When I left after the last workshop a bit would be dropped when shifting the accumulator in the moving lights demo. I was able to move failure to a different AC bit by swapping AC cards. When I got to Infoage for this workshop Evan said that he had successfully ran the moving lights program for a demo sometime previously. I spend some time trying to get the problem to reproduce unsuccessfully. I tried wiggling the cards and using the margin check voltages without causing the failure. Moving the cards enough that they move in the backplane connector would cause faults but motion less than that didn't. I suspect it will be back sometime. At a previous workshop I replaced a number of diodes in the memory cards but didn't' have time to do the memory margin check procedure. This time the sense amp margin voltages that caused failure were 4.9V and 17.3V. The manual says the difference should be > 12V which it is. The slice failure voltages are 7.04 and 8.1V. No minimum range was specified in the maintenance procedures. I left the slice voltage set to the middle of the range at 7.6V. Power supply voltages were 10.7 and -14.6. I did notice that I was still getting double increments on front panel load/store operations occasionally. The machine has a debounce circuit that did not debounce sufficiently. I assume the switches are noisier than when new. I previously put a 47uf capacitor on it to increase the debounce time. I increased that to 100uf. That may be getting a little too large since it seemed to ignore really quick switch presses. People using the machine should let me know if you see the double increment or other funny behavior still and if you find it ignores switch actions for you. Good check is after or occasionally during check the address matches the expected address for next location you are loading. When I tried to local Focal programs last time I got ?11.35 error which is input buffer overflow. Focal has a design issue. When loading from ASR33 paper tape it prints what it reads in without any flow control so if printing is slower than reading it will overrun when the small buffer is filled. I compared the RX bit timing from the teletype to the TX clock and the TX clock was slower. I adjusted the R450 clock to prevent the overrun. Teletype seemed to be happy with the faster clock. I was able to load lunar lander without errors. My attempt to land ended with a crater. Known issues are 1) The machine will start running sometimes at power on when the auto restart switch is off. It seems to start running before the voltages are stable so it corrupts memory when it runs. Didn't look at this. Easy workaround is to leave the single step switch on at power on. 3) The W300 delay line cards still need to be fixed. I think the machine is usable for demos. I suspect sometimes the demos won't work but will next time. We need to keep track of what issues we see and is anything correlated with when it acts up such as temperature to help in further fixing. I left it with the moving AC light demo, the punch block letter text on paper tape demo, and FOCAL lunar lander demo. I'll look for what else might be of interest. Getting the plotter working and making an interface on it will hopefully be a winter task for me. Having the plotter making some interesting plots should be a good background demo.