I have made some nice new rollers.
The heatshrink test made the plain rollers about twice as grippy, but still not good enough. Also, the heatshrink skin started to 'ruck' up so I would have to use a glue lined shrink. Move on to the ribbed rollers.
I tried them last night, and they grip the filament really well. The problem was I made them in aluminium (quick, easy, available barstock, soft so prototype only - brass will be the material of choice). The 4mm axle holes are a bit rough - simple drillings - and don't run well on the stainless axles so I ordered some nice needle roller bearings from RS that arrived today. I nipped over to the workshop, opened the roller bores to 8mm and slipped in the bearings.
The result? wonderful grip, smooth rolling action. Questionable feed rate under load. Hmm.
Encouraged by the grip etc, I assembled the 'nut' assembly into the 'fusion reactor' housing for the first time to give it a spin. This revealed one or two minor problems. Minor problem no.1: the radial springs crash into the frame where I added ribs to carry the alternate bearing system so no revolving can occur until I cut them off. Minor problem no.2 : the frame squashes the nut which stops the rollers rolling.
Solution to problem 2 will involve A: machining the rollers to the correct length (I had left them slightly long because the way I parted them off the barstock left a nice extra grippy rib. Now I know that 3 grip so well, the original 2 will be ok). B: possibly getting the correct flanged ball bearings to take radial and axial thrust; or C: making up some spacer tubes for the teeny M2.5 screws that sort of hold it all together.
Option A and C is the best/lowest cost but B is the costliest but correct design. Anyway i will have to wait until next week for B's bearings do A/C it is !
On a related front - last night I also pulled apart a metal can stepper motor from an old printer paper feed - I will mod it for hollow shaft operation; I have a plan, a drilled out rotor and no idea at all if it will have enough torque. It will be super compact.
The stripped motor is a unipolar winding but runs the rollers through a long geartrain. The printer carriage motor is identical except it is bipolar but is direct onto the printhead with a toothed belt. I suspect the latter has more torque.
When I get this back together I will run it with the filament lifting a 1.5 litre water bottle. My calcs indicate a 20 bar nozzle pressure (from the forum) needs 14N thrust on the 3.2mm (worst case) filament. I may need to increase my nut 'thread' angle a bit to reduce the chance of the mini grooves I am making becoming munged together. My rollers are already have the ribs set 0.16 progressively offset to compensate for them each being 1/3 of a rev along the thread on the 2.7 degree angle I am attempting.
Much more testing to do . . .
PS. I have machined a small groove near both ends of each 4mm axle - the springs no longer go ping-boing.