Negative Ohms Readings - Keithley DMM6500 updates November 26, 2018 Page 2

[omegaone]

Hello All,

When testing lead resistance, I ran into an issue first with FLUKE  289 at selection LoΩ.
Lead made 14 AWG, Teflon sleeve 600 V, 200 C.
Resistance discrepancy between from 9 mΩ  positive reading on one meter, and 9mΩ negative reading on another  meter.

First time, I ever witnessed the negative ohms territory on a practical test.

Since then, I have been trying to get some answers as to why. I have sent the log files to FLUKE product specialist, they are still working on it.

If anyone has any suggestions, please give some feedback.

Note:

I have also received an evaluation unit DMM6500. So I have repeated the test with the leads. To my surprise, DMM6500 read negative 5mΩ in 2 wire mode. ( Agreeing with one of my fluke 289 which read negative 9mΩ)

I am also getting Keithley to look into the same.

[alm]

If it's in two wire resistance mode (the Fluke 289 has no other option) it could just be that the short you are using has a lower resistance than the one used for calibration. If it was calibrated with a 30 mOhm short, anything below 30 mOhm will read as negative. External voltages (e.g. Seebeck effect) can also cause a negative offset.

What is the uncertainty Fluke specifies for this range? 1 mOhm resolution is really pushing it without Kelvin sensing or offset voltage compensation in my opinion.

[commongrounder]

There have been other threads on this forum discussing negative readings in 50-ohm mode on the 289.  My 289 also does this, and I have found it to be temperature dependent.  Measuring resistance at these low values, with only two-wire mode, you have to allow for noise, as well as thermal other reactions due to dissimilar metals.  I personally think it does pretty well, considering the applications the meter was designed for.

[threephase]

As per the others, you are asking an awful lot from a 2 wire resistance measurement in terms of practicality.

You can take a look at the calibration manual for the Fluke 289 at;

https://dam-assets.fluke.com/s3fs-public/287_289_cmeng0100.pdf

The basic accuracy on page 7 for the 50 Ohm range is 0.15% +20.

If you look at the performance tests for the resistance range on page 19, it doesn't actually give a value for the 0 Ohms test, only that a stable reading is required. For 0.2Ohms applied, there reading should be between 0.18 and 0.22 Ohms. Interestingly, if you scroll back up to page 18, it give a 0 Ohms test applied to the 500 Ohm range with an expected value between -0.01 and +0.01 Ohms.

The calibration of the 289 recommends the use of a Fluke 5520a calibrator. Looking at the specifications of that for the 0 to 11 Ohms function, there is a 1mOhm +35ppm uncertainty applied.

The DMM6500 looks better as it has a 10 Ohm range, with 0.002% rdg + 0.02% range specified best accuracy. However, reading the small print for 2 wire resistance test, compensation mode has to be applied and even then there is a 100 mOhm uncertainty.

For 4 wire measurement on the DMM6500 you can go down to a 1 Ohm range, but the accuracy is at 0.008% rdg + 0.02% range.

The above only applies to the DMM6500 if you are within 24 hours of the lat calibration and temperature is within 1 degree C.

The problem with the function is the low test current of 10mA, with such low resistances the reading will be susceptible to noise and thermal stability of joints.

You are really in the realms of a source measurement unit or a dedicated micro-ohmmeter that can increase the test current and give a more stable reading. The downside being that the increased current affects the measurement of the test item to some extent by heating it.

Kind regards

[omegaone]

"If it was calibrated with a 30 mOhm short, anything below 30 mOhm will read as negative. External voltages (e.g. Seebeck effect) can also cause a negative offset"

Very good point. At present, I do not know what was used in calibration.

I will look into it further. Thank you.

[omegaone]

Good to know you noticed similar on your meter.
I was comparing two Fluke 289 side by side.  As per fluke engineers, I did warm up the meter then made sure lead to meter temperature was consistent before taking the reading close to room temperature.

[threephase]

omegaone, it looks like I have the same shunt as you, so I have tried it in my Fluke 8846A, the meter was factory calibrated in April 2018.

In 2 wire mode, the reading was 0.00549 Ohms. In 4 wire mode, this dropped down to 0.00389 Ohms. The physical setup was the same, I just switched between the 2 wire and 4 wire mode. Having the short connected directly as per photo 3, the reading was 0.00489 Ohms. I did not get a negative reading at any point, but the ohmic values seem to be similar to yours.

