Fluke 8840A - calibration button depressed during startup

[Rax]

Hi all,
I've just been handed over a bunch of instruments from someone in my neck of the woods that is downsizing/moving.

One of these is a Fluke 8840A, in very good shape. Upon starting it up yesterday to see what's going on with it, it displayed something line 0.00 and I noticed the "CAL" notation on top of the screen. Grabbed the manual, which states the calibration button should not be depressed until one intends to perform it, and never during a power cycle.

Oopsie, I thought, turned it off, switched off the calibration switch and turned it back on. Well, everything seems normal... Self tests are successful, it measures voltage from my bench source pretty accurately (I don't really have confirmed/calibrated instruments to check - just my "nothing wrong with them" bench DMMs - but I have no reason to think anything is wrong with the value displayed).

Shall I worry about its calibration memory being wiped? I am not sure what else could happen during the mishap at first turn on.

Thanks a bunch for input!

[Rax]

Nothing wrong I can find with this unit (DCV and R tested this far)... Maybe I was just incredibly lucky.

One aspect I could use other owner's inputs is the values displayed when it sits idle (DCV). Upon starting, it went to just under +100mV DC. Did all tests, it then dropped to under 1mV, though now - a few minutes later - it fluctuates between maybe +10 and -15mV - though right now creeping on the negative side. This is without leads. If I attach leads and short them, it goes down to maybe 0.001-0.002mV (and squares on zero now and then).

I don't expect it to sit at 0.0000, but just wondering if I should worry about this much mV observed. A few mV sounds about right to me, but almost 100mV feels a bit odd.

[Rax]

I let it go for maybe a couple of hours, and the idle voltage measurement creeped to about -.78V. Almost one volt!  If I short the leads, it resets to basically zero, and then the cycle starts again.

Not a huge deal, as actual measurements seem correct, but just an irky thing is all. I'd be good getting to the bottom of it. None of my other DMMs do this.

[lowimpedance]

The input impedance on the 200mV, 2V and 20V ranges is much greater than 10M , so this drift behavior is normal. Go to manual ranging and select the  200 or 1000V range which is spec. at 10M and see that it will settle back to near zero like shorting the leads.
 With a bit of luck the CAL constants are not corrupt, only way to know for sure is to cross check with known sources and another trusted meter.

[bdunham7]

Powering it up with the CAL switch in will wipe the calibration constants.  On the 8840A/8842A models I try really hard not to do that even if I'm repairing something that will require some calibration later, such as an AC board.  The reason is that the ADC calibration process is lengthy and fiddly--and usually should be left how it comes from the factory.  The default settings won't typically be terribly far off, so the readings may not be all that wrong.

The ramping up of the DC voltage is completely normal and a result of the very high input impedance.  It's not broken.  None your other DMMs do it because they have a 10M resistor across their input amplifiers.

[jonpaul]

I recall the Fluke 8840A..42A do NOT write CAL constants unless

1/ CAL button is in

2/ A particular CAL routine is completed.

So I think the old CAL is retained.

I would check at least the 2V range with a known good DVM or Voltage standard.

The manual is available.

Bon Chance,

Jon

PS: bdunham7 the ADC cal is not difficult but a good Kelvin Varley divider and voltage reference is needed.

The process is iterative and sucessive passes improve the ADC cal.

[Paceguy]

The same thing happened to me with a Fluke 8842A. I turned it off, pushed the CAL switch to the off position and turned it back on. Everything appeared to be OK, so I checked it out on my 5100B and everything tested within spec. I could be wrong, but I believe that as long as nothing is pressed on the key pad and the CAL procedure has not been started, you may be good to go. It doesn't look like anything was wiped out

I was given a Fluke 45 earlier this year, That when powered up showed CAL on the display, and was in the calibration mode. It appears that someone had started CAL previously to it being given to me or someone put it into the CAL mode and started pressing buttons. With that one, I had to put it on the calibrator and calibrate the DC portion. Once that was done, the CAL indicator was off, and I proceeded to verify it completely on the calibrator. All was within spec. It appears that only the DC part of it's calibration was effected.

[Rax]

Thank you all, I'm bouncing up an down between heaven and hell with every new post... 

I will try more ranges and compare with my Fluke 189 - most similar unit I have - though the same bounty included a Fluke 8502A which, in case I fix it, may be my new bench reference. Interesting in this context, the lab having this other unit (8502A) - actually all these units come from this source -  physically removed the calibration button, probably to avoid any mishap by junior engineers...

Now, one of my day dreams is to find a hobbyist's calibration facility (or someone's "turbo-bench?..."). Are there any known providers of calibration services for those on a budget? I though I'd ask... This would be in the US, but I don't mind shipping if that's the only way.

[bdunham7]

The manual explicitly says that the meter will be uncalibrated if it is powered on with the CAL switch in.  It is stated in multiple places in different ways.  If that doesn't actually happen, that would be interesting to know.  Perhaps different firmware versions behave differently?

I would expect the greatest uncalibration error to be in AC and resistance ranges.  If those closely match your 189, then it may indeed not be uncalibrated. 

I see you are in SoCal.  If need be, I can calibrate it--an 8840A is right at the edge of what I'm sure of with what I have.  As for your 8502A, there's another member with one that is also fixing his right now, so there should be thread somewhere on that.

[Rax]

OK, I may have good news?....
For AC voltage, I used my AP to output 2V @ 1kHz, which its own RMS voltmeter sees at 2.005V.:

• Fluke 189 - 2.0053V
• Fluke 8840A - 2.0045V

I'd say close enough for comfort on the calibration constants being there, no?...

A 2.2k 1% resistor shows as 2.2015k. This also seems close enough for comfort.

Shall I conclude all is good?... My take is yes, but would like your vote.

Thank you all for your help.

[bdunham7]

That does look promising.  I wouldn't expect the AC to be that good without proper calibration constants, but the resistor could be due to chance since it is a 1% tolerance.  You might want to check as many ranges as you can by cross-checking with the 189. 

[Paceguy]

The manual explicitly says that the meter will be uncalibrated if it is powered on with the CAL switch in.  It is stated in multiple places in different ways.  If that doesn't actually happen, that would be interesting to know.  Perhaps different firmware versions behave differently?

I would expect the greatest uncalibration error to be in AC and resistance ranges.  If those closely match your 189, then it may indeed not be uncalibrated. 

I see you are in SoCal.  If need be, I can calibrate it--an 8840A is right at the edge of what I'm sure of with what I have.  As for your 8502A, there's another member with one that is also fixing his right now, so there should be thread somewhere on that.

It's possible that Fluke states that the meter will become "uncalibrated" if the CAL switch is activated in the manual without going into too much detail because it is possible that in certain cases such as depressing buttons, etc. and they want to avoid the possibility of operators doing so then accusing Fluke of not giving sufficient warning. I do agree with you that they do state this in several places and you have reason. I always avoided activating that switch due to this. My only experiences have been with 8840/8842A's and 45'sthat I received with the CAL switch already on and the CAL indicator on. Maybe I have just been lucky up to now.

[Rax]

Two points on this:

• I think Fluke was indeed being as precise and diligent as they could in their language over this control. But electronically, it may not necessarily mean the records are being immediately dumped (or "zeroed," so to speak). Often, with non-volatile memory, old records persist until new ones are written over them (regardless of other actions). My 2c.
• Take this with a grain of salt, but I'm more and more convinced the records survived. I am also working on a 8502A, and one of its core precision voltage references is measured by the 8840A as "7.0000V"

Hope this empirical data helps others in similar situations.

[Paceguy]

I tend to agree with you, it's similar to the conclusion I have also come to.