fluke 187 / 189 dmm nonlinearity issue.

[Martin Miranda]

i used to have a zener voltage reference (4.9963v) that is measured from a 8 1/2 dmm. this is the reference i use to sanity check my multimeters. but recently i got a new voltage reference. it's a 10.000v... then both my fluke 187 and 189 gave a false reading. i am not so sure if this is an issue with other 187's but i made a plot. (plot was made using a power supply and a 6 1/2 digit dmm)

i googled if this is a issue with other fluke 18# dmms but there is no result.
if there is anyone that can verify this. let me know.

[Martin Miranda]

i tried recalibrating but the calibration span is from

0v
5v
50v
500v
500v

but the drift is from 5v to 30ish v.

[mendip_discovery]

Is the 10 V exactly 10 V?
Does it have a temperature drift?
What is its Specification?
Have both meters been kept in the same temperature environment as the references for at least a few hours?

Have you taken into account the specs of the meters? They don't always have to agree.
Also check reversed as well. + / - then - / +

[Martin Miranda]

take a look first in the graph attached. the offset non-linearity is just too big and inconsistent.
voltage is 10.0000v
yes. room temp is 23c (aircon turned on) temperature is from a old dial temperature guage.

[bdunham7]

Could you tell us how you are setting this all up?  You say you have a 4.9963V reference and a new 10.000V reference but you show test voltages from 5 to 50 volts.  How are those produced???

You are showing both meter off about half a volt at 30 volts.  I think that is extremely unlikely.

[Martin Miranda]

the power supply used is a kenwood PA 36-3A
checked with a agilent 34401A multimeter.

You are showing both meter off about half a volt at 30 volts.  I think that is extremely unlikely.
you are right. that is why i wanted to know if there is a way of calibrating this or this is a recurring issue.

[bdunham7]

the power supply used is a kenwood PA 36-3A
checked with a agilent 34401A multimeter.

Clarity is essential here.  Are you using the Kenwood power supply as a source with the 34401A connected to it at the same time in parallel as you are testing the 187 and 189 and comparing the two?  Or are you doing something else?  Did you or anyone else actually adjust the calibration constants in the 187 and 189?  You didn't mention the power supply before, only the voltage references.  Are those involved somehow?

[mendip_discovery]

Connect all 3 meters to the power supply at the same time.

Adjust the power supply until the 34401 shows the desired voltage. Then record measurements on all 3 meters.

[Martin Miranda]

here is the drawing of how i connected the fluke dmm.

[Martin Miranda]

Adjust the power supply until the 34401 shows the desired voltage. Then record measurements on all 3 meters.
that is what is in the spreadsheet.


Did you or anyone else actually adjust the calibration constants in the 187 and 189?
i manually calibrated the meters.

You didn't mention the power supply before, only the voltage references.
i used a diy voltage reference for sanity check. but when i connected it to the 10v reference. it gave me a wrong reading. causing me to investigate more. using a power supply and a 6 1/2 dmm.

[mendip_discovery]

Unless I got something wrong your Fluke Meters are well within each other until the 35 V range which looks to be a typo.



So now this begs the question is the 34401 or the pair of flukes wrong. If someone has adjusted the pair of them using the wrong reference then this could cause this.

[Martin Miranda]

Unless I got something wrong your Fluke Meters are well within each other until the 35 V range which looks to be a typo.
you could be right. i'll grab another fluke i have a 77 iv also. let me check for a moment.

[Martin Miranda]

this is from a fluke 77 iv. (also manually calibrated)

[bdunham7]

Did you or anyone else actually adjust the calibration constants in the 187 and 189?
i manually calibrated the meters.

Unfortunately something appears to have gone wrong in that process.  I'm not sure what would account for that particular error and I didn't know the calibration constants could be that far off.  To answer your original question, I'm not aware of any common issue like this with these meters.

What procedure did you use for calibration--did you follow each and every step in the manual?  It almost looks like something happened in the 300mV or 3000mV step. 

[Martin Miranda]

Did you or anyone else actually adjust the calibration constants in the 187 and 189?
i manually calibrated the meters.

Unfortunately something appears to have gone wrong in that process.  I'm not sure what would account for that particular error and I didn't know the calibration constants could be that far off.  To answer your original question, I'm not aware of any common issue like this with these meters.

