I'm restoring my old HP 410B VTVM, and I'm interested in seeing how much the resistors have drifted since it was built, particularly the precision resistors in the input voltage divider. I don't have volts-nuts caliber equipment (well, there is a busted HP 3456A on the shelf waiting to be repaired someday), just a Fluke 8050A and a Fluke 27/FM.

I didn't expect to have much trouble making consistent measurements as I don't think the 8050A has the resolution to see temperature coefficient changes or thermocouple effects. But I'm seeing some odd results on the higher resistance values. First, I seem to see some contact resistance effects: I don't get consistent measurements just using the probes as the count varies a little with contact pressure and probe placement. The contact resistance would have to vary by thousands of ohms for it to affect the meter; I can't believe that could be the explanation. However, I'm able to get consistent measurements by slipping alligator clips on to the probe tips and clipping on to the range switch terminals. Maybe the old solder is so oxidized that the contact resistance can really vary that much?

Second, the Fluke 27/FM measurements track those of the 8050A better than the spec'd limits, but I see some odd behavior in the last digit of the 8050A. The last digit of the resistance value varies with the direction of the current through the resistor, and in one direction, it bobbles up and down about three counts. In the other direction, the reading is stable. The bobble doesn't seem to be sensitive to placements of the test leads.

For example, the 6.837M Ω 1% resistor (R6) measures 7.037M Ω one way, and between about 6.994M and 6.996M Ω when I reverse the 8050A test leads. That's a difference of nearly 0.6%

The 2.163M Ω 1% resistor (R5) measures 2.220M Ω one way, and between about 2.215M and 2.217M Ω when I reverse the leads, for a difference of about 0.2%.

The 683.7K Ω 1% resistor (R4) measures 697.9K Ω one way, and between about 697.5K and 697.7K Ω when I reserve the leads, for a difference of about 0.05%.

I did try switching off some potential nearby RFI sources - fluorescent lights, switching power supply, laptop computer - and saw no difference in behavior, although I didn't do an exhaustive search for RFI.

Finally, I did some quicky measurements with the Fluke 27/FM about two months ago, and the current measurements seem to be a bit off (I don't have the old recorded measurements handy as I write this, but I think they are outside the accuracy limits spec'd for the 27/FM). This is a non-climate-controlled New England basement, so the temperature is probably up about 5 degrees C and the humidity has shot up recently. But again, I wouldn't think my instruments are good enough to notice these environmental effects on the components themselves.

Any ideas as to what's going on? How can I improve my measurement procedure to get repeatable results? Do I really need better climate control even at the 3 1/2 or 4 1/2 digit level of precision? What's with the polarity sensitivity of the 8050A resistance measurements? Suggestions and advice would be gratefully accepted.

Best regards,
-Steve

--
Steve Byan stevebyan@me.com
Littleton, MA 01460

I Agree with Mitch on the "bobbling" being AC pick up. The other variations you are seeing are probably thermal EMFs. The EMF will change with different locations if the lead plating is different or you are contacting solder or the switch. I's difficult to make these high resistance measurements accurately. Best way I found (before getting  proper high ohm meters) was a low value low resistor in series with the resistor being tested with a stable high voltage across the series pair (if nothing else a string of 9V "radio" batteries in series, the snap together, and monitor the voltage with your 27. Measure the voltage across the low R with your 8050 and calculate the current. Don't forget to allow for the 10M of the 8050 across the resistor. Then apply Ohms law. Do a zero offset with the high voltage source off to remove the thermal EMFs.
 
HTH,
Robert G8RPI.

From: Steve Byan stevebyan@verizon.net
To: volt-nuts@febo.com
Sent: Monday, 17 June 2013, 18:50
Subject: [volt-nuts] Semi-precision high resistance measurement

I'm restoring my old HP 410B VTVM, and I'm interested in seeing how much the resistors have drifted since it was built, particularly the precision resistors in the input voltage divider. I don't have volts-nuts caliber equipment (well, there is a busted HP 3456A on the shelf waiting to be repaired someday), just a Fluke 8050A and a Fluke 27/FM.

I didn't expect to have much trouble making consistent measurements as I don't think the 8050A has the resolution to see temperature coefficient changes or thermocouple effects. But I'm seeing some odd results on the higher resistance values. First, I seem to see some contact resistance effects: I don't get consistent measurements just using the probes as the count varies a little with contact pressure and probe placement. The contact resistance would have to vary by thousands of ohms for it to affect the meter; I can't believe that could be the explanation. However, I'm able to get consistent measurements by slipping alligator clips on to the probe tips and clipping on to the range switch terminals. Maybe the old solder is so oxidized that the contact resistance can really vary that much?

Second, the Fluke 27/FM measurements track those of the 8050A better than the spec'd limits, but I see some odd behavior in the last digit of the 8050A. The last digit of the resistance value varies with the direction of the current through the resistor, and in one direction, it bobbles up and down about three counts. In the other direction, the reading is stable. The bobble doesn't seem to be sensitive to placements of the test leads.

For example, the 6.837M Ω 1% resistor (R6) measures 7.037M Ω one way, and between about 6.994M and 6.996M Ω when I reverse the 8050A test leads. That's a difference of nearly 0.6%

The 2.163M Ω 1% resistor (R5) measures 2.220M Ω one way, and between about 2.215M and 2.217M Ω when I reverse the leads, for a difference of about 0.2%.

