PC-jr PSU (again!)

I dug out the PC-jr Techref (in order to look up details on the joystick), and I also looked at the PSU diagrams (page B-23 in my edition). We had a thread on this about a month ago, mainly about the external transformer unit which I don't have. Now, the input connector is a 3-pin thing. The centre pin is connected to shield ground, and then via L3 and L4 on the system board to logic ground. It's thus a DC connection to the system ground. The outside 2 pins go through a filter to the AC terminals of a bridge rectifier (CR7 - CR10) and the output of this goes (via the power switch) to a 1500uF 25V capacitor (C13). The -ve side of this capacitor is also connected to system ground. Thus, IMHO it makes no sense at all for the transformer to be centre-tapped with the tap going to pin 2 on the connector. If it was, the winding would be paritially shorted out by the diodes in the bridge rectifier, which would cause a lot of damage. Also, since C13 is only rated at 25V, the maximum input rms voltage is 25/sqrt(2) volts, which is a lot less than 34V. I therefore still believe that it's a plain 17V transformer between the 2 outer pins on the connector. Incidentally, the +12V output comes from a standard linear regulator, the +5V output from a switching regulator (Z2 is the control IC, Q2 the chopper, and L1 the switching inductor). The -12V output is supplied by a secondary winding on L1, which is then rectifier and smoothed. It's always possible that IBM used several versions of the PSU card, but the schematic in my Techref seems to agree with the one in my machine. -tony