In line with my Mettler PC440 repair the other day I also finally came around to look at my Sartorius 1003 scale. This one is quite simple, it uses a strain gauge.
No calibration trim done yet, that’s why it is off slightly.
Power Supply
The power supply (12V/100mA) is simply a small transformer with a thermal fuse, a couple diodes, resistors a capacitor and a LM317.
The capacitor was way out of spec from what was probably decades of being plugged in. I replaced it with one from Frolyt I had in the component box, the latter also having a higher temperature rating (105C) than the old capacitor (85C).
I also added a small 125mA/T fuse in line with the thermal fuse to be more easily tripped on transformer winding failure and also added a Y2 filter capacitor on the primary side. I glued the fuse in with some hot glue, to not wiggle around.
Polarity protection
The scale uses the evil power plug layout of inner as 0V and outer as 12V. I read on a German forum that people wrecked their scales when the power supply failed and they used another supply unbeknownst to them about the plug polarity.
So I figured I add some polarity protection, just to be safe. I put a small Schottky diode in line with the +12V plug input. For good measure I also added a 14V MOV to act as a light surge protector.
Stability Problem
When turning the scale on the scale sometimes gets stuck in the “L” state (weight below pan present weight). Quite often it would then spontaneously have a bias of ~40mg and only go to 260mg before going off on “H”. With plenty of troubleshooting I identified that putting a small capacitance over one of the trim contacts of the OP07 operational amplifier solves this issue.
I desoldered the original PMI OP07DP, put in a precision socket and installed a newer TI OP07CP to check whether the problem persists. It did. My guess this is related to the (proprietary) ADC and its getting stuck somehow (SCR latch up type problem?) and it needs the signal to raise at a slightly less speed initially.
So I just put a 22n capacitor from the trim contact to GND and done, issue solved.
I couldn’t even find a datasheet for the microcontroller they used. Its a TMS1000 series but for ultra-low power operations and with builtin LC display driver (TMS2220). Only found it mentioned in some old IC catalogues but no pinout or specs.