100G -1000G Resistors

[jwet]

Looking for very high value resistors- 1 TOhm ideally with a 5kV rating, <=1/4w fine, <=250 ppm/C and decent voltage coefficient.  I can stack a couple but the capacitance quickly gets out of hand.  I plan to air wire this stuff on Teflon standoffs but don't want to resort to anything more exotic.  The application is high impedance voltage measurement around 2kV nominal- consulting job.

Stackpole, Murata, and Ohmite make the kinds of parts I want but they appear to be unobtanium or are $100/unit.  Victoreen, Cleveland used to make parts like this but they got bought by Fluke and I can't find them anymore- maybe internal use only.  New Old Stock in decent shape would be fine if they can provide 100'+s.  One time build of a few hundred units is most likely.  I've looked on Ebay and see some very old NOS Russian/Bulgarian beauties from early cold war, probably 70 years old.  Qtys are low but am asking suppliers, will buy some to test with.  The Bulgarian units are sealed in glass, tolerances are loose (5-10%), no specs other than ohms and quantities are limited (so far).  Does anyone have any experience with these- photo below.  They're about $5 in singles which is in the ball park.  I used to get these from Victoreen for about $1. 

Any other suppliers, leads or ideas would be appreciated.

[Benta]

Did you try Vishay (didn't check myself)?

[TimFox]

The Bulgarian unit shown is very similar to the older Victoreen deposited-carbon, sealed-in-glass units that were used in systems such as Keithley electrometers.
I'm not sure they were rated for 5 kV.
A modern source for 5 kV units is Ohmcraft  https://ohmcraft.com/wp-content/uploads/2022/02/HVR-datasheet.pdf  but I don't know their pricing or minimum order.

[jwet]

Thanks- funny you should mention Keithley- another Danaher, Tektronix, Fluke, Victoreen, Fortive?  head spinning acquisition circle.  As a Maxim FAE, I used to love going to Cleveland for Victoreen, Keithley and AB/Rockwell - I did several studies of industrial electronic companies and Cleveland was ground zero- great products and engineers.

Did Keithley make those resistors in house?  Victoreen had them made by Dale/Vishay I think but also sold them as components to get their volumes up to where Vishay cared.

[wraper]

The Bulgarian units are sealed in glass, tolerances are loose (5-10%), no specs other than ohms and quantities are limited (so far).  Does anyone have any experience with these- photo below.

This is not Bulgarian but Soviet and has datasheet.
https://www.quartz1.com/price/PIC/480Q0717900.pdf
http://www.155la3.ru/datafiles/kvm_kim_klm_tu_1977.pdf
You can google translate it and ask if something is not clear, I speak Russian. КВМ are vacuum composite resistors. +1000/-2000ppm/^oC TC, -60/+85^oC, 100V max. Rated to drift not more than 30% during 10k hours of operation and not more than 25% during 12 years of storage.

[TimFox]

I thought those resistors were made by Victoreen, but used in Keithley equipment (from the same city).
An interesting 1965 article in Rev Sci Instrum mentions Victoreen electrometer resistors on page 4  https://escholarship.org/content/qt3v68c7vw/qt3v68c7vw_noSplash_eac5c3bdc7d1f17286456241063f249c.pdf
However, that circuit has maybe 10 V across the Victoreen range resistors (max 24 V supply).
In grad school (1970s), we had a good supply of high-voltage resistors made from a helix of carbon on ceramic pillars (not precision, but good voltage ratings), from Resistance Products Co, now part of Vishay/Dale.
Unfortunately, that series seems to go up to only 500 megohm.   https://www.vishay.com/docs/31039/bt.pdf
None of these carbon resistors will have a good tempco or voltage co-efficient of resistance.
Even the Caddock resistors have a noticeable co-efficient at these voltages.