What procedure did you use for calibration--did you follow each and every step in the manual?  It almost looks like something happened in the 300mV or 3000mV step.

i borrowed another fluke 187 and it has the same issue. the unit is old/used but factory calibrated. yet has the same issue. we calibrated it again with wavetek 9100 calibrator. and the issue is still the same. we double checked our calibration mV and Vdc. but same result.

[tstp]

The measurement of 5.0000V is eventually using the 5V DC range of Fluke 187/189, and the 10.000V to 50.000V is using the 50V range. So I suggest to test 1.0000V to 3.5000V to see whether the result is similar to the issue of 10.000V to 35.000V or not.

[Kleinstein]

A point to check may be the power supply. Some superimposed AC could through things off. This can be in the normal output voltage, but also a common mode signal.  With a large superimposed AC signal both meters may be off, possibly even way off.

For such a linearity test / comparison of 2 meters a chain of batteries (e.g. AA chells) is the better choice. The points don't have to be at round numbers.

[miro123]

I've just crosschecked my old an trusty Fluke 197 against Brymen 969s and fluke 8842A.
In worse case I have seen 0.02% disagreement. We have bunch of 197s at work and no one have found such issue.
As Kleinsten suggest , You are most likely struggling with grounding issues.
Please connect one 9V battery and measure the voltage with two multimeters

PS: I have seen many miscalibrated fluke's. The worse are Fluke 45 from e-bay - broken cal sticker means 90% of the cases miscalibrated unit

[Martin Miranda]

hello, if can you please do a sweep between 5v to 50v.. i guess this is happening with old fluke that we have. some of them meters we calibrated does not exhibit this error.

[joeqsmith]

Linear supply connected in series as shown by OP.   Cables not twisted.   Approx 1.5meter total length.   No warmup.   189 was purchased used.  I've never attempted to adjust it.    HP34401A was purchased brand new and has never been adjusted.  I did have it checked for calibration 15 or so years ago.   

HP34401   Fluke 189   deviation   %error
6.7758   6.776   0.0002   0.002951681
13.507   13.509   0.002   0.014807137
20.083   20.086   0.003   0.014938007
39.872   39.878   0.006   0.015048154
51.413   51.421   0.008   0.015560267

It would be easy enough to automate and sweep up to 500V but I don't recall ever noticing a problem. 

[bdunham7]

i borrowed another fluke 187 and it has the same issue. the unit is old/used but factory calibrated. yet has the same issue. we calibrated it again with wavetek 9100 calibrator. and the issue is still the same. we double checked our calibration mV and Vdc. but same result.

You have something weird going on to cause this issue and it has to be with your test setup.  Since your Fluke 77 matches your 34401A almost perfectly and I'm assuming you have not 'calibrated' either of those, I'd say the problem is not your reference meter.  The AC interference issue Kleinstein referred to is one possibility, although your deviations seem extreme.  The F187 and F189 do not have some widespread issue that I have seen.  On the last F187, did you do the complete performance check with the Wavetek 9100 before you adjusted the calibration constants?  Given the issues you are having, I would want solid confirmation of any error from a completely separate test setup before attempting an calibration adjustments.

[bdunham7]

Just as another example I set up my 189 (AFAIK still with factory cal settings)  with a PD5020 power supply and a 8846A (known to have factory cal settings) and checked each value from 0.1 to 4.9V at 100mV intervals and then 5-50V at 1V intervals.  Every point was 4 counts or less deviation from the 8846A.

[RYcal]

Do you have all 3 meters connected at the same time?......this could cause loading issues
Are you using a traceable device to "calibrate/adjust" these meters?
There is so many variables in this method.

You need to be able to trust your reference equipment before even beginning to trust anything you are testing. This includes cables and setup etc.

I preform sanity checks for most people free of charge just to highlight the need for good traceable reference equipment. 99.9% of the time there gear is good without issue, the issues are from them trying to "hack" a solution at home.

[Martin Miranda]

with everything i posted above. this is about the fluke 187/189 meters that fails test. it is adjusted OK. so far we have 6 out of 40 meters tested with this problem. it passes the adjustment/performance check.

1. the technician performs the performance test based on the calibration manual (see table 4, on page 18 of fluke 187/189 calibration manual)
2. we are using a in house calibrated meters, the technicians trusted the meters that it is calibrated by me/us.
3. when i tested the meter using a 10V reference it failed. (see table 4, on page 18 of fluke 187/189 calibration manual) as we do not do sweep.
4. yes, we already tested a factory calibrated meter, with this problem.

