Shielded test leads and Guard effectiveness on DMM's W/O a guard (e.g. 34465A)

[kj7e]

Got around to making some proper low thermal PTFE shielded 4 wire test leads.  As an experiment I tested using the 3 and 10 amp current jacks on my Keysight 34465 as a Guard, works quite well;

https://youtu.be/UTCKNm-PYWs

[cellularmitosis]

brilliant! 

[montemcguire]

Good shielding job - quite effective. The only effect that seems to remain is the triboelectric effect of the insulation. This is a residual voltage generated by flexing or moving the dielectrics relative to the conductors. If there is some sort of way to get a conductive, flexible filler around each of the 4 internal conductors and between these conductors and the bare shield, this effect will be greatly diminished. In the past, fillers such as carbon impregnated paper or cloth have been used, and also carbon loaded polymer tubing. Not sure how this could be done easily now, but it's the last source of error in those really excellent cables.

Did you build the cable up from individual PTFE wire? If so, perhaps you could find some sort of conductive 'soft stuff' to cable along with the four PTFE conductors, and arrange for them to also contact the inside of the shield in order to drain away the charge generated by flexing the cable. Maybe something as simple as cotton string that's been rubbed with graphite could be cabled along with the PTFE wires and be effective?

Anyway, nice job! Cables and shielding really are important, as your video demonstrates.

[kj7e]

The wire is 18 awg stranded silver plated 4 conductor with PTFE jackets and a braided shield with an outer PTFE jacket.  The spades are copper with a direct gold plating, the 4mm hollow Z-plugs are gold plated beryllium copper.  This was the DIY kit from https://www.ab-precision.de/products/accessories/

The kit had enough wire for 3 cables, two the length shown in the video, and one twice that length.  Since my best meters use recessed 4mm plugs (Keithley DMM7510 and Keysight 34465A), I needed to use the lowest thermal banana plugs I could find.  The z-plugs fit very solid and have great contact area.  I have not decided on the connector combination for the other two cables yet.  I have extra spades and z-plugs for when that time comes.

[BradC]

Huh. Fascinating. I've been wondering if that was possible on a 3457a. Now I need to actually test it. Prior to your video I could comfortably just pontificate.

[kj7e]

I recall reading in a manual on some DMM where it did not have a guard, to use the current jack.  I cant find that now, it very well may have been in the 3457A manual.

[GEOelectronics]

BINGO!
a big "Thanks"

Would you mind testing the guard for effect to one or the other negative terminals and possibly the chassis ground?\

Thanks
Geo

[kj7e]

George,

My 34465A is tied to my earth bar with a 1" braid,  I did test with the cable shield open at the meter but with a jumper from the reference guard jack to my earth bar, nearly the same results as in the video.

[GEOelectronics]

Good. Thanks a lot. Very helpful.

Is your electricity floating (add: With proper earthing of course) and can you comment on the +/- of that scheme somewhere?

Geo

[kj7e]

The AC mains are not floating, I have to be cautious of ground loops with low level measurements.  Sometimes using the earth bar helps, sometimes it can add noise.  The ground differential will change depending on AC mains loading and other factors.  In general, I found having my gear all tied to the same earth bar does help, but many times the DUT should NOT be tied to it.

[ManateeMafia]

I noticed a similar effect measuring high ohm resistors. IIRC, putting on an anti-static wrist band will also provide the same effect. I did this measuring my Fluke 5450A and also a 1G home-built resistor.

[blackdog]

Hi kj7e,

I use the same technique with different 4-wire test leads and some measurement boxes in my LAB.
Almost always this lowers the noise level.

It's a good tip for those who haven't tried it yet.

Kind regards,
Bram

[TheSteve]

Have you experimented at all with some high power HF signals? I have my 3458A's a decent distance from my HF rig and amp and the antenna is on a small tower but keying up with even 100 watts on 20 or 40 meters does cause changes in my measurements. I won't even get into what happens when I tx with a kilowatt.