[Someone]

vacuum composite resistors. +1000/-2000ppm/^oC TC

Its a nasty problem, air is a G ohm resistor, but vacuum kills the thermal performance. Porcelain/glazed resistors + dry purged enclosure would be a start. Honest suppliers note the TCR and VCR ratings fall apart at higher voltages:
http://www.token.com.tw/resistor-hv/index.html

[pcm81]

vacuum composite resistors. +1000/-2000ppm/^oC TC

Its a nasty problem, air is a G ohm resistor, but vacuum kills the thermal performance. Porcelain/glazed resistors + dry purged enclosure would be a start. Honest suppliers note the TCR and VCR ratings fall apart at higher voltages:
http://www.token.com.tw/resistor-hv/index.html

Funny you mentioned that. At work we have to do certain bonding/ground tests in an environmentally controlled room, because high humidity air, here in Florida, can make a 10GOHM test fail.

[Someone]

vacuum composite resistors. +1000/-2000ppm/^oC TC

Its a nasty problem, air is a G ohm resistor, but vacuum kills the thermal performance. Porcelain/glazed resistors + dry purged enclosure would be a start. Honest suppliers note the TCR and VCR ratings fall apart at higher voltages:
http://www.token.com.tw/resistor-hv/index.html

Funny you mentioned that. At work we have to do certain bonding/ground tests in an environmentally controlled room, because high humidity air, here in Florida, can make a 10GOHM test fail.

Some systems a dry nitrogen purge isn't adequate! then you're off into the weird and wonderful.

[HighVoltage]

I have been using these in the past:

100GOhm KVM Russian military vacuum glass resistor NOS
https://www.ebay.com/itm/265524132036

[magic]

This is not Bulgarian but Soviet and has datasheet.
https://www.quartz1.com/price/PIC/480Q0717900.pdf
http://www.155la3.ru/datafiles/kvm_kim_klm_tu_1977.pdf
You can google translate it and ask if something is not clear, I speak Russian. КВМ are vacuum composite resistors. +1000/-2000ppm/^oC TC, -60/+85^oC, 100V max. Rated to drift not more than 30% during 10k hours of operation and not more than 25% during 12 years of storage.

This is correct, including the voltage rating, and I seem to recall that people tested them and found them to fail instantly at less than 1kV.

I have a few of those resistors and their tolerance after all those years is quite sloppy, particularly the 100GΩ ones. Most were 90~95GΩ and the worst ~70G, all measured by applying 100V in series with a 10MΩ DMM and doing the obvious math.

[magic]

I can stack a couple but the capacitance quickly gets out of hand.

I don't understand this remark. Connecting capacitors in series decreases end-to-end capacitance, and resistors have low capacitance to begin with.

In fact, I think a physically long string of 1GΩ resistors would have much less capacitance than one small 100GΩ resistor.

[jwet]

Tim Fox- I could be wrong about who made Victoreen R's- I was told this somewhere along the line.  Victoreen had a lot little specialty shops in that big place near Shaker Heights.  There was a little shop for making those view through electrometer based dosimeter pens, quite an operation- saw it around 1986.

[jwet]

Magic- My capacitance concern is node capacitance to "ground", not across the R's- even on a teflon standoff, its difficult not to pickup a pF of stray.  1000G x 1 pF makes for s 1 sec RC product, even at lower resistances like 10G,  its easy to get into motor boating- low frequency oscillation in anything with feedback.  Guarding can help but it all gets kind of ugly at these voltage and currents.

[jwet]

pcm81- I think something like you're talking about is what my customer wants to do but he wants to do the test in the field.  He doesn't want to issue his field tech a fancy Keithley electrometer and the cheaper megger like gadgets available won't go down low enough in leakage.  End equipment is Ion mobility spectrometers- the detectors used at airports for explosives.

I'm going to rethink my approach.  When you get yourself designed into a box like this, its best to back out and try something else.  I originally wanted to do it with an integrator but they didn't like the slow reading speeds it would give them (1/sec).

Thanks all.

[magic]

Not sure what sort of circuit this resistor goes into, but Johnson noise and stray capacitance may make significantly faster speeds infeasible anyway, so the customer may simply need to adjust his expectations.

[TimFox]

Not sure what sort of circuit this resistor goes into, but Johnson noise and stray capacitance may make significantly faster speeds infeasible anyway, so the customer may simply need to adjust his expectations.

