a revisited version/traveled to recover it yadda yaddA?