Repair/upgrade of Fluke 8840A Bench DVM
I picked up an Fluke 8840A Bench DVM at an estate auction. It "works", except that the resistance readings were wrong, and the AC "Option 09" wasn't installed.
Repair of ohms function
This repair turned out to be easy. The problem was oxidation on the pins of the big switch used to select either front panel or rear panel input jacks. I sprayed some contact cleaner into the switch and exercised it a few times and that seems to have fixed the Ohms function.
I determined it was switch "crud" because the ohms readings were OK on the higher ranges, but off in the low ohms range. This pointed to a fixed resistance in the input probes. 4-wire measurements were OK, which told me that the crud was specifically in one of the two 2-wire connections (i.e. the primary test input).
The missing AC "Option 09" function
AC measurement ranges are handled by an optional circuit board, which some Fluke customers did not want, so many 8840As in the wild have no AC functions. Fortunately for me, Fluke publishes the schematic of the AC board in the manuals for the 8840A. Further, there are photos online of the AC board, so one can get an idea of how Fluke laid out the components.
I figured "how hard could it be to duplicate the Fluke Option 09 board? The components are mostly commonly available, with none that are unobtanium (or can't be substited).
I entered the schematic of the Option 09 board into KiCad. It turns out that Fluke made at least two versions of the board, and I believe used it in other products besides the 8840A (specifically the 8842; perhaps others?) I made my Kicad version have options so that either version can be built.
I've used OshPark in the past for my circuit boards, and while the do a great job, they are expensive. As a test, I uploaded this design to JLCPCB (in China). The price of JLCPCB is much cheaper! A bit over a week later, I got my five circuit boards back.
I populated one of the boards (5 boards is the minimum buy from JLCPCB), and perhaps unsurprisingly, it "didn't work". Some study and testing revealed that I had copied the circuit of the Option 09 board used in the 8842 meter, and the pinout on the 20 pin connector was different between the 8840 and 8842 versions. The changed pins include the pin that is used by the motherboard to tell that the Option 09 board is plugged in.
Two bodge wires later, and now the board was recognized by the Fluke. No more "Error 30". However, the board failed self test. Investigation revealed that the output of the board was about 3v, instead of the expected 0 (with no signal applied to the meter). More debugging revealed a missing ground in the first op-amp stage, which was letting it oscillate, generating an AC voltage that the AD637 was dutifully converting to 3v DC.
Another bodge wire, and the oscillation was fixed. And then -- the board was working! With no calibration at all, the results are reasonable, maybe even sufficient.
The populated "version 1" board. The mounting holes had to be reamed out and "moved" a little bit, there's a missing ground that's been bodged in (upper left), and on the back, two bodge wires to swap pins 1 and 8 from the 20 pin connector. Also missing, the hole for mounting the shield.
The (version 1) AC Option 09 board mounted in "service position", showing it and my hand-held DVM both measuring an approximate 2V signal from my signal generator. (The version 2 board has the ribbon cable connector flipped so one doesn't need a twist in the cable.)
I've since fixed the errors in the Kicad files, including adding solder-blob options that allow one to select the proper pinout for the 8840 version or the 8842 version.
I also moved the mounting holes a bit further apart and enlarged them to 6mm. I also added a fixing hole for the metal shield, plus many silk-screen layer fixes.
I've ordered another batch of five boards from JLCPCB. Once I declare those "ready for production", I'll make the Kicad files available here, so you can make your own Option 09 boards.
Most of the parts are easily acquired. Some, like the AD637 RMS Converter, I buy as "used pulls" from vendors in China; so far, I haven't been burned. They're not cheap at $15 (used) but not stupidly expensive, either. The rest of the ICs are "common". Most of the resistors can just be 1% (or even 5% in some cases) 1/4 W parts.
The only tricky ones are the three in the custom ceramic laser trimmed thing, but I faked those "funny" values by using two resistors either in series or parallel (depending on the value), and measuring the end result. For example, the 1.11Meg is a 1.1M in series with a 10K.
The caps aren't special, either. The difficult one for me was the input cap, which has quite a high voltage rating; I bought one on eBay for what I thought an exhorbitant price, but then, I only need one. The cap footprints on the PCB are all sized to the caps in my parts bins; yours won't fit as nicely. Sorry. The only "interesting" cap is the 1.2pF; I used a "gimmick", as they're called: two short pieces of wire twisted together.