[kj7e]

Have you experimented at all with some high power HF signals? I have my 3458A's a decent distance from my HF rig and amp and the antenna is on a small tower but keying up with even 100 watts on 20 or 40 meters does cause changes in my measurements. I won't even get into what happens when I tx with a kilowatt.

Radiated RF from the Radio, Amps, feed line is almost no issue at all as tested in to a good dummy load (Palstar DL5K), radiated RF from the antennas, even though they are on the other side of the house is an issue.  Depending on what I'm doing I stay off the radio.   Especially 2m/440 with the handheld in the shop.  I fought trying to keep RF out of my 10v reference, got it down 5' distance at half a watt was okay, best I could do.

[ap]

A few words re. my cable sets in addition to the very instructive video:
The standard setup as used in the video is suitable for most applications, including the 3458A (use DELAY as indicated by the quick 3458A setup guide, although I have not measured any big differences). The cable shown is based on PTFE wire insulation, the highest insulation resistance material available (ETFE is another choice for precision cable insulation but is slightly worse), leading to minimized leakage currents. For measurement of very high resistances, the use of a common cable for all 4 wires is not recommended. It is then better to spilt up the setup into two cables, one for the positive and one for the negative signals. Details are e.g. given in the Datron 1281 user manual (4 wire ohms measurement instruction section), the 1281 even has a second guard signal for the high side to minimize leakage.
The cable not only dampens electric fields through the shield but also magnetic fields (as e,g, generated by mains currents in power lines or transformers) through the internal twisting. The best reduction can be achieved here by using two opposing wires for the positive and 2 for the negative signals, resulting in a two-signal cable built from a 4-wire configuration. This way the active area that any magnetic flux can induce a voltage in is minimized. But a two wire cable, as offered alternatively, is fully suitable too.
Depending on the souce of noise in a lab setup, common mode disturbances can also be an issue. In such cases, a magnetic core can be used close to the meter inputs with the cable wrapped arround a few times. Shrink tube helps fixing it.
And generally speaking, in a precision setup, the surrounding noise sources should always be minimized as general practice of course and ground/earth loops be avoided.

[kj7e]

Very interesting, I downloaded the Datron 1281 user manual and saved just the 4W Ohm connection example pages, attached here.

[hwj-d]

I too ordered the 2-wire twisted, shielded ETFE cable from ap.

That AWG20 cable feels a little bit strong in a positive sense. The result measuring my LTZs are awesome, they are clear like a whistle, now. No touching issues anymore, like you shows in your vid, if I connect the shield to the black minus jacks. Case shielding is floating, not connected in the moment.

From where do you have these nice plastic grips for the hollow bananas?

[Dr. Frank]

That shield connection is simply tied to the internal AGND of the DMM, over the shunt, probably the 100 Ohm one.

So it's not a true GUARD connection, and you could as well connect the shield to the LO jack of the 34465A, to achieve exactly the same shielding effect.

Very instructive, this setup, and your video, anyhow..

To make precision measurements on 10 kOhm and 100kOhm with OCOMP ON and especially a PTFE cable, you also should choose at least 1 second delay for 10k and 5 sec for 100k.
Otherwise, especially at NPLC 10, you would get up to several ppm error due to the high dielectric absorbtion of PTFE.

Frank

PS: I'm nitpicking on the terms 'GUARD' and 'GROUND', I know.

But I'd like to refer to Keithleys, 'Low Level Measurement Handbook' 7th edition, where the guarding and grounding is explained correctly: download.tek.com/document/LowLevelHandbook_7Ed.pdf

Guarding (of the shield) aims to mitigate leakage currents, see page 2-6: 'By definition, a guard is a low impedance point in the circuit that’s at nearly the same potential as the high impedance input terminal.'

Whereas grounding (of the shield) suppresses electrical field disturbances, and the potential of the shield is then, in contrast to guarding, at zero, or GND.

As the 3458A is mentioned also, distinct GUARD and analog ground jacks (the Lo jack, actually) are available, and these can be tied together by the front key.
It is not recommended at all to make a direct galvanic connection from case / earth ground into the measurement circuit.. that always leads to stray loop currents, which can hardly be controlled.