Making a 1G Ohm industrial accuracy standard resistor for the workbench.

[GEOelectronics]

Project goal- build an industrial grade accuracy 1G Ohm shop standard resistor to help compare/sort stock 1G resistors for other projects (Ohmite 1G 1% Slim-Mox 1 and 2 watt versions).

The standard resistor is built with 10 ea. 99M Ohm 0.1% HV resistors in series contained in a temperature lag filled with Dow dielectric compound. Trimmed with an auxiliary 10M series resistor to 1G Ohm for when used in it's primary function.

A convenient daily secondary function is to use it at 990M, as a HV divider for a 10M input digital panel meter set up to monitor a regulated 1kV bench power supply.

Here is a picture of the 990M module in the silicone bath for further testing.

Nothing fancy, but this will bring my workbench up a notch in the high-impedance HV and High MegOhm measuring accuracy department.

George Dowell
K0FF

[GEOelectronics]

Thanks for the reply, any insights, experience, suggestions, procedures on the subject would be welcome and appreciated thanks.

I do have access to some HV generators and High M Ohm resistance meters and bridges, but no calibrated electrometers.

George Dowell

[GEOelectronics]

Dow dielectric compound

How you measure resistance of:
* Dow compound.


Good question. It has pretty good specifications:
Dowsil DC-4 / Dow Corning® 4 Electrical Insulating Compound -

https://www.ulbrich-group.com/chemical-technical-products/TDS_DOW_CORNING_4_eng.pdf


especially
CTM 0114 Dielectric strength, 1.27mm gap V/mil 460 (460V per millii-inch)

CTM 0249 Volume resistivity at 23°C (73°F) Ohm.cm 1.1 x 1015

Well beyond my measurement capabilities (I think).

George Dowell

[GEOelectronics]

Dow dielectric compound

"Good way - remove BNC's, use separated PTFE insulated wire.

If you want coaxial - use only high quality shielded cable with strong and fat PTFE insulation, to remove EMI effects and keep wire-to-wire resistance always high. Try not use any BNC connectors(or use connectors only with PTFE).

"

Good point and I will do that today, thanks.

George Dowell

[GEOelectronics]

PS. see that video from Transmille https://youtu.be/5LqH_GyV6Ug

That video says it all.

Will do much more testing soon, thanks.

George Dowell

[GEOelectronics]

PS. see that video from Transmille https://youtu.be/5LqH_GyV6Ug

Testing was interesting. Do you know the HV Voltage used in the video's instrument?

My insulation tester is 500V, measures to 1000 M Ohms. It has an analog dial so is only good for rough testing. But of good quality for an industrial meter, made in Japan.

After an overnight measuring run on my DMM, the test setup (all new made by Pomona or HH Smith) in my picture is stable at 899.995  989.995 M Ohm on the DMM, just what was hoped for and expected.

Then, removed from the DMM, test lead + resistor stack setup as in my picture, attached to the insulation tester reads ~1000 M Ohms. Without the resistor stack it reads infinite.

Then I tried reading some of my workbench test leads, red to black wire, twisted together like in the video,all test infinite at 500V, even the 1.98 USD Harbor Freight test leads, all read infinite.

Tomorrow I'll break out the HP/ Yokogawa teraohmmeter and test everything again
@ 1kV.

Thanks.

George Dowell

[GEOelectronics]

PS. see that video from Transmille https://youtu.be/5LqH_GyV6Ug

Testing was interesting. Do you know the HV Voltage used in the video's instrument?

After looking at the video again and researching the instrument used in the video, I can see now that the HV used was 300V, but the instrument used was an electrometer.

George Dowell

[trobbins]

It is sad that the video just glossed over the effect of bundling of the test leads - and categorically stated that those leads could not be used for high accuracy testing - which is a poor summary, as that type of lead can be used, but with an appreciation that they should not be touching. 

It would also be appropriate that any terminal(s) have sufficient voltage rating/insulation resistance for (a) any test equipment that may apply a high voltage, and (b) the level of reference resistance trying to be measured.  And hence BNC connection would have to be a concern.

[GEOelectronics]

It is sad that the video just glossed over the effect of bundling of the test leads - and categorically stated that those leads could not be used for high accuracy testing - which is a poor summary, as that type of lead can be used, but with an appreciation that they should not be touching. 

It would also be appropriate that any terminal(s) have sufficient voltage rating/insulation resistance for (a) any test equipment that may apply a high voltage, and (b) the level of reference resistance trying to be measured.  And hence BNC connection would have to be a concern.

Yeah, that's pretty much why I included the word "industrial grade" in my project description, not metrology grade : just good safe practice using the best economically feasible materials.

For this standard resistor project I went ahead and mounted it in a Hammond cast aluminum box, coated on the inside with silicone HV "tape" and the connectors are  2 high quality PTFE MHV type.

