Mettler PK 4800

[ConnorGames]

Recently, the thought of getting an 80,000 count something off of ebay for $80 convinced me to finally get a balance. I have more than occasionally wanted to know how much boards or other small items of mine weigh, and this scale, which from the ebay listing seemed to be "semi-working", peaked my interest. The balance operates and seems reasonably accurate (within a few % of a $10 "fish scale" I have), although it needs something placed on top before it will zero, presumably due to the missing weigh plate (I will buy or fabricate one, possibly along with some calibration weights, after I get the balance operating to my satisfaction). Unfortunately, the zero drifts significantly over time (faster right after start-up, but still .01g every few minutes after running for an hour). Linearity is seemingly quite good, with my phone weighing 178.65g before adding and tearing a 1.1kg textbook, and 178.64g or so afterwards, although the least significant digit varies some due to the aforementioned drift, and also repeatability that seems not to be great (+/- about 3 least significant digits, although air currents certainly affect this). Pressing the calibration button currently only results in "Cal Err" on the display, although I previously was able to make it display -CAL-, but was never able to get it to proceed beyond that. I don't know what caused the change. Either way, the balance seems to need some work. Judging by the 1982 datecodes on the chips, testing and replacing electrolytic caps and checking the reference for stability will probably be my first steps. The reference is an LM299, so checking stability will probably require me to locate a nice DMM on campus here at college. The ADC appears to be implemented with discrete logic chips, and the computer is based around a Z-80. Calibration is stored in a battery-backed SRAM. I intend to change the battery as soon as practical (with a diode and external power source so as not to loose SRAM contents), and I may also attempt to dump whatever contents are read at startup using a logic analyzer. Interestingly, the balance also seems to have some kind of electro-mechanical servo -- humming from the transformer increases with the mass on the balance, and an internal component can be seen to move and then return to an equilibrium position when a load is applied.

I don't have time for pictures tonight, but I should post them soon. Also, if anyone has a good diagnosis of what I should check based on the above symptoms or a better way to dump the cal RAM, let me know!

[texaspyro]

Mettler scales are famous for faulty power supply caps.

[ConnorGames]

The forum just ate my previous attempt at this post, so I'll keep this brief. I just dissembled the scale and tested the caps. They all appear good. The SRAM battery is also dead, maybe it loaded factory cal from its PROM? I also got some pictures. Next step will be to reassemble and measure power supplies, then try to use my hot air station at 100 Celsius to provoke the drift to see if I can localize the problematic area. Otherwise, I'll start looking at thing with a 6+ digit DMM. Does anyone have a service manual for a similar scale or a theory of operation? I want to know how the DeltaRange architecture works (does it generate a reference voltage for zero? Should I check for drift there?), and also to understand the aforementioned electromechanical servo. The last photo included here is the back side of a board with some kind of optointerrupter on it, I did not remove the board out of rear of misaligning it (and judging by the fine-pitch threads that appear designed to adjust its position, I don't want to). If I can find the source of the drift, I'll recap the power supply and get a new SRAM battery.

Once again, any ideas?
The forum keeps refusing to let me post pictures and going to a "new post" page if I try to do that, so I'll try to attach them to a separate post

[ConnorGames]

Photos are too big to upload, so here is a link: https://photos.app.goo.gl/vhy1VmdM15WiRJZZ9