Found the docs, and indeed I was right the first time around. I just didn't have the nyloc far enough in, but then there is not a lot of spare room. I've adjusted it by taking the idler off and aiming for the teeth to align with the hole for the filament, but I'm concerned that the nyloc just won't have enough grip that far out. I'll try it this way, but if it starts unwinding itself, I will instead place a regular nut at the appropriate space, glued on with epoxy.
Curiously: I homed all axes while heating was ongoing, and there were no temperature readings in the meanwhile. Such single-threaded. Very modal. Much block. Wow.
And I was right. Just a little bit of extrusion and the bolt holds still while the gear merrily unwinds. The old bolt was born with the end nut, so there was no way it could come off. To the epoxy!
Epoxy takes about 24 hours to harden. Heating it reduces the hardening time and strengthens the result, according to the packaging, so I left it on the heated bed at 40 for the afternoon and evening.
In the morning, I tried to fit it, but due to some epoxy seeping out, the nut didn't go as far in as it should. See how the hobbing is off from the filament hole below it.
Filing down the epoxy (carefully disposing of the dust) got me back to a good, solid bolt with an appropriate spacing:
The test extrusions are very different from with the previous bolt. Where before it would build up a string that would then fall over, now it folds down like when laying down a rope. Oddly satisfying.
I did a 10cm extrusion with a mark, and it fell 7.6 mm short. I guess I should re-adjust my e_steps_per_mm multiplier. While checking for whether this should go on top of my previous multiplier, I ran across Triffid Hunter's calibration guide, which actually gives a formula - and it's different from what I had. My Wade's is a 9/47 according to the receipt, while the formula in the Marlin code uses 11/39 - that's quite a difference, no wonder I had a big multiplier. I expect the hob diameter is the same, roughly. (39/11) / (47/9) = 0.6789.
Note to self: The correct Marlin firmware is the one in hardware/Marlin, not one of the 2+ other Marlin instances I have floating around.
The 100mm extrusion test was off by 17mm, probably attributable to variances in effective hob diameter. New version:
#define E_STEPS_PER_MM ((3200 * 47.0)/(9.0 * 6.75 * 3.142) * (100 / (100-17)))
Writing it as
#define E_STEPS_PER_MM ((3200 * 47.0)/(9.0 * 6.75 * 3.142) * (100.0 / (100.0-17.0)))
should have more of an effect. And it does, but it's still 5mm off. Two test cubes show increasing lack of extrusion, and another extrusion test misses the mark by 34mm! If the epoxied nut is slowly coming off, I'm going to be annoyed. There's no sound of skipping or anything.
Triffid Hunter's guide suggests tightening the extruder idlers as much as possible. I won't quite go to the "my fingers hurt from tightening" stage, but close. Even with that, the amount of extrusion is way off. The shape of the test extrusion has also changed, from the "coiled rope" back to the "toppling wire". Either the gear is coming off or the nozzle is getting clogged again. The gear looks good, and the way this is slowing increasing speaks more to clogging. Time to go atomic? Let's give it a shot.
Now there's a strange thing. In preparation for the Atomic, I untightened the extruder idler screws. Just then, at 10C above the goal of 90, the hotend temperature increased by about 5C before falling some more. Coincidence? Maybe. There's some logic that prevents overshooting the temperature, but I haven't seen it do this before.
The atomic didn't work, only a short piece came out. I'm mystified.
I'm concluding that the nozzle is getting clogged again. I'm concerned that it's just happening, without any obvious reason.
Also, after unscrewing the hotend, it doesn't come out easily. Once out, I ran wire through it at high and low temperatures, but I don't feel confident that that's enough. Taking off the extruder idle to check on the alignment (which is still good) I see evidence of severe stripping:
Now is it stripping because it got clogged and had too much back-pressure, or did it get clogged because it was stripping and dust came down into the critical area? Or did some of the previous dust start the clogging, which then turned into stripping and thus more clogging?