The transistor complement is comprised of 3 each of two parts. One is the common 2N3906, which you probably have in your junk box. The other is an undocumented NJFET, which in some Fluke literature is called "selected", Fluke #707968, and sometimes the literature says it's the same as "industry" part J2655. Good luck finding info on that part. Some discussion on the Fluke mailing list suggests that the J109 FET is a good sub; in the meantime I'm using the PN4343. The three FETs are all used as switches, so you want to select for low "on" resistance.
For the two reed relays, I'm using "HAMLIM HE751A0510 REED RELAY" that I buy 6 of for about $15 on eBay. These are form A relays with a decent high voltage rating, though probably not as good as the one that Fluke used. I don't work with high voltage, and if I did, I wouldn't use this meter.
For the plastic rivet-things that Fluke used to attach the board: I went to the FLAPS (Friendly Local Auto Parts Store) and bought a set of clips to hold plastic trim and body panels in cars. Mine are: "Needa" brand #458220, nominal 1/4 inch "body grill headlight rivets, comparable to GM 14019205, 14019206, 16514113, and Chrysler 34201621 or 6504090". The holes in the Fluke mounting bosses look to be 6mm, so these are a tiny bit too thick, but they work, and you get 9 of them in a pack, so enough for you and a friend, plus one to replace the one you drop that rolls under the radiator.
The body panel retaining clips I use to hold my board in the instrument.
The real Option 09 boards come with a shield covering the sensitive bits in the middle of the PCB. Some versions appear to be cast zinc or aluminum. This is massive overkill, which is good, since we can't do that.
We can, however, make a shield by folding up aluminum or copper (if you're posh) sheet in a shape similar to the Fluke shield. I thought about trying to duplicate the shape of the Fluke shield and went so far as to make a paper model, which photo is below. I decided that was way too complex, and settled for a simple five sided box with a partition added in.
Prototype paper shield model that I made as an experiment. I decided it's complexity was unnecessary.
I used some aluminum "cane metal" that I had in stock, and used "pop" rivets to hold it together. I used a threaded standoff to hold it to the PCB, and also soldered two small loops of wire to the PCB to keep the shield from rotating around the fixing screw.
Prototype shield installed on blank PCB.
Underside view of shield showing divider.
Another view of underside of shield.
Shield installed on the Option 09 board which is installed in the meter, showing plenty of clearance between top of shield and motherboard components.
Prototype shield installed on populated version 1 PCB.
Detail view showing soldered "ring" used to locate the shield so it can't rotate. There is another identical ring soldered on the other side.
There is also a shield on the "bottom" side of the board; I've seen two versions (in photos) of different sizes. The smallest shields just the first input opamp and associated components. I made one by taking a piece of thin aluminum "flashing" and gluing it to an old credit card (for insulation) and then bolting this to the bottom of the standoff that holds the top shield.
The bottom shield mounted on the board.
Finished rev 1 board installed in the instrument.
Builds by Others
Fabio P. from Italy built one, and he was kind enough to send me his photos. Note the snazzy Rev 2 PCB.
Fabio's board with it's shield in place.
Fabio's board sans shield.
Make Your Own
Below, two links. The first is a zip of ALL the Kicad files, in case you want to fiddle with the design (for example, change footprints to match the capacitors you acquire.) The second zip is only the Gerbers and drill file that you would give to JLCPCB or another board house to get boards made. (I believe OshPark wants the Kicad .pcb file.) In any event, download both zips as the "big" one also has the BOM files.
All the Kicad files
Q. Will you sell kits/boards ?
A. I have a few spare bare boards. Email "bill" at this domain for details. Other than that, no. (stock as of 2023 02 16: 2 old, 2 new boards)
Q. Will you fix mine?
A. No, I'm retired and so not interested in a job.
Links to this article
This is just my documentation for my repair. I don't claim that doing this is safe or recommended.
Soldering irons are dangerous, be careful. Oh, and don't eat the solder.
There are live mains voltages in this unit. Be careful when poking around in the power supply area.