Good day to any and all! I was wondering about a phenonenon occurring with high resistance measurement with ohmmeters and multimeters that has baffled me for years. And that is on the instability of measurements at very high resistances (namely >100 Megohms also into Gohms and even Tohms). Has anyone noticed that multimeter readings usually becoming unstable and float up and down continuously between the set point. For example, if I'm measuring 500 MΩ with a Fluke 179 multimeter, it will undulate to say 497.7 MΩ then slowly up to 503.1 MΩ and then back down again in a slow repeating wavelike fashion. I have noticed this effect in almost every handheld multimeter and to some degree even in higher quality benchtop multimeters. Does anyone know why this occurs and why it's so difficult to stably measure extremely high resistance? Thanks.
Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
Largest Manufacturing Technology Community on the Web
Stay Connected:
Measuring High Resistance and Commonly Unstable Readings
- Thread starter Natronius
- Start date
- Replies 25
- Views 18,347
Mark W. Ingalls
Aluminum
- Joined
- Jun 5, 2012
- Location
- Little Rock
Dear Natronius-
When you are trying to measure such high resistances, and particularly at such low voltage levels, you are basically counting electrons, y'know?.
So, there are many sources for error, including noise, but what you are describing sounds more like a possible averaging function built into the DMM?
In order to make *accurate* low current measurements, e.g. TOhm, you need to employ special techniques, such as four-point Kelvin probing ("force voltage, measure current"). Here is a great resource from Keithley: WB/259: Request your FREE Low Level Measurements Handbook - Keithley Instruments Inc.. I personally think they are one of the best for this type of equipment.
Be well,
Mark
When you are trying to measure such high resistances, and particularly at such low voltage levels, you are basically counting electrons, y'know?.
So, there are many sources for error, including noise, but what you are describing sounds more like a possible averaging function built into the DMM?
In order to make *accurate* low current measurements, e.g. TOhm, you need to employ special techniques, such as four-point Kelvin probing ("force voltage, measure current"). Here is a great resource from Keithley: WB/259: Request your FREE Low Level Measurements Handbook - Keithley Instruments Inc.. I personally think they are one of the best for this type of equipment.
Be well,
Mark
The real Leigh
Diamond
- Joined
- Nov 23, 2005
- Location
- Maryland
four-wire (Kelvin) connections are used for making low-resistance measurements, where the resistance of the test leads and the resulting IR drop can affect the accuracy of the readings.
High-resistance readings need to use isolating and guarding techniques to reduce sensitivity to environmental noise.
Any decent high-range ohmmeter will have a Guard terminal in addition to the input terminals. Check the instructions for proper use.
Also, accurate measurement requires a reasonably high voltage.
The HP 3456A 6½-digit DVM compliance voltage is 5.5v on the 100MΩ and 1GΩ ranges.
Meters with lower native sensitivity use higher voltages.
Meggers, which are designed to measure high resistance, use a compliance voltage of 500v.
What exactly are you trying to measure? That might influence the choice of equipment.
BTW, the highest resistance range on your Fluke 179 is 50MΩ. Don't know how you're measuring 10 times that.
- Leigh
High-resistance readings need to use isolating and guarding techniques to reduce sensitivity to environmental noise.
Any decent high-range ohmmeter will have a Guard terminal in addition to the input terminals. Check the instructions for proper use.
Also, accurate measurement requires a reasonably high voltage.
The HP 3456A 6½-digit DVM compliance voltage is 5.5v on the 100MΩ and 1GΩ ranges.
Meters with lower native sensitivity use higher voltages.
Meggers, which are designed to measure high resistance, use a compliance voltage of 500v.
What exactly are you trying to measure? That might influence the choice of equipment.
BTW, the highest resistance range on your Fluke 179 is 50MΩ. Don't know how you're measuring 10 times that.
- Leigh
Mark W. Ingalls
Aluminum
- Joined
- Jun 5, 2012
- Location
- Little Rock
Dear Leigh, et al.-
I looked at Keithley AN 312, "High Resistance Measurements." If you force current and sense voltage in a (differential) four-wire setup you get an accurate high-R reading. I apologize for switching the force-sense sequence earlier.
Mark
I looked at Keithley AN 312, "High Resistance Measurements." If you force current and sense voltage in a (differential) four-wire setup you get an accurate high-R reading. I apologize for switching the force-sense sequence earlier.
Mark
The real Leigh
Diamond
- Joined
- Nov 23, 2005
- Location
- Maryland
Four-wire measurement of high resistance values is only used in specific special circumstances.
The example they give on page 2-36 is measuring the resistivity of semiconductor materials, where the resistance between the
probe and the material being tested is quite high.
The four-wire technique is not used when measuring high-resistance values in the general case
unless shielded test cables are used, in which case the shield is guarded.
Refer to the section on guarding starting on page 2-40 of the Keithley manual.
Unshielded test cables are generally preferred if environmental noise conditions permit because
they eliminate the time constant associated with shield capacitance.
