Fluke 8840A/8842A and AF Differences?

[Chris56000]

Hi!

Can any Members who are "au fait" with the Fluke 8840A/8842A and AF suffix models summarise the differences between them, what the suffixes mean, and suggest which model to chose for bench hobby use please?

Quality of the VFD isplay is now obviously not a problem as I've bought one of the new LED Display Replacement Kits!

Chris Williams

[AVGresponding]

The A/AF is a US military specced version (Army/Air Force) that has the capability to do ratiometric measurements using the rear terminals and an external v-ref: https://www.eevblog.com/forum/testgear/fluke-8840aaf-vdc-ratio/

[J-R]

For hobby use, I think you will be fine with any of the four permutations, just make sure it has the AC module installed.

[alm]

The 8842A has tighter specs than the 8840A due to better resistors, and possibly better selected parts. But honestly the specs of any of the 8840 models are great for a 5.5 digit meter. The 8842A also has some lower ranges (20 mV DC, 200 mA DC and 20 Ohm). Check the data sheet. Indeed pay attention that AC and GPIB were optional, and that the option check mark on the back may not be 100% reliable. The 8840A/AF came with the AC option installed, but that doesn't mean it can't have been removed later.

[AVGresponding]

The low current range on the 8842 is 20mA.

They are exceptionally good meters; I was lucky and got my 8840A with AC option (but no GPIB) for £100 shipped. I recently saw a parts only 8842A with broken display and obvious signs of being put together from multiple machines go for £160 + shipping!   

[bdunham7]

There's some incomplete or somewhat incorrect info in this thread, so I'll try to do a bit better as briefly as possible.

The 8840A and 8842A are similar except that the 8842A has an additional x10 input amplifer for 1 addtional lower range in each DC function--20.0000mV, 200.000mA and 20.0000 ohms.  It also has somewhat better accuracy specs and actual performance. Both meters only have one current shunt of 0.1R and the x10 input amplifier on the 8842A only works with DC, so the AC current range is 2A on both models.  Neither has the AC board installed by default and the AC boards do not interchange between models without some minor modifications. 

The 8840A/AF is a special model for the US government and is essentially an 8840A with the AC and GPIB boards installed as standard.  It also uses the rear terminals for a special ratio measurement function. 

[AVGresponding]

Huh, don't know why I got it into my head it had a 20mA range, my bad.

[J-R]

To confirm, 8842A/AF does not exist, just 8840A/AF?

[bdunham7]

To confirm, 8842A/AF does not exist, just 8840A/AF?

I've never seen or heard of one.

[J-R]

I guess I don't recall seeing an 8842A/AF either, and searching comes up with no proof.  Strangely, there are mentions kicking around, such as here: https://groups.io/g/TekScopes/topic/7644813
Probably the person posting that was just not being clear, and actually meant the 8842A.

I found a section of the Fluke catalog online which covers all the details about the 8840A vs. the 8840A/AF and also the 8842A: http://socal.mavin.com/pictures/ec/Fluke8842A.pdf

The 8840A/AF has:
- "expanded environmental envelope"
- meets some MIL spec
- true RMS is standard
- 1000V RMS AC input (vs. 700V)
- enhanced EMI/RFI protection
- DC voltage ratio

As mentioned previously, the 8842A offers the 20mV, 200mA and 20 Ohm ranges, enhanced accuracy and extended cal cycles, all due to "new hermetically sealed thin film resistors".

I have two 8842A units and one has a bit more drift than the other when using the additional lower ranges.  Also, you have to manually select these ranges, as it will not auto-range down to them.  To get meaningful results from the last digit, you will need to use offset.  The low ranges are very slow in S rate, although M rate seems good enough to use instead.

Tinkering a bit back and forth between the 8840A and 8842A, it's a tough sell to convince me the 8842A is really that much better.

I don't see that the GPIB board is a standard option on the 8840A/AF, both from the catalog and from image searching.

So which model to get?  I vote for whichever one is known working and comes with a current calibration.

[Original]

Is it possible to convert the rear inputs of an 8840A/AF to normal?

[J-R]

Is it possible to convert the rear inputs of an 8840A/AF to normal?

The 8840A/AF has other hardware changes and a different firmware, so I don't see this being a realistically possible mod.  Also, you'd end up with a hacked up instrument with multiple new limitations.

[bdunham7]

The 8840A/AF has:
- "expanded environmental envelope"
- meets some MIL spec
- true RMS is standard
- 1000V RMS AC input (vs. 700V)
- enhanced EMI/RFI protection
- DC voltage ratio

I don't see that the GPIB board is a standard option on the 8840A/AF, both from the catalog and from image searching.

