Fluke 87V & 28II problem at the ohms range .

[Kiriakos-GR]

Well I just discover an true problem with the Fluke 28II  by comparing it with the Agilent U1272A and the Fluke 8050A (bench top) .

And the problem is that the Fluke 28II adds  100mOhm on the 600 Ohms range , because it just does so .
The same thing was happened also at my Fluke 87V and now I know why … because it just does so too.

And why it did bother me ?  simply because all those tests that I did on test leads and their tiny resistance,
I was getting an a resistance  of  00.15  when I was crossing the leads and I was believing that the meter was saying the truth .
But no it was not !!
The true resistance  was 100mOhm less.
And in this example would be  00.050  that is just 50mOhm

I did a simple test by measuring the inner resistance of my own Resistors decade box .
With a fixed set of cables with bananas at the ends.
And the numbers that I got from them : ( with No REL) (cables & switches & bridges)
Fluke 28 II  ……………. 0.15  ( High res 20.000 counts )
Agilent U1272A …..….00.049  mOhms   (30.000 counts)
And the Fluke 8050A    00.06  (07 fluctuates)    (20.000 counts)
And the point is that all of them was measuring about 50 mOhm !!

And I got smokes from my ears , because I had take about a hundred  of pictures with measurements so far,
and I had posted them in the EEV for over one year now, with both the 87V or the 28II , and all my measurements today looks wrong.

Are those Fluke meters damaged? , hell no but there is a bug in them ..
Do they measure accurately?  Yes if you use the relative first so to manually zero out the resistance of the leads .

And then where is the problem ? 
The problem is when you need to measure the resistance of the leads it self.
This additional 100mOhm makes the measurement to look fake.

Is there a cure ?
No there is not,  the only thing that I can think of , are that in order to measure the resistance of the test leads,
I will had to use a dual banana bridge on the DMM, so to zero out  ( RELative)  first , and then to add the leads so to measure their own resistance.
But this sounds crazy too .. 
Because even if I need to get close to the correct measurement, this bridge banana it should be made of gold !!   

Before you call me as nerd because I care that much about micro ohms, just think what you are when you care that much for the micro amperes.

And now that we got even, I am ready for your comments if you have any. 

[Joshua]

Wow, that's kind of dissapointing. You'd figure that if it's consistent like that, they could almost fix it in software.

[Richard W.]

I don't see any malfunction. Lets do some math:

We want to measure a resistance 50m?. The Fluke 87V uses a measuring-current of about 1mA.
According to ohms law this means a voltage of about 50µV at the resistors terminals.

Then consider things like:
~ thermoelectric voltage (because not every contact has the same temperature and doesnt consist of the same material, there is copper, brass, solder, ect)
~ electrochemical voltage (humidity, residue of fingerprints,... ect..)

Your Agilent U1272A has a function called Smart ? which seems to compensate some of this effects.

I could not find any data to the measuring current of the 8050A, but i guess its higher than 1mA.
This also lowers the influence of thermoelectrical and chemical voltages in the contacts.

[Kiriakos-GR]

Well for the moment, what matters is to make a list of the models with this awkward behavior .

The Old Fluke 8050A looks healthy .

The 87V and the 28II is not.
I would expect from the members in our forum,
to do similar tests at the older versions of the 87,  like  like 87 & 87III.
Or at any Fluke meter that has 20.000 resolution at list.
And report your findings here. 

Just cross the leads, and tell us the reading. 


 

[Kiriakos-GR]

Your Agilent U1272A has a function called Smart ? which seems to compensate some of this effects.

Hi Richard, about the Agilent and Smart Ohm, this is another function by it self on the ohms range and needs to be manually activated.

All that I do here is to compare the ancient classic Ohms range, and nothing more. 

[Alex]

Kyriako, did you check the manual of the 28II first? How about you Joshua?

Look at the the resistance specifications on page 48. In the 600 Ohms the resolution claimed is 0.1 Ohms aka 100 mOhm. Also, the accuracy quoted is 2 counts when using the relative function. Despite these specs you are asking it to accurately measure 50mOhms?! If anything, it is exceeding its stated specifications.

http://assets.fluke.com/manuals/2x_2____umeng0100.pdf

Alex

[Richard W.]

Well for the moment, what matters is to make a list of the models with this awkward behavior .

