Fluke 189 calibration problem

[dcharette]

Hello All,

I have a Fluke 189, which I bought new about 18 years ago. And it's still main main workbench multimeter to this day. The calibration on it is still relatively accurate based on comparison measurements to bench equipment I have, and it's close enough for my work.

I recently purchased an eBay 189 in pretty good shape. All of the measurements when compared to my original 189 and also a bench Fluke 45, seemed to be within less than 1% of each other, which is acceptable for me, with exception to the resistance and capacitance measurement. When I first was going through the ranges, the resistance measurement when shorting the test leads produced a 3ohm reading, that then took maybe 30 seconds to settle, and never to close to zero as a normal meter would. Also the capacitance range read higher than the other 189 with the same cap measured and also took 30 seconds or so to settle.

I dissassembled the unit, and removed the original supercap which appeared to be a little leaky, and replaced it. I was sure to clean the PCB completely, and even submerged the lower half of the meter up to the rotary contacts in an ultrasonic bath. Now, the meter on resistance settles normally after a second or two, but shorting the leads always produces 1ohm now, instead of 3 prior to the cleaning.

I've made a set of calibration resistors by going through my bins and selecting enough to get all the values listed in the calibration manual and with my original 189, and have confirmed they're close enough for my use. All, well below or within 0.1%. All these resistors I've attached to dual banana connectors for insertion directly in the meter without using the leads.

I've repeatedly tried to calibrate the resistance range using these static resistors, but cannot seem to get the meter to measure anything below 1ohm when using the dead short banana.

In the calibration manual, it states to NOT use a shorting bar on the mV or resistance, 0mV and 0ohm step. Also, I've performed the mV calibration prior to the resistance calibration a couple of times, and still the same result.

I'm not trying to get a dead nuts on accurate meter, just need that sub 1ohm resistance range to be a little closer to the actual calibration point as I use this range for looking for near and dead shorts on PCBs.

Well, I thought I'd just ask advice here to see if anyone can shed some light on why this 189 won't accept the calibration in the way that I'm trying to do it, or at least why the 189 won't get that first 0ohm point zeroed out.

Thanks for any help,
Dan

[coromonadalix]

are you using this manual ?

Fluke-189_Fluke-187_calibration-manual_Elso.pdf
https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwjotofDh8_-AhWukIkEHaHEDEgQFnoECAwQAQ&url=https%3A%2F%2Fwww.elso.sk%2Fmedia%2Fdownload%2Fpdf%2Ffluke%2Fman%2FFluke-189_Fluke-187_calibration-manual_Elso.pdf&usg=AOvVaw2YpdOW4WSeSU8TXezVKjmF


page 19  note  4     ...


You seems obseesed by the zero ohms, very low values  loll     in the specs  you dont see   much precision 500 ohms down to zero   at  0.01ohms precision they say,  0.5%    ?

The only thing i do at near zero values   is to use the REL function, you neuter the probes values .... and go with it  ... and yes  probes quality  paly an important role, tried many and saw sometimes huge differences, im using high quality Flukes  form the 289 series bought in surplus


They say you need a calibrator at his precision :  Accuracy: + 0.0125 %  ??

I have 0.035% precision resistors and the meter seems to follow   ... in y case i dont mind.


Unless mistaken  theses meters "talk"  with a calibration system to be done correctly ??

[dcharette]

Hello,

Thanks for the comments,

I'm using the calibration procedure that's available from Fluke, from the 189 support page.

I do use the relative function to "zero" out the probe leads as well when I'm really trying to track down PCB short errors and in some cases this is helpful.

I'm also using Fluke TL71 test leads when using the meter normally, but for this calibration, I've made up a series of selected resistors attached to dual banana connectors, and insert them directly to the meter without using probe leads to help counter act any measurement errors made if the probes were used. I hand picked these resistors and combinations by using my other known calibrated meters, another 189 and a Fluke 45 to do this, and the final resistors measure close enough for my needs.
 
What I'm really trying to understand with regards to the newly acquired 189 is why it does not act and respond like my original 189? And in the calibration procedure for resistance, I feel like the new 189 is not recording the values I give it using selected resistors using another known accurate meter.

I'm trying to understand why is it that on the zero ohm calibration step, it specifically says in the calibration manual to NOT use a shorting bar.

What I would like is to be able to calibrate this lower ohm range to somewhat get in the same feel of my other meters, without needing to use a calibrator or calibration service. Not that it's not an expense I'm willing to pay, but why shouldn't the calibration procedure at least get me close using the method I've described? Seems like it would be a reasonable first step that any electronics tech would take before committing to a real calibration service.