- Leigh
The example they give on page 2-36 is measuring the resistivity of semiconductor materials, where the resistance between the
probe and the material being tested is quite high.
The four-wire technique is not used when measuring high-resistance values in the general case
unless shielded test cables are used, in which case the shield is guarded.
Refer to the section on guarding starting on page 2-40 of the Keithley manual.
Unshielded test cables are generally preferred if environmental noise conditions permit because
they eliminate the time constant associated with shield capacitance.
- Leigh
No, actually not. But, for the reading to continuously, but slowly change in one direction is almost always the case, and is referred to in the literature as "polarization." Fluke has several manuals that explain high resistance measurements in considerable detail that you can find with Google.
A separate problem is, how do I know the 127 GOhm that my bridge circuit -- yes, Leigh, it is a bridge circuit -- is measuring actually is 127 GOhm, and not something else, like 103 GOhm? Simple, you say, check it with a calibration resistor. Ah, there's the problem. With the arms of the bridge balanced on its highest resistance range my General Radio "Megohm Bridge" will measure 1000 TOhm, but the largest calibration resistor I own is only 10 GOhm. It's not that I haven't tried to find resistors with higher values, but I've given up looking. Dale claims they will make them to special order, but in fact they won't. So, up to ~10 GOhm I know my GR bridge is giving me the correct value, but much higher than that I just have to believe...
p.s. My G-R bridge has a number of voltage ranges to select from, from 10 to 1000 V. My two modern digital meters test up to 20 GOhm, one with voltages of 50, 100, 250, 500 and 1000 V, and the other with 500, 1000, and 2500 V. I have an old Biddle-type megohmeter that measures up to 1000 MOhm using 1000 V generated by cranking a handle. The reason for being able to select the voltage is tests at a variety of voltages can often let you identify a possible problem that would be missed by a fixed voltage.
Last edited:
The real Leigh
Diamond
- Joined
- Nov 23, 2005
- Location
- Maryland
You need to buy some higher-value standard resistors, or build them from sets of ten lower-value components.
NIST certifies standard resistors up to 1 TΩ, so they must exist.
Above that, if you want any level of process control, you have to build it yourself. That would be a complex and expensive
process, considering you would need an entire climate-controlled environment (particularly 35±5% RH).
That brings into question your current measurement environment and climate control.
I think the highest standard R we have at the cal lab is 1 GΩ. It's seldom used due to the extreme environmental requirements.
BTW, the Biddle megger is not a particularly accurate instrument, being designed for testing insulation resistance and such.
- Leigh
Ref: resistance measurements
also http://www.nist.gov/calibrations/upload/tn1458.pdf
The real Leigh
Diamond
- Joined
- Nov 23, 2005
- Location
- Maryland
Follow-up on this.
When calibrating standard resistors, NIST uses four-wire techniques for values up to 10,000 ohms.
They use two-wire for 100,000 and 1 Megohm resistors.
For 10 Megohms up to 1 Teraohm they use a three-wire connection, which is two-wire plus guard.
Since they define accuracy (here in the US), I think we can accept their protocols as definitive.
- Leigh
Ref: http://www.nist.gov/calibrations/resistance.cfm specifically Table 9.2
As I wrote previously, Dale lists high value precision resistors in their catalog, but despite that they don't sell them. I was told by someone in their sales department that the listing must be left over from a time when they did. Also, just because NIST offers a calibration service up to 1T doesn't mean any company currently offers such a product for sale. However, if you are aware of a company that does sell precision resistors of value 100 GOhm and higher, I definitely would like to know.
As for building them myself from sets of ten lower value-components, 1000 TOhm is 100,000x higher than the max. 10 GOhm I was able to buy. That is a lot of resistors to string together. Clearly, even if they were 0.0000001% resistors, what you suggest is not a practical solution. Nor is building one myself.
The real Leigh
Diamond
- Joined
- Nov 23, 2005
- Location
- Maryland
Not my job to find products for other people. You need 'em, you find 'em.
Sometimes problems arise that do not have simple solutions.
- Leigh
Sometimes problems arise that do not have simple solutions.
- Leigh
It was a rhetorical "request" based on your statement that since NIST offers a calibration service someone must sell them. I already know that none of the usual suspects do, nor do any of the unusual suspects I could locate.
Even more than that, accurate measurement requires an accurate calibration standard.
Madis Reivik
Hot Rolled
- Joined
- Jan 23, 2011
- Location
- Estonia
Actually, the fluctuation is below 1% which is negligible. Such high resistances are usually voltage, humidity etc etc dependant so dont expect high precision. So use the result, its good enough.
What are you measuring, btw ?
niemi24s
Aluminum
- Joined
- Sep 26, 2010
- Location
- South Of Lake Superior, USA
There's a possibility you may find suitable high resistance components by searching the surplus market. Some years ago a acquired a few 1.485GΩ ±3% resistors, but I've no recollection about where I got them - except it was one of the many surplus electronics places found online. I remember them being advertised as new excess from a maker of radiation measuring equipment.