[Martin Miranda]

we just discovered this problem with fluke 187/198. and don't blindly follow the (see table 4, on page 18 of fluke 187/189 calibration manual) performance test. it's better to also check for linearity. as we became complaisant doing the same work again and again.

[bdunham7]

Do you have or can you get a reliable precision power supply like an SMU or something and then carefully check it with a scope for output ripple, noise and stability?  And then do the same thing as you were doing with the Kenwood model?  You could also use the calibrator, but given the issues here you might want to bring in something else.

If you connect one of the malfunctioning meters to a well-vetted PSU, or your Wavetek calibrator and then select the same set of voltages that I tried--0.1 to 4.9V by 100mV steps, then 5V to 50V in 1V steps--then you would have a more complete idea of what might be going on.  I'd also do the same with the 34401A (presuming it has not been recalibrated in your lab) just to verify whatever you are using.  You also might want to switch the 34401A to AC at each step just to see if there is any ripple.

When you say you have found 6 out of 40 meters to have this issue, do you mean you've tested all 40 and 6 of them behave like this?  If so, do those 6 have any common history or use?  I would definitely stop trying to recalibrate anything until you figure this out.  I'm pretty sure there is no widespread issue with the 187/189 series of meters, but yours could have been damaged by some unusual use or something else in their history.  Gross errors like that are pretty weird.

I didn't ask before--what equipment do you have available? 

[joeqsmith]

Looking at my other three, they all appear close to the 34401A.  You could check the SN or see if there was a common hardware revision for the six you found.   

[ramon]

Do you have or can you get a reliable precision power supply like an SMU or something and then carefully check it with a scope for output ripple, noise and stability?

I strongly suggest to test exactly this. Two years ago I was testing LM399, LT1027, LTZ1000, ... with my R6581T. I found the problem to be my Kenwood PWR36 power supply.   

[Martin Miranda]

please read reply #8. thank you.

[ramon]

Sorry, I haven't read all post in detail. Could it be that those fluke devices have some linearity issue that nobody has ever tested/detected before. But I still would double check the setup.

9V PP3 batteries can be easily connected in series. Just only need four or five, and some resistors.

Also, tstp said to compare the 5 and 50V ranges. Could it be that the 50V range has some component/trace with issues?

[Martin Miranda]

yes. these are 3 year old battered/dropped/hit/left in the sun field multimeters. our mistake in testing that is we follow the performance test based on the calibration manual of fluke we did not bothered to check for linearity.

[Martin Miranda]

here is my test regarding the non-linearity, i'm still sick and get can't back to office as of the moment.
https://youtu.be/RJSmQ4StKm8

[Martin Miranda]

don't just follow the adjustment procedure, always sweep for linearity. 

[bdunham7]

don't just follow the adjustment procedure, always sweep for linearity. 

I would add "don't calibrate a broken multimeter".  And in most cases if your (modern) Fluke handheld is off enough to need calibration adjustments, it is more likely that it is broken than it is that it has drifted off.  This doesn't necessarily apply to your batch of problems, I have no idea what is going on there.

[Kleinstein]

The fault looks like a wrong calibration for at least one of the ranges. This could be due to a hardware problem (e.g. overload, corrosion) or a faulty calibration.

A normal calibration should check at some 5 V (near the end of the lower range) and some 20 or 50 V for the higher range. This should detect the problem already. As the calibration should check all the ranges.

A calibration should ideally check the function at a few more points than the minimum number to get the scale factors. This would be especially point to  check for more common defects. Normally there are cal procedures to tell the points to test as a minimum.
I don't expect such meters to developement a linearity problem - a more hidden and not so often checked problem is more like additional input current needed. With manual reading the result extra tests const time and thus money - though maybe less of a problem with lower wages.

[Martin Miranda]

i don't think you understand the issue. 

- the technicians are using the multimeter. they don't know that the meter is bad.
- we know the meters are NOT broken and in good working condition.
- we followed the calibration manual, (page 18, Table 4. Performance Test Steps)
- we discovered the non-linearity issue. when we tested the meter with a precision voltage.

