I'd try replacing that 18u cap, too.
Looks like a cheaper (cost reduction) alternative to the original 47u/50.
It might have developed reduced charging quality over time, series resistance increasing with increasing voltage. Which would explain the linearity problem.

I don't think the actual capacitance is very critical, as it probably will work fine within the range of ~15 to ~63 uF.
If unsure, just take a fresh bipolar cap 33u/50V, or two normal ones in series, connected in the middle with both positive terminals (the bipolar electrolytics are just this).

Every 5151 I've seen had the 18uf part in the H drive path. Not sure if 47uf was from a different chassis revision or what.

The easiest way to replace it I found was to stack 2 film caps in parallel

Thanks! I'll grab some correctly rated NP caps and film caps and report back.

Currently considering two of these in parallel for 20uF total. Is anyone able to confirm whether these should do the trick?
https://au.element14.com/illinois-capacitor/1 … dial/dp/3389023

These are perfect for this job, far better than a np-electrolytic.

Arkanix wrote on 2021-11-12, 09:33:

Currently considering two of these in parallel for 20uF total. Is anyone able to confirm whether these should do the trick?
https://au.element14.com/illinois-capacitor/1 … dial/dp/3389023

2 of those should work great

these are what I used:
https://www.digikey.com/en/products/detail/pa … E1106JF/2567774

Awesome, Thanks!
Will report back when they come in.

It appears that our guess was incorrect. I have finally gotten the caps from my supplier, soldered two 10uF film caps in parallel and the picture is virtually unchanged from the 18uF bipolar electrolytic cap. Does anyone else have a clue what could be going on with the H-Linearity?

maybe try reducing the width of the raster a little and adjust the coil again?

If @maxtherabbit's suggestion does not help, maybe look at the linearity coil closely.
In the very old TVs their limiter, limiting how deep you can turn the ferrite screw into, were made of wax, like the windings themselves.
Sometimes, when a circuit was borderline, it was necessary to screw the ferrite a bit deeper, cutting into the wax (and falling through if you drilled too deep, which then needs some annoying work to get it and drill in again properly so it just does not fall through).
If it is made of plastics, it will likely break, which you will not want.

Any of those adjustment rings on the tube that have come loose? Forgot what they are called... sit around the neck with a tab on them, have a little magnet on them or something to dial the tube in. Usually have some goop/cement to hold them in place, but that can crack up and they submit to gravity.

I think I might pack it up and call it a loss for now. Tried adjusting the adjustment rings and they didn't help with the linearity problem. I then tried adjusting H-WIDTH using L503 and I think the ferrite core cracked while attempting to adjust it so it doesn't want to budge and the cursor is now half off the left side of the screen.

I was thinking in general terms, linearity in the analog realm is accomplished by using the straight bit of an S curve, or between +/- 50% of a sine wave. This means that components have to be used in the middle of their range. Ergo we might expect non-linearity when components are aging out, have got overheated, or otherwise damaged. Yeah, thanks captain obvious, but, failing caps would tend to shorten a time period rather than distort it, so I tend to start thinking that transistor 21 22 or 23 is maybe drifted in spec and we're looking at the top of it's current S curve, when we should have been looking at the middle of it's designed one. However, we could be looking at the top of it's designed one if it's bias has shifted. It may be running kind of hot, this may be due to being overdriven, too much current, or failing. Hitting them one at a time with freezer spray (or inverted air duster) might show a change. Caps downstream of them may be guilty, not for timing purposes, but for going too conductive allowing too much drive or bias. But anyway, can build a straw house of maybes.

Might be worth a try when I get courageous enough to open it up again. Put it back together and lubed up/set head stop on an IBM Tandon drive for my 5150 instead, helped me feel a little better after messing up the ferrite in the H-Width coil

let this serve as a PSA not to go large gorilla banana mode on these old adjustable inductors, if you can't break the glue without excessive force maybe just accept the image being slightly distorted

Maybe just get a worn-out 5151 and take its inductor.
If it also has this nail polish, you can try softening/dissolving it.
But first try at another place whether the material can resist ethanol, IPA or even acetone.
Using ethanol is safer, but takes some time to soften the glue.

maxtherabbit wrote on 2021-11-25, 14:06:

let this serve as a PSA not to go large gorilla banana mode on these old adjustable inductors, if you can't break the glue without excessive force maybe just accept the image being slightly distorted

No visible glue holding the ferrite in place, it was just super fragile and got snagged after the first 1/4 of a turn.

EDIT: wasn't wrenching on it super hard either, I have a feeling that the tool I was using may be too shallow and wasn't engaging enough surface area.

Or it was of metal. Only plastic tools should be used for adjusting coils.

retardware wrote on 2021-11-27, 11:37:

Or it was of metal. Only plastic tools should be used for adjusting coils.

Yeah this ended up being the issue, I should have done my homework.

Now attempting to locate a replacement coil and will get some proper Delrin adjustment tools.