The Old Fluke 8050A looks healthy .

The 87V and the 28II is not.
I would expect from the members in our forum,
to do similar tests at the older versions of the 87,  like  like 87 & 87III.
Or at any Fluke meter that has 20.000 resolution at list.
And report your findings here.

Just cross the leads, and tell us the reading. 

We can do this but you can not compare the readings. There are so many influencing factors:

thermoelectric voltages and Electrochemical voltages
to avoid them you have to have the same temperature and humidity and materials.
All multimeters must be in the same room and need enough time to acclimate.

electromagnetic interference
Here you need also the same conditions for all the meters. Because not every meter reacts equally -> Zero Radiation

internal components
mechanical stress can cause problems.
Just bend a pcb with a perecision reference ic. Several 100µV difference are usual.
Note: temperature and humidity can stretch and compress a board. (Coefficient of thermal expansion)

Age of the internal components
The specs are drifting, mostly according to the quality and the knowledge of the engineers.

And last but not least: are the batteries fresh? 

Accuracy is not infinite, even in greece 

[alm]

The test also depends on the real resistance of the leads (need to do four wire wire measurement to verify). If EMI is an issue, the loop area and orientation may influence the reading. If you just press the probes together, the pressure on the probes will also be a factor.

I agree with Richard about the accuracy specs. I would also like to know how you're so certain that the 87V/28II is off, and not the 8050A? Based on these results, my guess would be that the 8050A is 100mOhm low in this measurement, but you need more data (eg. a four wire measurement, preferably with a recently calibrated meter) to be sure.

[Richard W.]

The test also depends on the real resistance of the leads (need to do four wire wire measurement to verify). If EMI is an issue, the loop area and orientation may influence the reading. If you just press the probes together, the pressure on the probes will also be a factor.

The test leads are fixed to his resistor-decade-box.
-> The resistance from one bananaplug to the other is always the same.

Another factor is the inner diameter of the 4mm-jack on the multimeter. The tighter, the lower the resistance.

@Kiriakos-GR:
To be sure about the resistance, you need a 4 terminal measurement (as alm said)!


If you don't have a meter with 4 terminal function:
~ set your power supply to a fixed current (about 1A; don't need to be accurate)
~ connect a multimeter [1] and your resistor-box with the fixed testleads in series to measure the flowing current
~ measure the voltage over the banana-plugs (the ones on the fixed testleads) with a second multimeter [2]

U (multimeter [2] ) / I (multimeter [1] ) = R

This should be more accurate as measuring the resistance direct with the multimeter

[Alex]

Ehem...does any of this matter for the original post? The manual basically says you can't measure 50mOhms.

[alm]

The test leads are fixed to his resistor-decade-box.
-> The resistance from one bananaplug to the other is always the same.

But it does make the suggestion for other people to also measure lead resistance totally useless.

~ set your power supply to a fixed current (about 1A; don't need to be accurate)

It has to be stable within the few seconds it takes to do a measurement, though. Another issue is that DC current accuracy, especially for higher currents, is usually worse than DCV/resistance. Apart from that, it should be more accurate than a normal four wire measurement, due to the higher current.

[ElektroQuark]

Ehem...does any of this matter for the original post? The manual basically says you can't measure 50mOhms.

No, it doesn't matter, but they are playing and threads have their self life, you now

[nukie]

Anything in the mOhm, move over to 4 wire measurement please.

[Kiriakos-GR]

Kyriako, did you check the manual of the 28II first? How about you Joshua?

Look at the the resistance specifications on page 48. In the 600 Ohms the resolution claimed is 0.1 Ohms aka 100 mOhm. Also, the accuracy quoted is 2 counts when using the relative function. Despite these specs you are asking it to accurately measure 50mOhms?! If anything, it is exceeding its stated specifications.

http://assets.fluke.com/manuals/2x_2____umeng0100.pdf

Alex

High Alex ,  you are correct about the bibliography, but those are the specs for the standard 6000 counts resolution.
At 20.000 counts I do my comparisons.
If I accept the fact the the 87V and the 28II can do measurements from 100mOhm and above,
the conclusion would be that the extra resolution on those DMM's are meaningless. 
 

[Kiriakos-GR]

Anything in the mOhm, move over to 4 wire measurement please.