The article I linked above goes into great detail about optimizing the speed of a sensitive electrometer circuit with these resistors.

[TimFox]

Tim Fox- I could be wrong about who made Victoreen R's- I was told this somewhere along the line.  Victoreen had a lot little specialty shops in that big place near Shaker Heights.  There was a little shop for making those view through electrometer based dosimeter pens, quite an operation- saw it around 1986.

I looked at the BOM (1977 manual) for my Keithley 261 Picoampere source (precision voltage source driving selectable high-value resistors), designed to calibrate electrometers.
The "deposited carbon" resistors have manufacturer's code 80164 (Keithley), but that may be since Keithley measured them individually for calibration.

[jwet]

Tim Fox- I just started digging into that 1965 Article- thanks very much!  You're kind of on your own with a lot of this esoteric stuff.  Keithley made a low level measurements handbook that is very good too and I have it but its more about appplications (selling instruments) than design.

[jwet]

Magic- I agree, I think they're hoping for >5 per second for pot tweaking or other adjusting.  At these currents, you're getting down to statistical currents sort of like shot noise.  I'm in the proposal and discussion phase now- I haven't got a contract yet.  I have to be fairly polite still...  I might pass- I have a week to decide.

[jonpaul]

OP: The combination you spec is impractical.

The tradeoffs of voltage coeff, R, precision at over 1 G Ogm are unobtainable.

We used Victoreen and Caddock in 1980s, on 12 kV Avaionics HV.

The 1000M were +/- 20%

Rated to 20 kV.

I would revise your   approach to avoid the unreachable requirement.

Jon

Jon

[magic]

By the way, I still don't know what you want to use this resistor for, but if it's just low level current measurement like in the article linked by Tim Fox then I don't see why you would need a 5kV rated part.

At first I thought you want it for a high impedance voltage divider or something like that.

[PartialDischarge]

2kV is not that high, so you could use a follower opamp connected to the input and floating (bootstrapped) at the input voltage. Bandwidth is going to be high, but also is going to be the cost as you need a DCDC converter that gives 2KV although low current

[strawberry]

2kV / 1Tohm = 2nA

[jwet]

Magic- I explained as much as I could above.  The box will measure voltages up 2500V with very light loading and  measure very low currents (10e-14A) for their field techs that service Airport Ion Spectrometers- explosives detectors.  Lab gear exists that can do all this but it has a lot of features that they don't need and costs about 4k and is not easy to source.  They are doing a make/buy analysis to see if they can get something custom designed at 150 pieces for less than the cost of buying 150 $4K instruments, about 1/2 million and not really up to the task of knocking around in a tech's van.  Their current solution is to FED-X a Keithley and an elecltrostatic voltemeter out to the field which is a pain too. This ultra low current, high voltage stuff is a funny niche that I did a while ago.  I'm doing a 40 hour trade study right now to give them their options and they'll go decide.  I'm only on the hook to deliver a study and this one of several avenues.  I'm also looking at other consultants (better than myself) that specialize in this field.  We'll see.

Thanks all.

[bdunham7]

Lab gear exists that can do all this but it has a lot of features that they don't need and costs about 4k and is not easy to source. 

Just curious--what is the $4k candidate that can measure 2.5kVDC with a 1T input impedance and also measure current with 10fA resolution?

[strawberry]

 FET input capacitance fc = 1/(2*PI*1Tohm*10pF) = 0.0159Hz ~100s settling time
bootstrap preamp

[jwet]

bduham7- The customer won't tell me what they think is the $4k solution until after the study! I suspect its some kind of semi custom Keithley solutions like a modified 6500 type meter- some of these have a 200T Ohm input Z but only a 200V input range for DC voltage (divider?).  The old and still widely used solution in the lab is an electrostatic voltmeter mostly made by Sensitive Research- real old timey analog voodoo- they use a mirrored scale analog meter and come in a faux walnut box.  They're actually somewhat plentiful surplus but should never leave a lab, very fragile.  Input Z isn't measurable.

strawberry- that is the fundamental physics problem, you've got it- the C has be reduced by 1000x or more.  Like everything in the modern world, people have been chipping away at these tough problems for a long time.  Forget 1 pF, you need something that is .1 pf to start and that needs to be knocked down with circuit techniques.  Passive or Active guarding (bootstrapping), nulling voltmeter techniques and other techniques are used.  Look at the paper from 1965 that Tim Fox posted early in the thread- this adresses the speed problem pretty well.  This is not a first order or even second of third order type of problem.