Regards
Regards
Conrad Hoffman
Diamond
- Joined
- May 10, 2009
- Location
- Canandaigua, NY, USA
What are we trying to measure here? I'd think surface leakage would be a huge problem and pinning anything down at the many Gohm level would be an exercise in frustration. Does the manual for the GR bridge offer any clues, or is there a GR Experimenter with a write-up related to it? My guess is you have to infer calibration in some manner, not use an actual Tohm resistor. I think the GR 716C cap bridge can also do very large resistances, but my guess is that unless they are rebuilt with new insulators the leakage would too high.
The real Leigh
Diamond
- Joined
- Nov 23, 2005
- Location
- Maryland
Another source of error with high-voltage resistance measurements is the "voltage coefficient of resistance".
The actual resistance changes as the applied voltage increases. The numerical value must be obtained from the manufacturer.
This is similar to value changes resulting from temperature shift.
- Leigh
The actual resistance changes as the applied voltage increases. The numerical value must be obtained from the manufacturer.
This is similar to value changes resulting from temperature shift.
- Leigh
jim rozen
Diamond
- Joined
- Feb 26, 2004
- Location
- peekskill, NY
For the original poster - heed the questions about 'what are you measuring.'
At that level there are many factors that create shunt resistances and those sneak shunts
often are time-varying. For example if I put a standard resistor on my general radio
ohmmeter, I can see the value change by over ten percent by simply breathing on the
part.
A fingerprint or, worse, a puff of rosin solder flux vapor will reduce a samples resistance
well below 10e14 ohm. And again the sneak shunt will not be constant in time. So there
is a considerable amount of hygene involved in high resistance test and measurement.
Materials such as teflon or polystyrene are common. Avoid things like fiberglass (G10, FR4)
as they can be trouble based on their somewhat porous nature.
At that level there are many factors that create shunt resistances and those sneak shunts
often are time-varying. For example if I put a standard resistor on my general radio
ohmmeter, I can see the value change by over ten percent by simply breathing on the
part.
A fingerprint or, worse, a puff of rosin solder flux vapor will reduce a samples resistance
well below 10e14 ohm. And again the sneak shunt will not be constant in time. So there
is a considerable amount of hygene involved in high resistance test and measurement.
Materials such as teflon or polystyrene are common. Avoid things like fiberglass (G10, FR4)
as they can be trouble based on their somewhat porous nature.
A manual for a GR megohmeter (not the model I have, but similiar enough) can be found at:
http://www.ietlabs.com/pdf/Manuals/1644_im.pdf
If you take the time to read it you will see there is a calibration procedure to follow. But, in the end, when measuring a value of, say, 273 GOhm, it's a very nice check on procedure to be able to stick a known 100.0 GOhm resistor into the circuit and have it read 100 GOhm. Measure twice, cut once. Or something like that.
Yes, measuring these high values is difficult. But, one can infer from just the fact that GR sold so many of those bridges says that there is a need to measure up to ~1 TOhm, and that it's possible to do.
I'd given up on finding higher value resistors after extensively searching for them several years ago. However, this thread prompted me to check eBay. As a result, on its way to me from Bulgaria for a total of $9 including shipping is a Russian surplus 1000 GOhm, +/-10% resistor.
Interestingly, there's a decade box (1, 10, 100, and 1000 GOhm) box for sale as well that would be ideal, other than the $1880 price. The fact I'm not the only one interested in such high values of resistance, and that it isn't as nearly-impossible to measure as some of the posters think, is evidenced by the fact the guy had 7 of these available, and already has sold 4.
The smallest digit on my milliohmmeter on its most sensitive scale is 10 micro-ohm (10^-5), and the largest division on my megohm bridge is 1000 tera-ohm (10^+15). No other physical phenomenon comes to mind where only two fairly small instruments allow measurements to be made without any gaps covering 20 orders of magnitude.
The resistor was waiting for me when I got home this evening, so I immediately tried it out. It is sealed in a glass capsule and was packaged in a plastic bag, but if whoever put it there touched the glass with his bare fingers, the residue easily could drop the resistance. Also, I didn't take the time to set up the bridge for a proper super-high resistance measurement, but instead just used alligator clips from the two main terminals (without touching the glass).
So, what did I find? The resistance was a promising 650 GOhm. I'm about to leave on a trip, so it will be a week before I can clean it and connect the leads to make a proper measurement. But, so far I'm quite happy with what I got for my $9.
The real Leigh
Diamond
- Joined
- Nov 23, 2005
- Location
- Maryland
Very good. Those glass-tube resistors require a really good cleaning to achieve nominal resistance.
Victoreen used to make/use lots of those in their geiger counters and similar products.
- Leigh
Victoreen used to make/use lots of those in their geiger counters and similar products.
- Leigh