Very strange behavior of my Keithley 182 nanovoltmeter

[gamalot]

I got a Keithley 182 nanovoltmeter with a display failure a few days ago, which is easy to fix by making a simple 60V boost converter. Then I did some simple tests on this instrument. Unexpectedly, I got a very interesting (strange) result.

In short, in the 30V range, near 3.25V 5.2V, the voltage measurement has a sudden jump about 100uV. I have repeated this many times in the past, and the results are the same, no matter how I change the filter, integration time, or other settings.

 

[e61_phil]

Hi,

a couple of month ago I did some linearity tests with my K182. I attached the results. I can do further measurements if that helps (not before december).

First row is calibrator output voltage, second row is K182 reading and third row is temperature.

best regards
Philipp

Edit: The K182 isn't as bad as is looks here. The comparison is not fair in 1/3 of the range. Therefore I did another test in the 3V range with a divider. That showed +/- 0.4ppm. But that is offtopic, I think.

[gamalot]

Hi,

a couple of month ago I did some linearity tests with my K182. I attached the results. I can do further measurements if that helps (not before december).

First row is calibrator output voltage, second row is K182 reading and third row is temperature.

best regards
Philipp

Edit: The K182 isn't as bad as is looks here. The comparison is not fair in 1/3 of the range. Therefore I did another test in the 3V range with a divider. That showed +/- 0.4ppm. But that is offtopic, I think.

Thank you Philipp!

Thank you very much for the reply, it made me find out a mistake in my post: it should be 5.2V not 3.25V. At the same time, your reply also proves that my K182's strange behavior is not a special case.

I will do some further testing to try to figure out what is happening near to 5.2V.

If you don't mind, can I know the firmware version of your K182?

 

[gamalot]

I did a sweep up and then sweep down, I was surprised to find that there is a hysteresis, just feels like auto ranging.

[e61_phil]

I will try this sweep next week

My K182 shows the following on startup:

Keithley 182 SDV
A04 B03 A01

[TiN]

I tested my 182-M last night too. It is modified version with 9V battery powered front-end. FW version A04 B03 A01. Calibrated earlier this year using procedure from manual and 5720A/Keithley 262 setup.
INL data I got leaves me wondering too:

-3/+3V sweep on Keithley 182M 3V range. 100ms integration, Filters off, 6.5 digit mode. Reference meters in 10V range



Correct data below:

[Vgkid]

That is kind of weird.   
Thanks for posting.

[TiN]

del some nonsense..

Doing test with 30V sweep currently. (other meters on 100V range).

[David Hess]

What type of ADC does it use?  That looks like the glitch commonly found in synchronous voltage-to-frequency ADCs.

[Kleinstein]

It's linearly nonlinear  . Perhaps I'll change the battery in front end and test again some time later. 

Doing test with 30V sweep currently. (other meters on 100V range).

The linear part (like in the 3 V range of the 182) of the "INL" curve kind of indicates a problem in calculating the straight line to subtract. So this is not a property of the meter, but a kind of test problem.  So the 3 V and 0.3 V range look good - with possibly some minor problems in the 2.2 V range.

[gamalot]

Is it possible that the input protection circuit causes this strange result?

[Kleinstein]

I don't see a problem with the input protection. There may be an issue with the extra 30 V amplifier, if it's only used for more than 5 V instead of for the whole 30 V range.

[TiN]

Kleinstein right on money. Sure, there was a bug in the data plotter. Shame on me for silly mistake. Array with INL points was calculated by using poly from channel 5 (F8508 data) instead of channel 6 (K182M).

Pulled wrong pics from post above, will replot the proper ones tonight.

[e61_phil]

Sorry, I forgot my promise to take some measurement.

At first here are some old measurements for K182 INL. Measured against my Fluke 5440B. The 5440B was sweeping from 0 to 10V for best linearity. The 0 to 10V was then reduced to 3 to 10 by a SR1010-1k.

@TiN: Could you reproduce the hysteresis?

Now I will start some up and down sweeps to see if my K182 also has this hysteresis.

[gamalot]

Thank you guys! I am doing something really boring!   

[e61_phil]

I could confirm these hysteresis..

[TiN]

I see the jumps on 30V range too, at about +/-5.2 and +3.5 VDC levels. 



On lower ranges like 3V and 300mV there are no problems (at least in my unit). 

[e61_phil]

I also see it only in the 30V range. The measurement above with +/- 0.4ppm was done in the 3V range.

I've done some more runs this night. The "switching" points always appear at the same voltages.