Playing around with the instrument, I did get a negative reading in 2 wire, if I compensated and in 4 wire if I reversed the sense connections.

I then tested a 1mOhm shunt 0.1% accuracy. Again no negative ohms reading was obtained, although the readings were no where near where they should have been. The 2 wire measurement was high, which is what I would have expected, the 4 wire was low, which is not expected, so there may have been some setup error causing this.

A 10mOhm shunt produced a high reading in 2 wire mode, and a much closer reading in 4 wire mode. Which is pretty much what I would have expected, although given the tolerance of the 10mOhm shunt, also 0.1%, the 4 wire reading is too high.

Kind regards.

[threephase]

Pictures of the 10mOhm shunt being tested, as it got too big for the reply.

[threephase]

This is the 1mOhm shunt being tested on a 2450 SMU with different setups being used.

Using 4 wire measurement and a 100mA test current, the reading is quite poor, up the test current to 500mA and the reading is better but still not close enough. At 1A, I get a reading within tolerance, but it is still a little high in comparison to a reading from a micro-ohm meter injecting a 10A test current.

The last picture shows a 2 wire measurement using the SMU with a really bad result. So even an instrument with this level of accuracy and expense can have its bad moments.

Kind regards

[omegaone]

Thank you for all the pictures and valuable info.
I have been bit busy with my regular job and with myself and was unable to go through them all. But I intend to, soon.

In the meantime, I am attaching the log files, graph, thermal image, and pictures I sent to Keithley for your review.


Note: I have deleted the 4 wire resistance readings from this post due to erroneous connection of leads to keep confusion to a minimum!
 ( Thanks to Threephase and Commongrounder)
2 wire connection and negative ohm readings still valid at this point.( 24 November 2018)

[threephase]

Thanks for the extra information.

Not sure why the two wire measurement goes negative, as suggested above, it may be due to the way the meter is calibrated. If I was trying to make that measurement you are doing, I would zero the meter using the shorting plug and then measure the leads. The other question to ponder over, is what resistance value you should actually be getting. That would involve knowing the Ohms/m of the wire from the manufacturer and calculating to the lead length you have.

For the 4 wire measurement it looks like the sense leads are swapped, so 'high sense' is connected to 'low force' and 'low sense' is connected to 'high force', that could be responsible for the negative reading on the 4 wire test. You also need to make sure that the sense leads don't short to one another at the connection point.

The resistance values you are measuring are very low and also at the point where noise can affect the readings, especially when measuring test leads that are more susceptible to noise rather than the shorting plug.

Kind regards.

[omegaone]

Thank you for your response.

About the leads: My first test was to test the leads independently and if you look at the data, the numbers are very close. The connection was made as per the manual (minus the resistor) see attached picture.

And Keithley tech support has repeated my test on request using a common paper clip of approximately .02 Ohms using 4 wires.( Still separate operation from Tektronix )
I am also attaching their data here.

The Engineer performed the test was a veteran with 40+ years with Keithley just to give you some prospective.

I am also attaching a picture of calibration certificate here. It is fresh from factory.

About the noise:
You may have seen the thermal images I have attached. I took precaution to match the test piece to machine temperature the best I could.

Once, the resistance issue is solved, I have more important matter to pursue with microamps and DMM6500. As you know, the creator of this forum has put out his best engineered microcurrent product. I have one of the gold version sitting on my table ready to be tested.

Appreciate all your feedback.

Samuel

[threephase]

Looks like the Keithley engineer produced some good readings. Have you tried to repeat their measurements on your DMM6500?

Due to the angle of the photo, I am struggling to see how the leads are shorted together where they are ty-rapped. When I look at the two red leads, they look like they are tied together, but one is going to Sense Lo and the other to Input High?

I would be inclined to rearrange the leads to make sure that there is no chance of an accidental short between the exposed metal.

Kind regards.

[omegaone]

I apologize for the not so good pictures, but be assured the leads were connected as per the manual. I do repeat my tests over and over but not have time to take pictures each time.
So far my reading went to negative consistently. It also correlated to the AWG. ie. I went from 14 gauge to 12 gauge and the negative value increased as a function of gauge difference.

