Nearly there with IBM 029 KeyPunch

The theory said to change the resonant frequency of my keypunch's ferro-resonant power regulator from 60Hz to 50Hz I should up the 15uF capacitor to 21.6 uF. Well I tried 21.5, and the voltage went up by ONE volt. Hmm, scratch head, at least it went in the right direction. Search around for AC capacitors and find one in a defunct large electric lawnmower. Its 30uF, I think about replacing the 15 with it but if an extra 6.5uF only gained one volt, and I need 6 then it seemed reasonable to combine it and try 45 uF. That gives me 46volts, only 2 volts short. I throw in the 6.5uF too and I get 47 volts. I try the punch and it works much better. Not 100%, but that could well be down to other problems. For one thing, the alpha shift key does not latch, I need to hold it down. There are keys for numeric and alpha. If it was not meant to latch, why would they have a numeric key? Its so many years since I used an 029 (at university as my first job was at a paper tape using establishment) that I can't remember for sure. Anyway, thats sounds like an easy problem to find in relay logic, though these wire relays are difficult to see if they are active or not. Maybe I'll add LEDs (and resistors of course). Another problem is 1 to 7 print fine, but not 8 and 9. The alphabet print only their numeric component, 1 to 7 and again not 8 or 9. I tried using it in interpret mode, it fired like a machine gun but the card did not move, but the time I worked out how to stop it there was the smell of overheating solenoids. I don't think I damaged anything, just a bit hot. Thanks for all your help.