Somewhat tangential to electronics but... I wanted to build a FMES seismometer and the instructions called for a 0.05M K2HPO4 solution as part of the electrolyte, implying an accurate scale to weigh out a small amount of the dry salt to add to the solution.
Amazon has 50g x .001 (milligram) scales around $17 and it turns out, no surprise, that the last digit has no real meaning, although the firmware tries to make it appear realistic when near zero or a standard test weight value. So I stepped up to a $43 ebay "200x0.001g Analytical Balance Scale" with a possibly-effective-looking boxy draft shield around the pan. At first this appeared to work, BUT the reading would randomly jump from 0.000 g up to several hundred mg, then return to zero, even with the draft shield in place and the top-load hole covered. It would also consistently sense the presence of my hand as a ~ 100mg force even just hovering 5 cm above the shield without touching anything. Some tests showed it was not stray light, not vibration and (maybe) not RF. I think it is either electrostatic force or 60 Hz capacitive pickup through the plastic case. I simply wrapped aluminum foil around three sides and the top except for the loading hole, and that completely fixed the problem. FWIW I then tested it for a while sequencing through 100 mg, 0 mg and 110 mg test weights and so far this scale is repeatable at the +/- 2 mg level although its firmware also "cheats" within 10 mg of zero to hide the zero drift.