Because of this post and the other one from Alm about the 8050A calibration,
I will just post 10 pictures of my 20 years old 8012A with 2 Ohm + 20 Ohm range (2.000 counts)
Plus the U1272A , plus the 20 years old  8050A.

Just read the title of the pictures, of what is what.
I feel good that I have this lovely Fluke 8012A, even today it does put even newer DMM in to shame .. LoL
( At list in the mOhms range )   

Enjoy ... No four cables measurements.. 

[Alex]

Kyriako, did you check the manual of the 28II first? How about you Joshua?

Look at the the resistance specifications on page 48. In the 600 Ohms the resolution claimed is 0.1 Ohms aka 100 mOhm. Also, the accuracy quoted is 2 counts when using the relative function. Despite these specs you are asking it to accurately measure 50mOhms?! If anything, it is exceeding its stated specifications.

http://assets.fluke.com/manuals/2x_2____umeng0100.pdf

Alex

High Alex ,  you are correct about the bibliography, but those are the specs for the standard 6000 counts resolution.
At 20.000 counts I do my comparisons.
If I accept the fact the the 87V and the 28II can do measurements from 100mOhm and above,
the conclusion would be that the extra resolution on those DMM's are meaningless.

Yeap, thats correct as the manual specifications characterise the limitations/capabilities of the actual measurement hardware, whereas the # of counts should simply reflect that. In this case the firmware should not be displaying digits that have greater error than the digit value. In other words the number of counts is not correct in that range. I'll attribute this to Fluke and rest my case. You are a tough man to convince Kyriako! Pfff 

I haven't seen an audio casette tape for years, not to mention on a bench!   
We can tell your are putting the 1272 through its paces; its battery is running low.

[Kiriakos-GR]

You are a tough man to convince Kyriako! Pfff 

Yes I am the crazy one of my village, its true, but I have a talent on finding things,
or at discovering answers when I troubleshoot a problem.

The latest Fluke catalog 2010-2011 called as  11605-eng-01-a.pdf

Check the picture at the bottom.
I would declare my self as totally crazy, if Fluke send a message to me,
saying that the Fluke 28II does not have identical ranges, or resolution with the 87V.

I haven't seen an audio casette tape for years, not to mention on a bench!   
We can tell your are putting the 1272 through its paces; its battery is running low.

On the U1272 I use NiMH on it, so to check the battery life, and I do run battery consumption tests.
But this is a secret.. yet 

[Richard W.]

Thats resolution, this means this is the smallest step between to values.

look at this picture: http://www.mw-import.de/images/big/buegelmessschraube-25-50-klassik.jpg
resolution is 0,01mm, but the smallest possible value is 25mm

It's the same with your multimeter.

[alm]

Yeap, thats correct as the manual specifications characterise the limitations/capabilities of the actual measurement hardware, whereas the # of counts should simply reflect that. In this case the firmware should not be displaying digits that have greater error than the digit value.

Less resolution than accuracy is bad, about one digit more resolution is optimal (to reduce quantization error), more is not very useful from accuracy point of view, but may be useful for relative measurements (as long as the linearity is good enough). I believe we've discussed this issue somewhat recently. The extra resolution doesn't hurt, it just doesn't give you extra accuracy. It's up to the user to be aware of the accuracy specifications and calculate the measurement error. It's rare for a meter to be accurate down to the last digit.

[Alex]

On-screen quantisation error (we dont know whats going on inside) will be limited to half a least significant digit, i.e. 50mOhms for the 28II in the 600 Ohm range. We must also consider the 'counts' error which is 2 counts i.e. 200mOhm for the 28II in the 600Ohm range. If the multimeter had an extra digit (600.00 Ohms) then the counts specification would be wrong as the resolution would refer to the second decimal from the right and the count to the last decimal point.

So for the 28II, 600.0 Ohms range with an input of 0.1Ohms, according to the manual we have ±(0.2 % + 2) using the REL function ofc:
±(((0.1*0.2)/100)+0.2) Ohms = 0.1 Ohms ± 0.2002 almost equal to 0.1 Ohms ± 0.2 Ohms. For higher resistances the %of error drops off quickly as you would expect.

I would not use this DMM for accurate measurements below say 10 Ohms. The Fluke 289 (50000 counts) has a 50 Ohm range with 1milliOhm resolution, 0.15 % + 20 counts accuracy error. Even that is pushing it.