It appears you are right about the GPIB as the 8840A/AF has a spot on the back to indicate the presence or absence of the option.  I don't remember where I got that they all had this, but perhaps the government contract included them and only the few retail versions actually appeared without.  Here's one in real life:  https://www.ebay.com/itm/175756230145

The 1000VAC statement is interesting, but the one spec sheet I have says otherwise.  A 1000VAC range would imply that there is a third version of the AC board.  I don't have an 8840A/AF in hand to look at or test, but if anyone does and is willing to give it 800VAC to see whether the display flashes I'd be curious.

https://nscainc.com/wp-content/uploads/pdf/F_8840A-AF.pdf

Tinkering a bit back and forth between the 8840A and 8842A, it's a tough sell to convince me the 8842A is really that much better.

It is an incremental improvement to be sure, but needing offset to zero out a range with 100nV resolution isn't at all unexpected.  That is the equivalent of the last digit on a typical 6.5-digit DMM.  And I've found that out of about a dozen examples, the 8842A long-term calibration stability is noticeably better.  But I'll agree that any of these in good condition is an excellent choice for a budget bench meter.

[J-R]

I believe the 8840A/AF 1000V AC input is correct.
First, the labels on the 8840A and 8842A list both 1000V DC and 700V AC, but on the 8840A/AF it only states 1000V MAX and does not split out AC/DC ratings.
Second, if you look at the 1990 Fluke catalog, under the 8840A specifications section it states: "Maximum Input: 700V rms, 1000V peak or..." but under the 8840A/AF it states: "Maximum Input: 1000V rms or...".

Given the age of these units, maybe they have stabilized enough that there isn't as much of a difference between the 8840A and 8842A that there may have been in the past.  But really to clarify my point on which to select, I'm just saying I'd rather have a known-good unit with a fresh calibration and data rather than blindly choose one model over another.  Also, the speed comment is because one of the main advantages of the VFD and this model is the fast update rates, but the low range is achingly slow in S mode.

[bdunham7]

I believe the 8840A/AF 1000V AC input is correct.
First, the labels on the 8840A and 8842A list both 1000V DC and 700V AC, but on the 8840A/AF it only states 1000V MAX and does not split out AC/DC ratings.
Second, if you look at the 1990 Fluke catalog, under the 8840A specifications section it states: "Maximum Input: 700V rms, 1000V peak or..." but under the 8840A/AF it states: "Maximum Input: 1000V rms or...".

I agree it looks like there is some difference there, but the specifications also show the highest AC range to be 700VAC.  So I'd be curious to know if it will measure all the way to 1000VAC without flashing or if it is simply rated to withstand that much voltage without blowing up.  AFAIK there isn't an applicable schematic available to resolve the issue, so perhaps some day I'll get the opportunity to dissect one and see for myself what is in there.

I'm just saying I'd rather have a known-good unit with a fresh calibration and data rather than blindly choose one model over another. 

Hmmm, I sort of think that the whole point of getting these meters is that you can rely on them to be accurate without recent calibration, provided that they aren't broken.  Calibration sounds nice, but it costs money and cheap calibrations don't prove much.  I've yet to see one that is out of spec without having an obvious major malfunction, and in every case where I've repaired them they go back to being very accurate.  The only way I know of to get one to be slightly inaccurate is to swap in a different AC board.   

Of course, my opinions date back to when you could scrounge up a fairly decent 8842A with AC and GPIB for under $100.  With a handle.

[J-R]

I still couldn't see Fluke printing 1000V MAX on the front of all the 8840A/AF units, where the AC board is not optional, for military customers.  So that got me thinking more about this.

I went through the True RMS theory of operation on page 5-19 of the 8840A manual.  The block diagram shows how the input is scaled based on the selected range.  In all cases, the output from the TRMS board is always 0-2V to the track/hold circuit:

- 200mV, r1: divide by 5, multiply by 5, multiply by 10 = 2V
- 2V, r2: divide by 5, multiply by 5 = 2V
- 20V, r3: divide by 5, multiply by 0.5 = 2V
- 200V, r4: divide by 500, multiply by 5 = 2V
- 2000V, r5: divide by 500, multiply by 0.5 = 2V

It also states that the input is clamped by MOVs to 1560V.  So 1000V RMS (1414V) should be no problem, even for the 8840A and 8842A.  Of course different firmware would be needed, and I suspect some of the input protection is different on the 8840A/AF since 1560V might be a bit close to 1414V.


With regard to the calibration discussion, I don't see how you could rely on anything without a somewhat recent calibration (self or otherwise).  That is the entire point, provide evidence that it works correctly so you CAN rely on it.  In the context of this thread, the OP is not going to have the necessary equipment to calibrate it himself, so that means paying for it.

