My MU-140 was giving me fits this weekend. I would calibrate it and watch it as it heated up. The gm calibration would consistently drift low. Low enough to where the 5 KHz signal could not be turned high enough to read "100" with the 1600 Ohms in place. I pinned it down to CR12, one of those pretty green tablet diodes. I replaced it with a 1N4007, and went about calibrating, waiting, and re-checking to verify the fix. Well, son of a gun! the problem was lessened, but, still there! Hmmm. I put on my thinking cap, grabbed a cold 807, and took a few measurements. The voltage across the new CR12 was just over .4Vdc. It's a silicon diode, shouldn't be on yet, right? Well, the meter is 1mA FS, so, WRONG! It doesnt take much leakage current to start pulling the meter reading low.
My DVM puts 1mA through a diode in diode-check mode (I know because I helped design and test it when I worked at Tandy). So, I measured the forward voltage drop of the 1N4007 at 1mA and, lo and behold, it was ~.5V at room temperature. The new diode was effectively leaking too. To solve the problem, I put a schottky diode in series with CR12. This slightly raised the threshold voltage of the effective diode in CR12's position. Wow, what a change! The required setting of the 5 KHz adjustment (R24) is much lower and more stable - and gm readings on this old MU-140 are now rock-solid. I deliberately let it cool down and then went about calibration. Once the 12AU7 warmed up, the zero, emission, and gm calibration readings were all carefully re-adjusted. I then let it sit and warm up with a 6F6 running emission test for a half hour . Once it was fairly good and toasty, I rechecked the calibration - right on the money!
I know that there will be those who feel that it is just as good to remove CR12 and simply add two back-to-back diodes across the meter itself. For me, this solution keeps the designed-in meter protection (actual current limiting) in place and I am more than comfortable with the amount of protection it will afford. I hope that some of you find this useful.
Anybody want to buy some green diodes?
Edit - typo/ d12 is actually CR12 and R34 is actually R24.