It’s time to see what the flea market special $5 Fluke 8000A looks like up close
Step one: clean it up.
After blowing the dust off and cleaning the outside we can now check the basics. Main fuse is intact, current input fuse is blown. That could potentially happen in normal use with a simple overload, so not a sign of troubles just yet.
Step two: try powering things up.
Plugged it in, pressed power button- a big nothing in return. No lights or other signs of life. I guess it’s tme to check voltages. Looking at the Service manual courtesy of BAMA (Boat Anchor Manual Archive), there are three main power supplies, +5V, +15V and -15V.
Measuring those I get +5.5V, +15V and -12V. Something also smells hot. That something turns out to be a custom Fluke IC U3 (analog IC as they call it in the manual).
Step three: chasing the troubles.
So we now have a chip that’s running hot. For that it either has to be damaged and drawing more current or be loaded. To test the idea, I remove the IC. Immediately the -15V rail sprang back and became -20V. Ugh, I guess there are other troubles lurking. Now U4 (Digital IC) is getting hot, possibly because of that. Ok, lets remove it as well. Thankfully all DIPs are socketed and are easy to remove. And what do we have now? 15V rail is still +15, -15V is now -23V and what does +5V look like? A nice +6.5V. Not surprisingly U5 SN7447 is now warm and unhappy.
Removing that, we are now left with two analog only ICs that are still cold. In the absence of U5, +5V rises even more. What a nice design! Going back to schematic, it’s easy to see why 5V rail is so load dependent. It’s basically produced the way cheap wall transformers do that- a secondary winding, with a rectifier and a cap. There is simply no regulation! A bit unexpected from Fluke that they’d feed digital logic from such a source. Moving on to -15V, we see Collector of Q19 at -24V, Emitter at -23.5 and a base at -15.5V. The base reference is obviously doing its job, but Q19 is not.
In a moment of not paying attention I proceeded to desolder a wrong transistor, Q26 instead of Q19. Q26 is a 2N3904, but not your today’s 0.1 cent part:
The original part is 2N4403, a general purpose 600mA PNP transistor.
Measuring E-C resistance quickly proves that the part is indeed toast:
The only one I had in my junk bin was 2N3906, rated to 200mA. Figuring I can at least test with it, I swapped it in. Well, now the C B E read as -24.3V -15.5V -14.8V. Much better! At least analog rails are behaving. Placing U3 back in its socket, I have a tiny hope it will stop heating. That of course does not happen. The chip is still uncomfortably hot. We shall consider its magic smoke to have escaped many years ago. Taking U3 out and placing back U4 and U5, I hope to see at least some chips still alive. U4 is cold, but U5 is warm. Which is unsurprising given the 5V rail excursions towards 7-8V. Still no signs of life on the front panel. U4 does not seem to be doing much either- no scanning of the display digits, or segments. Removing U5, and grounding one of the segments while enabling the digit, I can finally light something on the front.
Another nice surprise was awaiting me when I touched a couple of traces with my finger and got a jolt. Those two traces from the fuse are live and open and are going to the power switch in front.
Conclusion: not such a good deal after all.
It seems that there is not much hope for getting this DMM to work again. Individual chips are custom made for Fluke, and being 1974 vintage are a bit expensive to find. U3 sells for at least $20s so does U4. U5 is the only standard part that can be easily replaced. I suppose the other approach would be to recreate the original functionality using a PSOC or a microcontroller, while leaving the rest of the input analog chain intact. That may be a fun project, but will have to wait till another day. While at It, I’d probably throw at least some type of a regulator on the 5V rail!
P.S. There is a discussion of this post over at EEVblog Forum