Thanks.

[bdunham7]

bduham7- The customer won't tell me what they think is the $4k solution until after the study! I suspect its some kind of semi custom Keithley solutions like a modified 6500 type meter- some of these have a 200T Ohm input Z but only a 200V input range for DC voltage (divider?). 

The only thing I haven't seen--and perhaps I missed it--is what they expect for precision.  This doesn't sound like a $4K field device to me, but I suppose if it only has to have a single or even a few ranges and someone has been quite clever, it might be possible.  Perhaps an entirely different sort of design would work, like an active differential system that floats up a more sensitive meter to within a few volts of the target.  I'd want to see it to believe it.

[jwet]

Their stated goals are 2% basic accuracy on volts full scale on volts and 20% at .1 pA current, but there are other conditions - like Mil Std 810- shock and vibration- this will allow it to be hand carried and shipped on planes in checked luggage and not require a transit case.  Why not ask for the world is their philosophy- at least they're letting a paid market study contract.

I don't know if you've ever seen a nulling voltmeter- Fluke made them before sexy DMM's- they were kind of cal lab oddities.  On one side of a null detector was a HV source with a 5 or 6 decade kelvin varley voltage divider and on the other was the unknown input.  The null detector between the two drew very little current (zero if possible) often an electrometer was used.  After you balance things out - you've measured your voltage- digitally!  The null detector is also the current meter.   This isn't really much different than how an opamp works really and you don't have to do it at high voltage if you scale the cancelling voltage source, etc.  Long lever, short lever.  The nice thing about the nulling is its like a inverting op-amp where the common mode voltage is  near zero and the bias currents can be controlled with DC bootstrapping, like putting a resistor in the non inverting input equal to the feedback R only active.  There are also more clever methods with multiple stages of this stuff.

I posted the resistor question because this is one issue that is key to in an house first order brute force build- can I get an off the shelf 1 T-ohm resistor that didn't suck? I suspected No and I am more sure now that the answer is no and brute force won't get it done.

I'll post something when the dust settles, knowing how these things go, it will take them 2 years to decide and after I've completely forgotten about it, They'll want to do the worst option in a month for half of what I estimated 2 years earlier....

[jonpaul]

get a Sensitive Research  Electrostatic Voltmeter 

Avail as surplus, $40..250, in ranges 5kV ...100 kV, zero current.


Jon

[jwet]

Need quantity 150 and long term support for field service people.  I mentioned this above as an electrostatic voltmeter- a standard solution for 1/2 the problem but also pretty fragile for a tech servcing airport equipment.

[bdunham7]

not require a transit case.

Now they're getting silly.

I don't know if you've ever seen a nulling voltmeter- Fluke made them before sexy DMM's- they were kind of cal lab oddities. 

Yes, that is what I was thinking of, except that if they are always reading the same voltage you could simplify the design a bit.  It's not hard to design a device to put out a fixed or adjustable 2-2.5kV by using a lower-impedance feedback design, then null your measured voltage to it.  Insulation testers come to mind as a source.   I still have no idea how you could produce a tested and certified product like this in that low of a volume for well under $4k.  It's good they're paying you upfront.

[jonpaul]

There are small cased ESVM with 10..20 kV ranges for field use. They are rugged and are on ebay perhaps $40..100 ea


Jon

[jwet]

bunham7- I agree with everything you and others have said.  I even think that the customer knows its semi impossible.  They just can't buy something this expensive without a pretty good paper trail.  I think I'm a pawn probably but its ok- its interesting and has low risk- the only deliverable is a report.

