The battery in my 4957A was completely empty and had first signs of leakage at some pins.
So it had to be replaced immediately. Unfortunately this battery stack is not available any more. So I had to assemble one from four single 1.2V NiCd cells (150mAh) as Marc did in his video.
My cheap Amazon-based spot welder did a good job after I figured out how this is functioning.
I reused the pins from the old battery and wrapped the pack in shrink tubing. I also glued them together using some epoxy.
Perfect fit and full 4.8V for the buffered RAM.
This is by the way a picture of the original cells (for reference):
Next stop: cleaning and re-assembling
Not too bad after cleaning carefully!
The backplane rebuild with a new line filter (it really fits – you can buy the original part from Schaffner 30 years later – FN380-4-22). I also added a new fan.
Last preemptive maintenance was to replace all caps and RIFAs – just two were bad:
Switch it on
I tested the power supply on the bench and got very disappoint results. It did not start and made dangerous clicking noises.
After reading the service manual of the 4951C carefully, I found the hint that this supply needs a load. HP also explains where it needs to be applied…and then it worked.
BTW: The 4957A power supply PCB is an exact replica of the on board power supply of the 4951C – at least the one I found (HP branded).
Putting everything together….and three beeps and….a picture.
Unfortunately the image is still way too bright, which puts us back into error searching mode.
How does the brightness work
The signal is called intensity and puts a negative voltage of about -45V to one of the gates of the CRT. This deflects the electrons so that they do not reach the screen.
I firstly struggled to make some sense out of the CRT deflection circuit of the 4957A, but then I recognised that the flyback transformer is the same as in the 4952A – and they copied most of the schematic.
But of course not completely, so I had to re-engineer this part of the CRT driver (I used 4951A part numbers where possible).
The upper part is the intensity part, the lower part generates the focus voltage.
Focus Voltage & Acceleration Voltage
The focus voltage is around +180V. HP removed the focus trimmer and replaced it with a simple voltage divider (R4/R3) and some clamping diodes.
The voltage is borrowed from the acceleration voltage generated by CR11 (800V at the cathode) which gives us about 350V at C48.
Intensity
The raw intensity voltage is created by another flyback winding which gives us about -50V at the anode of CR12. This is stabilised by 42V Z-diode and buffered by C2. This -42V is then used feed the intensity trimmer.
Error detection
Measuring the voltages revealed that CR12 was working correctly, but after R34 it dropped to 0V. This means “full brightness” and explains the screen picture.
First guess was either the Z-diode or the tantalum cap buffering the stabilised voltage. Quickly measuring the two diodes detected no obvious faults.
Temporarily I replaced the tantalum with a 4.7uF cap from my spare hub…and…success!
What’s up next?
I ordered the replacement tantalum from Reichelt together with replacements for the broken Molex 396 connectors you can see in some pictures – they were all brittle and broke on the first disconnect attempt.
I still have to clean the disk drive and then put everything together for a first real function test.
Attention!
Working with the open switch mode power supply and of course the CRT with 800V DC at some points of the PCB is not a joke. It is seriously dangerous. Only perform these repairs if you are sure what you are doing and are equipped professionally (never ever without isolating transformer!).
