How to measure 1, 10 and 100GOhm resistors?

[jogri]

Hi guys,

i've recently bought a Keithley 610 that seems to have a few defect resistors (the resistance reading is way off when the device is using those resistors). Do you know a sensible way to measure a 1,10 and 100GOhm resistor without having to use another Electrometer? Sure, i could use a bunch of 10Megs in a parallel circuit for the 1G, but that won't work for the 10 and 100G resistors. I also thought about using the voltage drop across the unknown resistor with a bunch of 10Megs in series but i'm not sure if this is accurate enough (my goal is to find out if the resistor is still within its 1% accuracy spec).

Any idea how to tackle this? Not that i would ever use those ranges, but having a fully functional instrument would be nice.

[MadTux]

Transimpedance amplifier with high impdance Opamp such as LMC662 and like 10x10M (accurately measured) in feedback path?
https://en.wikipedia.org/wiki/Transimpedance_amplifier
Put unknown Gohm resistor on input, feed it like 50-1000V and measure output voltage with 6 1/2 digit DMM. Should be rather easy to get better than 1% accuracy.

[Renate]

I guess that no one thought of the obvious.
A gigavolt supply and a clamp on ampmeter.

[goaty]

You could just buy replacement resistors.

[The Soulman]

Supply safely a high voltage (U) across the unknown resistor(R) and wire in series a suitable dmm* on mV range to measure
the current(I), calculate R from the known U and I.

Suitable dmm means one with a known input resistance on mV range, typically only specified as a minimum or typical value.
measure it with another dmm to be certain.
Also be save with higher voltage's, respect maximum voltage to ground as specified for many equipment and cables.

[The Soulman]

And also, in dealing with giga Ohms the slightest contamination can throw your value's off, that includes resistor body and teflon stand-offs.

[DaJMasta]

Electrometers/picoammeters are really the only viable method for devices not designed for high voltages.  Sure, 10kV regulated would let you measure the current and do the math on a regular meter, but you're talking so high resistances that at a 1V or 10V test voltage (or anything else that a reasonable bench supply may manage), you really need to be measuring nanoamps accurately, at least, to get a good reading.


Something like a 6.5 digit or more meter with a good low range could at least give you a pretty good reading, as would an SMU as they typically will read nA and below without an issue, but you need something sensitive unless you're going to throw at least a few hundred volts through it.

[The Soulman]

Think of it as a Voltage divider,

unknown resistor 100.000.000.000 Ohms     100,0 Volt
dmm.                          10.000.000 Ohms        0,010.00 Volt  (4,5digit dmm on 200 mV range still gives 0,1% resolution)

High voltage is not always inevitable. 

[magic]

Supply safely a high voltage (U) across the unknown resistor(R) and wire in series a suitable dmm* on mV range to measure
the current(I), calculate R from the known U and I.

Yep, I have used this exact trick to measure some 1~100G resistors and got plausible and repeatable results.

The circuit:
100V - the resistor - standard 10MΩ DMM - ground

The math:
10MΩ / DMM reading * (100V - DMM reading)

For 100V I whacked together a simple generator consisting of voltage multiplier, series resistor and zener.

[jogri]

Thanks for the answers, i think i'll do the HV/DMM approach first as i already have the equipment for that. My 34970A can handle up to 250V, i am not going to apply a higher voltage just in case the resistor decides to break down while measuring. I think i could even get it to measure the applied voltage and voltage through the DMM at the same time and give me the resistance directly.

@Goaty: Why don't i just buy replacement parts?
-Because i want to avoid having to open it again in case the resistors aren't faulty. This unit is extremely sensitive and i want to keep the contamination to a minimum. I'll measure the resistors while they are still installed (it should be able to withstand 100V, the manual states that you shouldn't touch the input as it can reach up to 120V for high resistance measurements).

@The Soulman: I am going to wash the resistors with iPA to make sure they aren't dusty or have fingerprints/fat from the previous owner trying to find the fault.

[David Hess]

Transimpedance amplifier with high impdance Opamp such as LMC662 and like 10x10M (accurately measured) in feedback path?

I would probably try using an integrator instead with a capacitor replacing the feedback resistor and the integration constant calibrated using a known good input resistor.  The LMC6081 or LMC6001 are the precision versions of the LMC662 although what really matters is low frequency noise.

[floobydust]

I found any setup using a series multimeter and HV DC power supply quite susceptible to AC hum and interference, at the 10-22Meg level.
Your bench has to be electrically quiet, ground yourself and keep noisy LED/CFL lights away as well as any mains power cords.

[David Hess]

I found any setup using a series multimeter and HV DC power supply quite susceptible to AC hum and interference, at the 10-22Meg level.
Your bench has to be electrically quiet, ground yourself and keep noisy LED/CFL lights away as well as any mains power cords.

I never had any problems measuring 1 gigohm resistors against the 10 megohm input resistance of my digital voltmeters that way.  The multimeters were set to measure voltage so their 10 megohm input resistance acted as a current shunt.

The integration time of a good multimeter is selected to be an integer number of power line cycles so that power line EMI is rejected.

[The Soulman]

I found any setup using a series multimeter and HV DC power supply quite susceptible to AC hum and interference, at the 10-22Meg level.
Your bench has to be electrically quiet, ground yourself and keep noisy LED/CFL lights away as well as any mains power cords.

I never had any problems measuring 1 gigohm resistors against the 10 megohm input resistance of my digital voltmeters that way.  The multimeters were set to measure voltage so their 10 megohm input resistance acted as a current shunt.

The integration time of a good multimeter is selected to be an integer number of power line cycles so that power line EMI is rejected.

Also take care of cable routing, don't make a loop antenna.
Another tip, when in need of a low noise floating voltage supply, snap together a bunch of 9V block battery's. 

[floobydust]

I found any setup using a series multimeter and HV DC power supply quite susceptible to AC hum and interference, at the 10-22Meg level.
Your bench has to be electrically quiet, ground yourself and keep noisy LED/CFL lights away as well as any mains power cords.

I never had any problems measuring 1 gigohm resistors against the 10 megohm input resistance of my digital voltmeters that way.  The multimeters were set to measure voltage so their 10 megohm input resistance acted as a current shunt.

The integration time of a good multimeter is selected to be an integer number of power line cycles so that power line EMI is rejected.

Last I used 5GIG test resistor and 22MEG input impedance multimeter, at 100V I think that's about 20nA which is extremely low.
NMR (normal mode rejection) on DCV is 60dB for most multimeters I have. Not sure exactly but I gave it the hand test and readings move around due to capacitance when moving your hand (as expected), so I might have confused that with AC noise.
But I recall grounding me or keeping a hand steady near the wires/resistor-under-test affected readings.

[magic]

The sensitive node has only 10MΩ impedance, thanks to the DMM. You can see in advance how much noise you will get by connecting the input cable and leaving it open.

Effective E-field shielding can be arranged with grounded tinfoil and some insulator to keep it from touching things. Juice boxes combine both parts in one convenient material

[Zero999]

The multimeter method works well, other than the fact that the input impedance of some multimeters isn't that well controlled, but that's easy to calibrate out by using another mulitmeter to measure it or measuring a voltage, with and without a 10M series resistor and calculating the actual input impedance.

[Smith]

I usually use a Megger for quick measurements, and a HV supply (mostly use 1kV)  plus a picoammeter for more accurate measurements. Fortunately I have an electrometer at home.