The voltage nulling technique is looking pretty good and I think I get the electrometer for free- I'm supposed to show them all reasonable solutions with analysis of pros and cons.  One nice second order thing about the voltage nulling is that it reduces the effects of voltage coefficient as there is no voltage across the big input R.  I have found some very high value R's but the VCR's are such as to create huge errors (30%+!) if not cancelled with this nulling.  The best are 250 ppm/V- time 2kv gives you 50%!  This is not a brute force problem.

jonpaul- thank you, I have one in a faux wood box but I'll look on line at the ruggedized ones.  They do use things like my model now but they ship them to the techs and the techs ship them back to a cal lab on return for surveillance and checks.  Thanks

[TimFox]

Back in graduate school, we needed to apply about 10 kV to a test object and measure the (small) current through it.
We floated a battery-powered Keithley electrometer in a purpose-built acrylic enclosure, with acrylic rods to rotate the control knobs.
The case of the electrometer was connected to the power supply, so that any leakage or corona from the rectangular case did not flow through the electrometer's ammeter circuit.
To protect the input of the electrometer (a vacuum-tube input unit, with reasonable input voltage rating), we added an appropriate diode network across the input.

[exe]

some very old NOS Russian/Bulgarian beauties from early cold war, probably 70 years old.

Oh, those can have very bad tempco. Hope it doesn't matter for your circuit. Afaik, somebody evaluated them for a (pico-)ammeter application, but I can't find the thread. Also worth cleaning and drying it properly, any surface contamination ruins precision.

[Someone]

not require a transit case.

Now they're getting silly.

It ends up with the "product" they order being a branded pelican case factory fitted with the actual product inside. Several employers have done exactly this.

[jwet]

Tim Fox- That's pretty cool.  My first job was at General Atomics in San Diego in 1982.  As a new grad, I was supposed to work in three groups for 6 months each- they called it the Tech Grad Rotation program.  They were an aging company with a lot of Manhattan project type guys in the hierarhcy.  They were trying to get some young blood and were hiring a few dozen grads a years from good engineering schools.  General Atomic is still around and thriving- they did the  Predator drone- (good story).   My first rotation was in an instrumentation group that did radiation monitoring equipment for Nuclear Power Plants- our system was installed in most Westinghouse PWR's around the world- over 100 sites with 100's of monitors per system, all networked together in 1982- super state of the art-  software, digital, analog, physics and instrumentation problems.  I loved it and couldn't believe I was getting paid to go play there everyday.  I was in the Detector Physics group, working on analog front ends for all different kinds of radiation detectors- GM Tubes, Ion Chamber, Fission Chambers, PMT/Scintillator, CdTe-Solid State- the gamut.  I had some great old time mentors that had seen it all.  There were also a lot of commercial Nuclear Power guys mostly brought up through the Navy.   My next rotation was supposed to be to their Fusion Research Reactor called Doublet 3- a big Tokhamak that was hot shit for its time.  About a month before I was to transfer, I took a big tour and saw these guys operating instruments behind 1/2" lexan blast shields with3 foot lucite rods to turn scope knobs, etc like you're talking about.  I decided to stay put in my group and go out of the Tech Grad deal.  I've done a lot of ion detector and PMT work up to about 2 KV and have been on projects with 200 KV X-Ray power supplies.  That stuff gets weird.  I settled down to precision data acquisition and mixed signal stuff and after Nuclear dried up went to work for Maxim for about 25 years in their standard products group defining products like Analog Swtiches, Op-Amps, etc.  A lot of fun.  Retired a while back taking little consulting projects that come along.

[ZhuraYuk]

Tried to get similar 1T resistors to replace them in some old electromters. There are few options for cheap in stock:

https://mou.sr/4a3v7Sg   Mouser 1T Ohm 500ppm 5%

https://www.aliexpress.com/item/1005006645803409.html There is a lot of new Chinese high ohm resistors, all seems to be made of same material 200ppm.

https://twjohmresistor.aliexpress.com/store/912607391 This shop has large choice of Giga Ohm resistors to upgrade yuor old electrometers.