[alm]

With regard to the calibration discussion, I don't see how you could rely on anything without a somewhat recent calibration (self or otherwise).  That is the entire point, provide evidence that it works correctly so you CAN rely on it.  In the context of this thread, the OP is not going to have the necessary equipment to calibrate it himself, so that means paying for it.

If you need to prove to a third party that what you measured is correct, sure. Like in any safety critical industry.

But for a home lab, I disagree. In my experience this class of meters will usually be pretty close to its 1y spec, or way off. So if I could spot check on a few values, and those values are correct, I'd feel pretty confident. I say that as a person who in the past compared used 4.5-7.5 digit DMMs as best I could with improvised signal sources to look for inconsistencies. And the only deviation from 1y specs I could measure was a meter which had compents in the ADC circuit replaced, a meter which was just always reading zero for ACV, and some differences in high frequency ACV.

How often do you do a measurement where absolute accuracy down to say 50 ppm matters? For me the high resolution DMMs are more about relative measurements, like the change of a resistor with temperature, rather than absolute values.

[J-R]

Come on, you guys know better than to keep pushing back on the calibration stuff.  It's fine if you don't want it or need it, but to keep making these claims that just checking a few things makes you "confident" is just 100% false.  Don't use those words around others because you're just plain wrong, sorry to say.  Go ahead and say "good enough for my purposes" or something similar instead.  And it's also fine if you have the equipment to check everything, as do I, but most others do not and its a big step to get there.

These specs are not published by Fluke for just for the fun of it, the devices have their limits and need to be calibrated regularly if you want to rely on it.  There are no magical low-drift virginal vintage ones out there that you can just check a few data points and then assume everything is fine.

The 1 year rating on these units is atrocious if taken at face value, especially the extra low range on the 8842A.  The last two digits mean nothing.  I wouldn't have it around if it only met spec.  As an example, the 1 year rating for 2V would be 8 counts. The 8842A I self calibrated/adjusted almost 2 years ago still bests the 24 hour spec.  That is one of the big advantages of these older units, but it needs a somewhat recent calibration (SELF or otherwise).

Also, it is rare to find these with intact front and rear calibration stickers.  The back end can be taken off and the calibration switch manipulated without disturbing the front sticker.  So you don't know if there is some random function/range that is off because the previous owner was tinkering with it.  A professional calibration will typically have a lot more data points than just the published performance test points.

[bdunham7]

There are no magical low-drift virginal vintage ones out there that you can just check a few data points and then assume everything is fine.

Out of a dozen units or so I have worked on I have never found one to be out of spec if it powers up, reads nominally correctly and passes a self test.  The only time I have changed any calibration constants has been when I replaced, added or repaired the AC boards.  Has your experience been any different?

The 1 year rating on these units is atrocious if taken at face value, especially the extra low range on the 8842A.  The last two digits mean nothing.  I wouldn't have it around if it only met spec.  As an example, the 1 year rating for 2V would be 8 counts. The 8842A I self calibrated/adjusted almost 2 years ago still bests the 24 hour spec.  That is one of the big advantages of these older units, but it needs a somewhat recent calibration (SELF or otherwise).

Well that is an interesting perspective.  Aside from the zero/offset issue in the low ranges of the 8842A, these meters have better specifications than any other 5.5-digit meter that I'm aware of and they are much, much better than modern 5.5-digit models.  So if you 'wouldn't have them around if they just met spec', what would you have?  Of course they were expensive and 30 years ago they cost as much as a decent 6.5-digit DMM does today, so they are sort of in their own category.  I have 8842A models that look like a truck ran over them and they meet their 24-hour specs, 2 counts or better on the 2V/20V ranges.

So you don't know if there is some random function/range that is off because the previous owner was tinkering with it.  A professional calibration will typically have a lot more data points than just the published performance test points.

That's true, I suppose, if you only have one meter and no accurate voltage standards and no idea where the meter came from.  But if you have only one standard and a few parts, say 5V or 10V and some resistors, you can easily check all the ranges by comparing them with each other.  The nightmare scenario is if someone has attempted the ADC linearity calibration and not done it properly--that would be difficult to recover from without a calibrator.  This is one of the reasons that I would prefer a unit of unknown origin over one where the seller says "gee, I just calibrated it!".  Checked calibration=good, adjusted=no bueno, IMO.  Unless there is a convincing explanation.

As far as your offset/zero complaints go, this just goes with the territory when you have that extra gain with a meter that doesn't use autozero.  It takes a few extra steps to get accurate readings at very low levels, but it is perfectly manageable.  There's a reason that the autorange doesn't go down into those levels.