[Martin Miranda]

The fault looks like a wrong calibration for at least one of the ranges.
we double checked with some meter if we did something wrong. this is currently with a random batches of meter.

This could be due to a hardware problem (e.g. overload, corrosion) or a faulty calibration.
THIS is a hardware problem, but we sent out out of cal multimeters, no checking for linearity...

A normal calibration should check at some 5 V (near the end of the lower range) and some 20 or 50 V for the higher range. This should detect the problem already. As the calibration should check all the ranges.
our mistake is just following the calibration manual. not some 20 or some odd value. this is where we knew the issue.


A calibration should ideally check the function at a few more points than the minimum number to get the scale factors. This would be especially point to  check for more common defects. Normally there are cal procedures to tell the points to test as a minimum.
I don't expect such meters to developement a linearity problem - a more hidden and not so often checked problem is more like additional input current needed. With manual reading the result extra tests const time and thus money - though maybe less of a problem with lower wages.
they will send in the meters, knowing that it only needed adjustment it is in good working order. but is it also metrology should inform the client that when a multimeter should be rma'ed, warranty claim or just ber?

[Martin Miranda]

- a technician is only issued with 1 multimeter. (unless they check the meter with another technician) which why will not.
- in calibration, we tend to check off with the procedure. grab coffee and sign off.
i don't think this is a case of carelessness. we changed procedure since we discovered the issue.
it's only a matter of adding two points in between test.

[Martin Miranda]

- and lastly... THIS IS MY MULTIMETER I USE ALMOST EVERYDAY!!   

[bdunham7]

- we know the meters are NOT broken and in good working condition.
- we discovered the non-linearity issue. when we tested the meter with a precision voltage.

If they are not linear then they are broken, unless there is a severe calibration issue induced by some calibration procedural error.  If they were working properly in the distant past (when new) and they haven't been calibrated since then, that level of non-linearity would indicate some sort of failure, not the need for calibration.  Thus my warning "don't calibrate a broken meter". 

If you get a chance, I'd be curious to see what you get when you measure the output of your Wavetek 9100 with the Aneng meter at 50mV, 300mV, 500mV and 3.000V.  That's assuming this the exact calibrator that you used on the 189s and not the broken one you discussed in another thread.

[Martin Miranda]

If they are not linear then they are broken.
(broken is defined by unusable, blank display, crushed, erratic values... ) a user/technician won't know if there is a skew/linearity problem.


i'm only telling that don't JUST follow the calibration manual Table 4. Performance Test Steps and sign off.


broken one you discussed in another thread
  this is the broken calibrator. but trust me. it works now. 

[Kleinstein]

The test at the normal testpoints from the calibration should catch such an error as shown in the video. There should be not extra test points needed for this.
As the deviation starts at just the change of the range it looks like a wrong adjustment of the 50 V range. So likely 2 points seem to have gone wrong in the cal procedure:
1) the adjustment of the 50 V range
2) not catching the error in the testing after adjustment and detecting the first error right away.

It can help to do a few more test than those in the calibration manuals, but these add to the costs. So the additional steps are more like the simple ones, like check the zero drift apon warm up, or additional point in an automated procedure in the waiting time before the actual test.

Additional points are advised to catch know common issues with the meter, that may otherwise be missed. This may be part of updated cal procedures.

With a repaired calibrator there is also a chance that the calibrator may be at fault ! So the higher voltage range there may be off and causing both the wrong adjustment and not detecting the fault. This is why there should be an additional good meter checking the calibrator.

[alm]

Is the non-linearity also there in the DC mV range? What did these test points from the performance verification (table 4) give?


In the adjustment procedure there are 300 mV and 500 mV points that I imagine may be used for ADC linearity adjustment. The voltage dividers for higher voltage ranges should not need separate linearity adjustment.

[Martin Miranda]

good meter checking the calibrator

i cross checked my wavetek9100 with
- fluke 732b DC standard
- fluke 5790a AC Measurement Standard
- wavetek 1280 Dmm.

[Martin Miranda]

update with the mV.

[bdunham7]

If they are not linear then they are broken.
(broken is defined by unusable, blank display, crushed, erratic values... ) a user/technician won't know if there is a skew/linearity problem.

i'm only telling that don't JUST follow the calibration manual Table 4. Performance Test Steps and sign off.