Testing with the HP 4329A High Resistance Ohmmeter, even at 1kV applied shows correct, stable readings, but is not precise reading enough for my needs, so now its back on the KS 34470A to watch it cool off and re-stabilize after carefully soldering to the connectors.

George Dowell

[GEOelectronics]

Dow dielectric compound

How you measure resistance of:
* Dow compound.
* BNC-Banana adapter.
* BNC-Wire adapter.

?

It increase measurement uncertainty.
Good way - remove BNC's, use separated PTFE insulated wire.

If you want coaxial - use only high quality shielded cable with strong and fat PTFE insulation, to remove EMI effects and keep wire-to-wire resistance always high. Try not use any BNC connectors(or use connectors only with PTFE).

I did test them all, the Pomona BNC to clip lead looked good by many of the dual banana plugs didn't. Replacing them with silver single banana plugs cleared up a lot of the instability.

Now I'll go ahead with the PTFE coaxial soon as I find the roll.

Thanks a lot, I would never have suspected high quality plastic parts like those dual banana would have significant leakage!

George Dowell

[GEOelectronics]

Dow dielectric compound

How you measure resistance of:
* Dow compound.
* BNC-Banana adapter.
* BNC-Wire adapter.

?

It increase measurement uncertainty.
Good way - remove BNC's, use separated PTFE insulated wire.

If you want coaxial - use only high quality shielded cable with strong and fat PTFE insulation, to remove EMI effects and keep wire-to-wire resistance always high. Try not use any BNC connectors(or use connectors only with PTFE).

Lots of tests and improvements since we last talked. I changed the connectors on the 990M standard resistor (DIY of 10 99M 0.1% HV units)n to MHV. The resistor stack is inside a Hammond aluminum box, lined with HV silicone tape.

Testing with short single conductor silicone wires shows are reading of 989.80M Ohms, which is just right where I wanted (selected lower value resistors for the stack so I could trim it later).

This lets me test all my coaxial and other test cables by using an MHV TEE right at the resistor box, and using one of the ports to attach the cable being tested (far end non terminated).

Any change from 989.8 indicates a leakage path.

Come to find out every one of my right angle BNC to BNC male cables were bad. These were about 12 inches long and had high quality PTFE coaxial cable. I cut them up and tested them on a HP-Yokogawa High Resistance meter at 1000V . Their leakage came from the BNC connectors not the cable. Cut up about 5 of the "cleaner looking" ones before testing the one black covered one, which had no great leakage at all. Evidently somewhere along the line the clean ones were also black but cleaned by some method by a technician, which left some sort of residue inside.

Now I've set up a desk just for resistor testing with a Guarded Wheatstone bridge, an HP-4329 Hi Resistance, HP-4328 milliOhm meter, an ESI 232A high resistance guarded Wheatstone bridge (12 G Ohm capability), the usual Megger etc. Teflon sheeting, shielded boxes in the works, and so on as I learn more.

Now I can test all sorts of cables, actual resistors, wires,  etc. The question now becomes, what is considered a low loss for a test cable? The Pomona BNC to clip leads adaptor came in at above 85 or more X 10^ 12 Ohms. (HP 4328A4329A)

At this time I am making new pairs of test cables from Belden RG179 TFE cable. 2 x  6", 12" 18" 36". Even the 6" one shows a slightly different reading when used on the test jig made from my 990M resistor (KS 34470A)- the difference is from 989.8 down to 989.47 which seems to indicate a shunt resistance in the cable.

Next to try to calculate that shunt resistance in G Ohms.

George Dowell

[GEOelectronics]

It is sad that the video just glossed over the effect of bundling of the test leads - and categorically stated that those leads could not be used for high accuracy testing - which is a poor summary, as that type of lead can be used, but with an appreciation that they should not be touching. 

It would also be appropriate that any terminal(s) have sufficient voltage rating/insulation resistance for (a) any test equipment that may apply a high voltage, and (b) the level of reference resistance trying to be measured.  And hence BNC connection would have to be a concern.

Good points. Now that all my resistance measuring tools are in one place, it was easy to repeat the test of the "ordinary" test leads, and yes, there is definitely measurable leakage between them at 1000V. What this very high resistance means to the meter depends on the meter's input impedance. A normal 10M Ohm meter wil never really notice, is what I think. He was using his meter in "electrometer" mode, which is a fantastically high input impedance, so yest those leads would be worthless in that mode. But on the other hand, my brand new shop made Teflon coaxial cables also have measurable leakage, which is why I posted the update just previous about what is considered "good enough" or actually obtainable.

George Dowell

[ArthurDent]

Here's a photo of a 1000 Mohm resistor I have made by Victoreen.

[GEOelectronics]

Those are nice, thanks for showing. Have you measured them for accuracy or drift?
I do have a precious few by Victoreen in glass like that, with some fantastically high resistances marked!