I think my most challenging diagnostic problem was a channel hanging the system. I used a 360 I/O channel test box that connected to the channel and when a error or hang condition would occur it had recorded the last several instructions and the channel responses to those instruction. IBM sent a factory engineer to the site and it was determined that a test I/O instruction to a 3271 was generating a device end status when it should not generate any new status. The fix was a EC to the wiring of the 3271. My most unusual repair was not in the maintenance manual. One morning about 2 am I had a bad 12v power supply in a disk controller unit. I asked the customer to bring me a 12volt car battery. I rewired the 2841 to the battery sitting in the floor beside it. The 2841 powered up and ran for about 15 minutes before it started getting disk read errors. I asked the customer to bring me a battery charger which I connected to the battery. The computer ran two days with the battery supplying the 12v until I could get a power supply. Sometimes experience and good luck are the troubleshooters best tools. In 1988 I was in China helping the Chinese with their computers and received a call from The Bank Of China as their system was down and they had closed the bank because of it. My assistant Chen Wei and I walked into the computer room and I looked at the console. The console showed a MPX channel error. I looked around the computer room and saw several sections of the raised floor were pulled up which indicated the Chinese service company had been moving boxes around. Due to past experience I knew the buss and tag cables could be switched and would cause a MPX channel error. The first box I checked had the cables switched and Chen Wei switched them correctly. Everyone in attendance thought I was a true computer expert but it was experience and good luck.