OK, so your definition of broken is a bit more simple and lurid than mine.  Shall we say 'malfunction', whether or not said malfunction is evident to the casual observer?  I don't think we're totally in disagreement here, except I'm saying you should do your performance and linearity checks before any attempt at calibration.  It is very typical for me to take a meter that is 'wrong', repair the fault and then recheck it to find that no calibration adjustments are needed.  Even on older units with trimpots I would overhaul them before making any adjustments, and when I haven't followed that rule I've regretted it as it has made more work in the end.

i cross checked my wavetek9100 with
- fluke 732b DC standard
- fluke 5790a AC Measurement Standard

Not relevant to the potential miscalibration issue.

- wavetek 1280 Dmm.

At the voltages I suggested?

[alm]

update with the mV.

Okay, so it's not a linearity problem with the ADC. The performance verification does a good job testing that with 0, 20000, 30000 and 50000 count points across different ranges. It's that there's a problem with the divider for the 50V range. Does measuring the input impedance at 50V compared to 5V or 500V yield anything strange? How is the linearity in the 500V range?

I'm wondering if there's a protection component that's only in the circuit on the higher DCV ranges?

[Martin Miranda]

update with the mV.

Okay, so it's not a linearity problem with the ADC. The performance verification does a good job testing that with 0, 20000, 30000 and 50000 count points across different ranges. It's that there's a problem with the divider for the 50V range. Does measuring the input impedance at 50V compared to 5V or 500V yield anything strange? How is the linearity in the 500V range?

I'm wondering if there's a protection component that's only in the circuit on the higher DCV ranges?

It's that there's a problem with the divider for the 50V range.
thank you that is what i am only pointing out. that's all.
i can't move this to repair thread because it's just a metrology issue.

[Martin Miranda]

If they are not linear then they are broken.
(broken is defined by unusable, blank display, crushed, erratic values... ) a user/technician won't know if there is a skew/linearity problem.

i'm only telling that don't JUST follow the calibration manual Table 4. Performance Test Steps and sign off.

OK, so your definition of broken is a bit more simple and lurid than mine.  Shall we say 'malfunction', whether or not said malfunction is evident to the casual observer?  I don't think we're totally in disagreement here, except I'm saying you should do your performance and linearity checks before any attempt at calibration.  It is very typical for me to take a meter that is 'wrong', repair the fault and then recheck it to find that no calibration adjustments are needed.  Even on older units with trimpots I would overhaul them before making any adjustments, and when I haven't followed that rule I've regretted it as it has made more work in the end.

i cross checked my wavetek9100 with
- fluke 732b DC standard
- fluke 5790a AC Measurement Standard

Not relevant to the potential miscalibration issue.

- wavetek 1280 Dmm.

At the voltages I suggested?

I'm saying you should do your performance and linearity checks before any attempt at calibration.

the time i posted the error in my first post. i am confused if i am making a mistake in alignment/cal
"i do not check for calibration, i only follow the fluke's calibration procedure. i am not doing any linearity check."

i'm just saying, posting to this tread that FLUKE 187's has potentially has this issue and can't be aligned.
that's all. 

[bdunham7]

thank you that is what i am only pointing out. that's all.
i can't move this to repair thread because it's just a metrology issue.

I doesn't matter much what forum you put the thread in, we'll find you wherever you go! 

Perhaps you could do the same linearity sweep in the 500mV, 5V and 500V ranges, preferably with another meter in parallel--even a cheap one like the Aneng.

Unfortunately I don't think anyone has a schematic for the 189 so I'm not sure where to start troubleshooting. 

[jesuscf]

I had issues with my Fluke 187 measuring anything correctly.  It turned out to be broken pins in the terminal block, the ones soldered to the PCB, a very common issue with the Fluke 187/189.  There is a thread on this forum on how to fix that problem.

[alm]

I did a quick test with a Fluke 189 connected to a HP 6114A precision PSU (accuracy in the same ballpark as the Fluke 189) measuring 9, 19, 29 and 39 volt, and all where within a few counts from nominal. So like Joe, I'm not observing the same. So clearly there's something special with your particular sample of DMMs. Either a problem with that particular batch (were they bought at the same time?), or something specific to your environment, like them being exposed to overload beyond their ratings.

Could it be the the supercap leaking and creating a leakage part in the 50V range that affects the linearity of that particular range? I have no idea how close those components are on the PCB.