Thanks! 

[bdunham7]

What I'm really trying to understand with regards to the newly acquired 189 is why it does not act and respond like my original 189? And in the calibration procedure for resistance, I feel like the new 189 is not recording the values I give it using selected resistors using another known accurate meter.

I'm trying to understand why is it that on the zero ohm calibration step, it specifically says in the calibration manual to NOT use a shorting bar.

What I would like is to be able to calibrate this lower ohm range to somewhat get in the same feel of my other meters, without needing to use a calibrator or calibration service. Not that it's not an expense I'm willing to pay, but why shouldn't the calibration procedure at least get me close using the method I've described? Seems like it would be a reasonable first step that any electronics tech would take before committing to a real calibration service.

How much is the meter off by in the resistance ranges right now?  If the calibrator resistance standard (or test resistor) is off too much, then a new calibration constant will not be stored.  The threshold is ~15%, anything more and the meter software rejects the stimulus as an error.  It is important to realize that this could be an error with the standard, the wiring or the meter itself.

The software calibration system sets offset and gain constants so as to give an mx+b slope function to apply to readings.  Using a shorting bar instead of the calibrator might give you a better zero, but it will mess up the gain constant.  If you have a working calibrator and some wiring, you can be sure that the difference between the 0 ohm and 500 ohm settings on the calibrator will be nearly exactly 500 ohms, but you don't know what the actual resistance of the entire setup--including the wiring--will be.  If it was 1 ohm and 501 ohms for example, then using a shorting bar would give you a 501/500 ratio error, or about 0.2%.  If you get the gain constant right, you can REL out the residual resistance for low ohms measurements.  You can't fix a bad gain constant.  Using your methods, this probably isn't a big deal and has nothing to do with your issue.

The problem appears to be that you are trying to calibrate a broken meter.  If your DCV (and DCmV) ranges are good but your resistance range is not, there may be  a problem with the ohms current source or something else in the ohms circuitry.  First try to measure the current source output with your other 189.  The 500R range should put out ~1mA, the 5K range about 100µA.  If the current source is off or unstable, you won't get a proper resistance reading. 

[dcharette]

Hi bdunham7,

thanks for the reply and tips. I think I have some more information that at least starts to shed a little more light, at least for my understanding.

I have measured the current output in both ohms ranges you mentioned. I've done it using the 500ohm and 5kohm resistors that I made for the calibration purposes.

With the good 189, I put the unit in ohms mode, with the load resistor on a dual banana plugged into the input jacks, and then make a mV voltage measurement with another meter, a Fluke 45 in this case. And using that 500, I end up with almost exactly 50mV, which checks out with the 1mA at the low ohms range. Doing this measurement with the other 189, I get about the same. So, I've tried some lower resistor values, 10, 4, and 2. And what I get with the original and 2nd 189, is about the same, 10mV, 4mV and 2mV respectively. So, this leads me to believe the current source is Ok. Because, during that measurement, the original 189, showed the corresponding ohms values on the display correctly, but the 2nd 189 showed the values a little higher, 12, 6 and 4 ohms respectively.

With my original message above, my error was roughly 1ohm off, now for whatever reason today, it's a little more than 2ohms off.

Being it's different today, might be a sign pointing to the fault. But at least I feel the current sourcing from the meter is solid in the ohms function.

As well aside from the displayed measurement being a little more off today than my earlier post, the conclusions of this is telling me the calibration points for the lower range don't get updated correct when I do the calibration procedure.

And, at least with what you said about the 15% error... this might mean that because of the degree of the error I'm seeing, the calibration table isn't being updated. I get what you said about the gain function, and it makes sure sense now as to why you can't use a shorting bar.

With that in mind then, is this silly of me to think, but could I perhaps brute force that lower range little by little to get it within range to write the value I want? In other words, I could insert say 15% lower than what I see when a near short is there, for that low point, perform the calibration, see if it updates. Then, insert again 15% lower, see again if it writes, and so on until I get below 15% error for the 0 measurement, and theoretically, it would then update properly? Without a proper calibrator, would this actually work or is it silly? I realize this might be some crazy series of iterations into an asymptotic path of lunacy, never to be reached, but if I could get close enough to zero, maybe 10mOhms, I'd be happy.
I could use a pot with static divider on to fine tune it to within 15% by reading it with the other meter.

