Well I think I have an answer. Drlava was right about a sagging PSU, however the performance is hard to explain. When measured under virtually no load (ie. a multimeter) the pot adjusts the output from 21.5V. to 26 V., but under even under the minimal load of 5 P3s on standby (630ma) the output drops to 21.5V. and is no longer adjustable. Even at 4.0 A. the output hardly changes down to 21.1 V., but that's enough to start dropping out P3s and their diode loops. I have a custom built cabinet for all the power supplies and would like to avoid setting some lab supply on top of the cabinet just for the 445s. The available room for a replacement is 210mm x 115mm x 60mm. It needs to operate on 110V. AC, produce a minimum of 10 A. and isn't AFU! Any suggestions for a source would be greatly appreciated.