The 683.7K Ω 1% resistor (R4) measures 697.9K Ω one way, and between about 697.5K and 697.7K Ω when I reserve the leads, for a difference of about 0.05%.

I did try switching off some potential nearby RFI sources - fluorescent lights, switching power supply, laptop computer - and saw no difference in behavior, although I didn't do an exhaustive search for RFI.

Finally, I did some quicky measurements with the Fluke 27/FM about two months ago, and the current measurements seem to be a bit off (I don't have the old recorded measurements handy as I write this, but I think they are outside the accuracy limits spec'd for the 27/FM). This is a non-climate-controlled New England basement, so the temperature is probably up about 5 degrees C and the humidity has shot up recently. But again, I wouldn't think my instruments are good enough to notice these environmental effects on the components themselves.

Any ideas as to what's going on? How can I improve my measurement procedure to get repeatable results? Do I really need better climate control even at the 3 1/2 or 4 1/2 digit level of precision? What's with the polarity sensitivity of the 8050A resistance measurements? Suggestions and advice would be gratefully accepted.

Best regards,
-Steve

--
Steve Byan stevebyan@me.com
Littleton, MA 01460

volt-nuts mailing list -- volt-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts
and follow the instructions there.

Thanks for all the suggestions.

The Fluke 8050A schematic is hard to follow due to the complicated range and function switching, but if I did so correctly, on the 20M ohm and 2M ohm ranges it applies 1.2 volts through a precision 10.00M ohms reference resistance, then through about 2K ohms, then to the resistor under test. It measures the voltage across the resistance reference and then the resistor under test and reports their ratio. The 8050A has a conductance function as well as resistance, but the actual measurement circuit is identical to the resistance function; the dual-slope conversion is simply inverted.

For the 6.837M ohm resistor I measured, the voltage on the resistor under test would be about 0.5 volts, so I can't see thermoelectric effects being an explanation.

For the 683.7K ohm resistor, the voltage would be about 77 mV, so I could see the need to start to worry about thermoelectric effects, but the change in resistance for the smaller resistor is lower, so I think it must be some other effect.

I have a 2.00M ohm 1% wire-wound resistance standard in a nice oak box that I obtained from the recent MIT lab storeroom clean-out. I've yet to clean up the terminals, but measurements of it seem to exhibit similar polarity sensitivity, so I now think the cause must be in my measurement technique or in the meter itself. The 8050A reads about 10 µV DC across the resistance, so that seems like a thermoelectric effect, but one too small to affect the resistance measurement. The kicker is that the 8050A reads about 0.4 volts AC across the resistance, and it is only spec'd at ">60 dB" normal-mode rejection ratio for 60 Hz.

I tried twisting the test probe leads together, but I still get about 0.4 volts AC. I plan to try using coax next, with short leads to the test clips, and see if I can get the AC level down. I should also try moving the meter somewhere further away from the service entrance to my house; my workbench is only about 5 feet away from it.

Best regards,
-Steve

--
Steve Byan stevebyan@me.com
Littleton, MA 01460

Are you running the 8050 from mains or does it have the battery pack fitted?
 
Robert G8RPI.

From: Steve Byan stevebyan@verizon.net
To: Discussion of precise voltage measurement volt-nuts@febo.com
Sent: Friday, 21 June 2013, 0:48
Subject: Re: [volt-nuts] Semi-precision high resistance measurement

Thanks for all the suggestions.

The Fluke 8050A schematic is hard to follow due to the complicated range and function switching, but if I did so correctly, on the 20M ohm and 2M ohm ranges it applies 1.2 volts through a precision 10.00M ohms reference resistance, then through about 2K ohms, then to the resistor under test. It measures the voltage across the resistance reference and then the resistor under test and reports their ratio. The 8050A has a conductance function as well as resistance, but the actual measurement circuit is identical to the resistance function; the dual-slope conversion is simply inverted.

For the 6.837M ohm resistor I measured, the voltage on the resistor under test would be about 0.5 volts, so I can't see thermoelectric effects being an explanation.

For the 683.7K ohm resistor, the voltage would be about 77 mV, so I could see the need to start to worry about thermoelectric effects, but the change in resistance for the smaller resistor is lower, so I think it must be some other effect.

I have a 2.00M ohm 1% wire-wound resistance standard in a nice oak box that I obtained from the recent MIT lab storeroom clean-out. I've yet to clean up the terminals, but measurements of it seem to exhibit similar polarity sensitivity, so I now think the cause must be in my measurement technique or in the meter itself. The 8050A reads about 10 µV DC across the resistance, so that seems like a thermoelectric effect, but one too small to affect the resistance measurement. The kicker is that the 8050A reads about 0.4 volts AC across the resistance, and it is only spec'd at ">60 dB" normal-mode rejection ratio for 60 Hz.

I tried twisting the test probe leads together, but I still get about 0.4 volts AC. I plan to try using coax next, with short leads to the test clips, and see if I can get the AC level down. I should also try moving the meter somewhere further away from the service entrance to my house; my workbench is only about 5 feet away from it.

Best regards,
-Steve

--
Steve Byan stevebyan@me.com
Littleton, MA 01460

volt-nuts mailing list -- volt-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts
and follow the instructions there.