I have requested the Engineer to send me the same paper clip as a joke! May be he will. ( he is on vacation now)

In the mean time, I intend to test paper clip and will post results here. I find, I am able to get a better resistance value by using ty-rap. The reason is better pressure applied against the measuring surfaces and not impeded by other material ( ty-rap being and insulator)

I am not sure I have mentioned, I have a file with Fluke engineers as well ( Fluke Canada)

World famous Paper Clip update:  I did not have time to capture data, only posting the pictures with 2 wire vs 4 wire and (Temp, V, A)

I do not need to explain, pictures should speak for themselves about the Negative Ohms.
They did respond initially with a request to add temperature compensation between the lead and the meter, which I complied with. Now I am waiting for their response as well.

Note: I have deleted the 4 wire paper clip resistance readings from this post due to erroneous connection of leads to keep confusion to a minimum!
 ( Thanks to Threephase and Commongrounder)
2 wire connection and negative ohm readings still valid at this point.( 24 November 2018)


 

[alm]

Is the value of 200 pA really the current through the force leads. Not through the open current input jack? And 300 nV the voltage between the sense leads? If that's the case, then no wonder it's reading negative. At these levels, the Seebeck effect (1 uV/K between copper and copper oxide) and even sneezing at the other end of the room will swamp any signal. You'd need a guarded test setup to get reliable results.

I would measure the current through the force leads, and voltage between the sense lead, with other meters.

[threephase]

Thanks for the latest set of test pictures.

I am not sure what the secondary current and voltage measurements are relating to, as they don't seem to be representative of the resistance value obtained.

The thing that is of interest though is that the current is positive, but the voltage negative, I would have expected them both to be positive if the connections are correct.

I have tried to look at the test jacks and to me it looks like the top left (+ sense) has a black wire going to it but the top right (+input) has a red wire. But they should both be going to the same side of the paper clip. Or it could be that my old tired eyes are deceiving me yet again.

You could try reversing the sense leads and see if that gives you a positive resistance reading.

Alternatively, you could disconnect the left hand sense jacks and apply a low battery voltage across them. Battery positive to top left and battery negative to bottom left should give a positive resistance reading, if it is negative it may mean that the sense connections within the meter are reversed somewhere along the line.

I have done that test on my 8846A.

I have also tested a paperclip on my 8846A and get a comparable reading, although slightly lower, maybe you have a bigger paper clip than me

Kind regards.

[commongrounder]

Omegaone:  I have studied your photos, both the earlier set with the leads assembled with individual banana plugs, and the later set with the leads assembled with dual banana plugs, and observe in both setups that the polarity is reversed between the force leads and the sense leads. You need to connect red force and red sense to one end of the paper clip, and black force and black sense to the other end of the paper clip (as shown in the diagram you posted). Carefully trace out your leads and you will see a red lead in a black jack, and a black lead in a red jack on your meter. This will definitely give you a negative reading. Check it out and let us know if correcting the setup helps.

[omegaone]

Thank you for your correction. Both were incorrect just as you have pointed out. Relieved to see DMM6500 Does not go into minus in 4 wire connection.
I am attaching the new pictures 

I am also attaching Fluke 289 minus reading. I have used the same leads on DMM6500 and read in minus as well. See pictures attached.
 Have a look and any feedback would be appreciated.

[alm]

Glad you got the problem with the DMM6500 sorted!

As for the Fluke 289, on its 50 Ohm range it is specified for 0.15% of reading + 20 counts (or 20 mOhm), so for a 15 mOhm resistance, any reading between -5 mOhm and +35 mOhm is in spec, so expecting any kind of accuracy for resistances < 100 mOhm is very optimistic. Is the value of these leads (from jack to jack) <= 15 mOhm? You would need four write connections to the banana plugs to figure that out.

[2N3055]

Don't want to sound disrespectful, but you should really read F289 manual.
By specification, zero reading is not even specified for instrument.
It would make no sense for a 2W measurements because they cannot know what is resistance of cables used for measurement.
It is specifically said that only relative measurement is specified. Which is needed to compensate for cable resistance.
Calibration manual doesn't even specify what reading will be when you apply short on terminals before zeroing with REL function.
Even when you zero it out with REL it is specified as (as per FLUKE calibration manual):

46 LoΩ (289) 50.000 Ω range:  0.2 Ω applied = 0.2 Ω low limit:0.180 high limit:0.220
47 LoΩ (289) 50.000 Ω  range: 50 Ω applied = 50 Ω   low limit:49.905  high limit:50.095

Which is in line with specs: range 50Ω, resolution 0.001 Ω, Accuracy 0.15 % + 20. There is no specification for non-compensated zero whatsoever.

