Semi-gutless wonders can sometimes be helped, but only sometimes. 
This had saturn primary caps and fuhkyoos everywhere else. Automatic power supply roaster (~9V for fan) didn't help. The system was getting random errors and CDROM no longer read. Lotsa ripple getting into the servos and signal amps, ya know. Also looked like bad RAM, more on that later.
The line rectifier was rated for 2A, despite having the appearance of a KBL404 or a variant. NTC thermistor also only good for 2A, ceramic caps instead of UR Y caps, a missing Y cap from the negative side of the rectifier to EGC/GND (EMI), tiny wire on the common mode choke, #20AL for output leads (ampacity of #22CU), stupid heatsinks, too-small rectifier for +5 and diodes-on-a-bracket for +12. Anything else is blatantly obvious
+12 had no PI filter, so I added one. I also did not forget to move the feedback point to after the coil! Now, not having enough room to terminate all the +12V wires, I came out with a single #14 and spliced all the 12V wires to that. All the wires to the M/B connectors were replaced with #18CU, except the PG signal. The two main output rectifiers were replaced with SBL3040's. Even with the higher operating temps caused by their small heatsink, they are loafing along. The primary switchers never got excessively hot, thanks to the two new, good caps in their base bias ckt. The fan was hotwired. Here are the voltages after the improvements:
+5.15V/15.3A
+12.5V/6A
-5.3V/0.25A
-12.0/0.65A
In addition, two hard drives were connected along with the loads. Lookin' good!
Onward to the 'puter. The board was recapped, aw 'Jee.' Pure junk, those things. I added a few that were missing from 'cost cutting.' We're on the way, now... The front bezel's only holes were 5/32", starving the fan in the PS for makeup air. I opened them up, as they should've been.
Now since the RAM in these AT dinosaurs is powered straight from the +5V supply, all the trash from before the PS was fixed went right to the memory. Not so anymore. With 128MB from both SIMMs and a DIMM, memtest held for hours. The CDROM works again, and I just serviced it. Vacuumed out dust, fresh lithium grease, 2-26 on connectors/switches, cleaned lens, etc. The system was almost perfect at this point.
A 10GB HDD on the old BIOS required the EZDrive overlay. As usual, "32-Bit Transfer" was disabled. I just put my EZDrive floppy in and fixed that. Defrags at least twice as fast now. Gotta love lazy automatic installs, that 32 bit transfer was never enabled after all this time.
Now nearing the finish... I REM'd out some real-mode drivers in config.sys and autoexec.bat, as they are simply not needed in win98. Made sure all was well with the AWE64 and finished up.
Not wiping this one out as there's still data on it.
Hope you all enjoyed,
-Paul

This had saturn primary caps and fuhkyoos everywhere else. Automatic power supply roaster (~9V for fan) didn't help. The system was getting random errors and CDROM no longer read. Lotsa ripple getting into the servos and signal amps, ya know. Also looked like bad RAM, more on that later.
The line rectifier was rated for 2A, despite having the appearance of a KBL404 or a variant. NTC thermistor also only good for 2A, ceramic caps instead of UR Y caps, a missing Y cap from the negative side of the rectifier to EGC/GND (EMI), tiny wire on the common mode choke, #20AL for output leads (ampacity of #22CU), stupid heatsinks, too-small rectifier for +5 and diodes-on-a-bracket for +12. Anything else is blatantly obvious

+12 had no PI filter, so I added one. I also did not forget to move the feedback point to after the coil! Now, not having enough room to terminate all the +12V wires, I came out with a single #14 and spliced all the 12V wires to that. All the wires to the M/B connectors were replaced with #18CU, except the PG signal. The two main output rectifiers were replaced with SBL3040's. Even with the higher operating temps caused by their small heatsink, they are loafing along. The primary switchers never got excessively hot, thanks to the two new, good caps in their base bias ckt. The fan was hotwired. Here are the voltages after the improvements:
+5.15V/15.3A
+12.5V/6A
-5.3V/0.25A
-12.0/0.65A
In addition, two hard drives were connected along with the loads. Lookin' good!

Onward to the 'puter. The board was recapped, aw 'Jee.' Pure junk, those things. I added a few that were missing from 'cost cutting.' We're on the way, now... The front bezel's only holes were 5/32", starving the fan in the PS for makeup air. I opened them up, as they should've been.
Now since the RAM in these AT dinosaurs is powered straight from the +5V supply, all the trash from before the PS was fixed went right to the memory. Not so anymore. With 128MB from both SIMMs and a DIMM, memtest held for hours. The CDROM works again, and I just serviced it. Vacuumed out dust, fresh lithium grease, 2-26 on connectors/switches, cleaned lens, etc. The system was almost perfect at this point.
A 10GB HDD on the old BIOS required the EZDrive overlay. As usual, "32-Bit Transfer" was disabled. I just put my EZDrive floppy in and fixed that. Defrags at least twice as fast now. Gotta love lazy automatic installs, that 32 bit transfer was never enabled after all this time.
Now nearing the finish... I REM'd out some real-mode drivers in config.sys and autoexec.bat, as they are simply not needed in win98. Made sure all was well with the AWE64 and finished up.
Not wiping this one out as there's still data on it.
Hope you all enjoyed,
-Paul
Comment