Thanks again for taking the time,
Have a nice remainder of your weekend, 

[bdunham7]

I have measured the current output in both ohms ranges you mentioned. I've done it using the 500ohm and 5kohm resistors that I made for the calibration purposes.

It would be easier and less error-prone to simply put the good 189 in the 5mA or 500µA range, connect the two together and read the current directly.  If the current is within 5% of the specification and is stable, then I'd agree that isn't the problem.  Your method isn't necessarily wrong, just  bit more convoluted.  The fact that your ailing 189 reads higher on resistance with the same voltage measured across it is not a failure mode I've encountered.  The instability is also an issue that needs resolving before you can get anywhere.

With that in mind then, is this silly of me to think, but could I perhaps brute force that lower range little by little to get it within range to write the value I want? In other words, I could insert say 15% lower than what I see when a near short is there, for that low point, perform the calibration, see if it updates. Then, insert again 15% lower, see again if it writes, and so on until I get below 15% error for the 0 measurement, and theoretically, it would then update properly?

With the caveat that there is no available service information that would answer this question directly, I don't think that will work.  The typical operation of a software-calibrated meter is that it takes a 'raw' (uncorrected) reading and then applies the calibration corrections to it and displays the result.  The raw reading is always still there, hidden in the process.  During this type of calibration, meters typically totally ignore any current calibration constants and will reject values that deviate too much from the expected raw reading.  Another way of thinking about it is that the calibration correction is limited in magnitude to ~+/-15%, so any calculated value larger than that is deemed an error and not stored.

But again, trying to calibrate a broken meter isn't going to work out very well.  You need to find the actual problem.  Can you verify that the mVDC range is accurate and stable to within your ability to compare?

[dcharette]

Hi bdunham7,

So, I have a little more info and a idea on the cause of the problem.

I spent some time attempting to calibrate the mVDC range. I don't have a proper calibrator, but am using my other two meters and a power supply for comparison and the two meters are within 50uV of each other throughout a range of 0 - 3VDC.

So, I have been attempting to calibrate the mVDC on the questionable 189, and what I feel like is happening is that the lowest point(s) either the 0 or the 50mV, on the calibration isn't saving correctly. I've done this several times and can confirm the upper points are saving because I purposely offset the point by just less than 15%, and found in the end, the measurements were then normalized to what my offsets were, as you would expect when calibrating. And, it follows almost exactly as you said with the slope equation. The farther off I offset it in the higher ranges, I can see the slope changing and fitting the offsets, but it pivots from either the 0 or 50mV point where it doesn't change at all.

What I am thinking at this point, is that the mVDC calibration data locations for the smallest either 0 or 50mVDC or both, aren't being written, or, more likely, are being written, but the EEPROM isn't actually able to write it fully. I have seen microcontrollers and EEPROMS in the past where there would be a hole developed in only one bit or multiple bits due to repeated writing. So, my theory is that part of the memory location for this one point is just not taking whatever I record. I know it's probably not due to repeated writing, but maybe there is just a dead bit(s) there not allowing a change and hence, I'm stuck with whatever value is already there, or the bit is such that it's in the higher part of the number and whatever new calibration value isn't enough to change the number to make a difference.

While it may be a longshot, I thought that I would get a new replacement EEPROM. Remove the old one, and read it with an external programmer to at least preserve the calibration and serial number, etc. data that is already there. Then, program a new one with that data. Install it in the 189, and attempt the calibration again.

Not a huge expense and a couple hours of time I guess. What do you think, is this nuts?

Thanks for the comments thus far, they've been incredibly helpful.

Have a nice week! 

[bdunham7]

While it may be a longshot, I thought that I would get a new replacement EEPROM. Remove the old one, and read it with an external programmer to at least preserve the calibration and serial number, etc. data that is already there. Then, program a new one with that data. Install it in the 189, and attempt the calibration again.

Not a huge expense and a couple hours of time I guess. What do you think, is this nuts?

I assume you'd be trying to avoid the need for the IR serial cable and the Fluke 87/89-IV, 187/189 Service Software (Fluke P/N 676152) that is called for in the calibration manual, since nobody seems to have this software.  I don't know one way or another whether there is an EEPROM issue, but I'd be surprised if it wouldn't generate an error if there was.  What exactly does your setup look like for providing the 0/50/300/500/3000mV stimuli?  Since you don't have an actual calibrator, you might try a shorting bar for 0mV just to see if it results in the EEPROM updating.