That reading, no matter how annoying might look is not a problem per se.
Real question is: if you zero out cables ( Allow reading to stabilize 5-10 seconds) and then measure known good resistor, is it ok then?
If it is, than meter is Ok, and negative reading is just a quirk, but has no practical implications.

[omegaone]

You are very respectful and thank you for your feedback.

Your practical test sounds very reasonable to do and I have done few with this particular meter. ( 50 W 1 Ohm 5 percent tolerance - precision )
There is a reason why I have started with 5 % tolerance and I have tested 10 resistors of similar. Max difference I was able to see was just over 1% with  relative function in place.
( lead resistance taken out of the equation)

Now if you are wondering why such high wattage, I also need to test them on a power environment (6 V, 6A , 1 Ohm config) and 36 W of power dissipation. Then I add time 15 min, 45 min so far.  Now another way to see the resistance of these resistors are by temperature dissipation.

So is essence, if I have identical resistors they should dissipate equal power when put in parallel and share the load.

Before one do all these, it is utterly important, to get the proper resistance measurement. (Because it is a function of temperature )

Now so far I have used Fluke 289 and DMM6500.
Both start off in negative with 2 wire config.

Threephase and Commongrounder are both carefully studies the pictures and pointed out the error in my 4 wire connection. Big thank you for both.

[commongrounder]

When attempting to measure very low values of resistance in 2-wire mode, things like the metals used in the banana plug, and temperature start to come into play. The plating on the banana plug is likely dissimilar to the copper in the wiring, and the metal used in the meter jacks. This can induce voltages that throw off the reading. There are others on this forum that know a great deal more than I do about these effects, but suffice to say, you cannot expect accurate very low value resistance measurements in two-wire mode. You should definitely be in the market for some Kelvin clips for your meter! This will allow you to match resistors with confidence.

[omegaone]

You can see, my place is not a fancy lab.
So I start off by using the tools I do have.

As you have pointed out, the plating on banana plugs, wire, etc plays  a crucial role in conjunction with temperature. ( thermoelectric effect )

I am doing one test at a time, so far, 2 wire test goes into minus territory 4 wire has not!

I intend to post every finding here. So any feedback constructive / destructive all are welcome. Important part is we find FACTS and we all get to share the FACTS and be confident in what we do on our  day to day operations.

Thank you

[2N3055]

Omega,
If I was helpful, my pleasure.

Thanks for explaining what you're doing.
Since you need only relative measurement, you can get very accurate results if you connect several of resistors you want to compare in series and apply external current from PSU.
Current doesn't need to be exact only sufficiently stable. For your purpose, normal lab PSU will be fine. Connect and let it heat up and stabilise.
Then with a voltmeter measure voltage drop across each resistor. With DMM6500 you can get VERY accurate results if current going through resistors is sufficiently stable during measurements.

Regards,

[omegaone]

Actually what I have explained is only part of it.  It is not what I am doing.

Anytime a measurement is made, the reference is vital.
So partly why I have been doing resistance measurement.

I am also involved in micro Amps. I have purchased Dave's product and very amazed at its capability. Prime example be DMM6500 does not have a dedicated Milli amp port. As a result any measurement one try to take encounters a big resistance at the input fuse. ( design flow)  At 10 microAmp selection at DMM6500 the measured resistance is 10.12 killo Ohms. What do you think will happen at this resistance if I try to measure  few micro amps with DMM6500?
In comparison, Fluke 289 internal resistance at micro amp setting with its dedicated milli amp port is only 100.52 Ohms. I get better micro amp measurements with 289.

 


Dave's Tool  gives me eyes to see at higher resolution with just any multi meter.  ( Before I came to know about Dave, I encountered this issue, but there were not much research in finding solution for this  issue anywhere. So when I did come across his microamp project, I was so thrilled to see some engineer is doing the job right!!!)

Burden voltage is a real problem and much more significant problem than the industry has realized.

I am really thankful for him to produce a product that can be used in this respect.

So the method you have  suggested does work but not in my case.
I will be back to resistance measurement inaccuracies again in a circle.