REPAIRING THE B128 POWER SUPPLY
      Latest updates, additions and/or corrections: 1-17-2021

     I've only worked on two B128 power supplies so I don't have much 
experience with them and there are no schematics I know of. It is a 
switch mode type and, as old as these are, I expected to find some 
electrolytic caps dried up and failing. Not so. They all checked OK. 
     #1 PS was intermittent... it would come on and go off at random, 
sometimes "flickering" between states with an audible buzz or clicks. 
The computer didn't work at all. No burned parts and fuse was OK.
#2 was dead... no output and input fuse not blown. 
     Note that many old switchers will not run properly or will self
destruct without some kind of a load on them; at least 10 percent of
a full load is recommended. Early PC supplies were like that. For this 
one, I used an old automobile tail-light bulb as an artifical load. I 
put the heavier filament on the +5VDC (about a 2 Amp load), the lighter
filament (about 500mA) on the +12 and a small grain-of-wheat lamp on
the -12 just to verify it's working. It's important to meter all outputs
to ensure the PS regulates properly after repairs are done.

     #1. I suspected the main switcher transistor, a Motorola TO-220 NPN 
type with nothing but a "house" number on the case: 3821-2. That number 
didn't cross to anything in my data books or online sources. It helped 
that the PS was at least partially working so I could measure the 
waveform on the collector. It would peak at over 900 volts so I knew 
that standard switcher replacement devices wouldn't hold up. I found 
a generic replacement in a different case style (larger) that was 
normally used as a horizontal deflection output transistor or "HOT" in 
a tube type color TV set. Its generic number is BU508A. It's an 
equivalent to the Sylvania ECG 2300 (or Philips TCG 2300). 
     Because it is physically larger, the transistor would not mount
on the old heat sink. I drilled a hole just above the original one and 
mounted the new transistor with a mica washer (to isolate the transistor 
collector/case from the sink) and added heat sink grease. The transistors 
leads didn't easily fit the holes, so I cut them off near the body of the 
transistor and wired connections from the stubs to the board. The new 
transistor has a higher voltage and current rating and so should work well 
in this application. 

     #2. There were no burned parts and the fuse was good so I knew this 
PS just failed to start up. The problem was traced to a shorted Q1 
transistor, a 2N5190. That crossed to an ECG184 which I had in stock. That 
supply worked fine after repair.
     I decided to reverse-engineer the swicher side of this PS and drew a
schematic to aid anyone who must repair one. It's usually the input side 
of a switching supply that fails anyway. I listed the component ID's that 
were shown on the board but many of them cover the board markings... a 
stupid thing for the factory to do. 
     Bad filter caps or regulators on the secondary side can be easily 
checked if the supply works but only one output source is low or down. 
The secondary side of this supply is what I would describe as an 
e-plumbers nightmare. I didn't even try to make a schematic of that side.
I felt myself growing older just looking at it. 

The PS specs are as follows: +5VDC @ 5.5A, +12VDC @ 0.4A, -12VDC @ 0.3A
Pin 1 of the power connector is shown as "50/60 Hz" but without a 
schematic, I don't know what its source is inside the PS. 

Ray Carlsen