I bought a 4957A that was broken. One cap on the power supply board at the feedback control chip and that's all it was. I never did find the service manual for it.
Yeah, this one needed 5 caps on the PSU replaced (all 4 poly caps, and one electro).
Unfortunately, this unit also had some pretty catastrophic battery leakage on the "OPT" board, which was more or less destroyed, and caused some damage to the backplane. I discarded the OPT board, and cleaned up the backplane as best I could, but it still wouldn't work initially. Then on a closer inspection, I noticed that the contacts were missing for a few pins on the PROC and MEM boards (probably corroded through, and fell out at some stage). I circumvented the missing connections with some wire jumpers, and now it sorta works, but it's still a bit sad.
It won't boot "normally". You have to enable one of the diagnostic modes on the CPU card's DIP switches (doesn't matter which) then set the DIP switches back to "normal" and hit reset. I'm not sure what the cause of this is, all of the diagnostic tests show 0 errors. Also the tape drive is screwed, the motor has completely seized up. I don't think it's useful as a protocol analyser any more, but I'm now thinking about how I could create something new and cool out of it, now that I know the screen and keyboard work nicely.
So I probed around with my oscilloscope, and found that the while the Vsync is a common 60Hz, the Hsync is a more exotic 25kHz. Apart from the Hsync figure, it seems to act as a normalish TTL monochrome monitor. I wonder if there is some way to drive this display with a Raspberry Pi or similar... Hmm.
Here are some pictures, in case anyone's interested:
![](http://i.imgur.com/nLovYZxl.jpg)
![](http://i.imgur.com/0z9uCSGl.jpg)
![](http://i.imgur.com/loFX4k6l.jpg)
![](http://i.imgur.com/8Obcww9l.jpg)
![](http://i.imgur.com/2uJCXwVl.jpg)