[Martin Miranda]

I doesn't matter much what forum you put the thread in, we'll find you wherever you go! 
i mean, it is better to talk here with metrology members.

Perhaps you could do the same linearity sweep in the 500mV, 5V and 500V ranges, preferably with another meter in parallel--even a cheap one like the Aneng.

yes. i'll post that soon.

Unfortunately I don't think anyone has a schematic for the 189 so I'm not sure where to start troubleshooting.
yes. same here. but i'll make another thread. i'll try swapping the resistor with a good working unit.

[Martin Miranda]

So clearly there's something special with your particular sample of DMMs.

yes, correct. i think the damage is caused by shock. (dropping on a concrete floor)

https://www.eevblog.com/forum/metrology/fluke-187-189-dmm-nonlinearity-issue/msg3903305/#msg3903305

[Kleinstein]

I am still not convinced this is actually damaged hardware in the meter. It could still be just wrong cal constants for the 50 V range.

There is little use in checking the lower ranges. The interesting point is the 50 V range, as this was the part reading wrong.
This is check the Fluke meter in the 50 V range are some 10 V and 40 or 50 V.
Similar set the soure to 4 or 5 V and read the same voltage in the 5 and 50 V range.

[bdunham7]

I am still not convinced this is actually damaged hardware in the meter. It could still be just wrong cal constants for the 50 V range.

I don't see how.   Only the mVDC ranges have multiple calibration points that could be used for anything like a linearity compensation, all the other ranges are simply mx + b setups AFAIK.  The only possibilities I can see are some error at the 300mV and/or 3000mV calibration points or some really weird but consistent damage to the meter.

[Kleinstein]

The measurement with the nonlinearity showed good readings up to 5 V and than bad (considerably too high) values from 6 V on. The 6 V - 10 V are obviously in the 50 V range and no longer in the 5 V range. The readings up to 5 V would be in the 5 V range if in auto mode.

I don't know the cal procedure, but I strongly suspect that there is a testpoint at some 50V to adjust the gain (divider) for that range.



Chances are the calibrator will change ranges at some 10 or 20 V. So the 50 V calibration would be with a different range there. If this one is bad, it can be the source of a bad adjustment and than still reading "good" (as in wrong / wrong = good) in the test at 50 V, but bad in the later test at 6-10 V.

[bdunham7]

The measurement with the nonlinearity showed good readings up to 5 V and than bad (considerably too high) values from 6 V on. The 6 V - 10 V are obviously in the 50 V range and no longer in the 5 V range. The readings up to 5 V would be in the 5 V range if in auto mode.

I don't know the cal procedure, but I strongly suspect that there is a testpoint at some 50V to adjust the gain (divider) for that range.

OK, I looked at the original post and there are added trailing zeros to the 10V-50V readings that make it look like it was all the same range, but it wasn't.  So while I'd still like to see that the output of that calibrator is at the 300mVDC setting, I still have no clear idea of what might be going on here.

The cal procedure only specifies 5/50/500VDC for all the DC ranges above the mV ranges, and those use 50mV, 300mV, 500mV and 3V.

[alm]

Clearly there's a difference in linearity between the 50V (and up?) range and the lower DCV ranges. It could either be that 50V is off, or the lower ranges are off and have been adjusted wrong in the past. Maybe a varistor is damaged and is conducting at much lover voltages? Or maybe the switch is dodgy and making wrong connections?

There is a service manual with schematics of the 87V that may have a somewhat similar input circuit.

[Kleinstein]

The very first post shows a large error in the 50 V range and a rather small one in the 500 V range. For the 5 V range the meter is in line with the other cheap meter and the calibrator. So chances are high it is the 50 V range that is off. in the Fluke meter and that is it. So far I have not seen mich linearity problem inside one range.

[bdunham7]

The very first post shows a large error in the 50 V range and a rather small one in the 500 V range. For the 5 V range the meter is in line with the other cheap meter and the calibrator. So chances are high it is the 50 V range that is off. in the Fluke meter and that is it. So far I have not seen mich linearity problem inside one range.

That I'm not seeing.  Leaving the first point which is the 5V range, all the rest are in the 50V range.  There is a sort-of proportional error up to 30 or 35V, then above that the meter snaps back to being correct.  Definitely not linear and the 500V range is not involved as far as I can see.