Ultimaker 2 pancake stepper direct drive

[Anders Olsson]

So, I should present this project a bit, since several people have asked about it, even though it is just a prototype yet.

The intention was to make a 1.75 mm direct drive feeder which was light enough to not affect printing quality significantly.
I built it because the combination of 3 mm filament and bowden feeder does not have enough precision/performance for some of my projects.

One example of what I want to be able to print is tiny things with retraction using the 0.25 mm nozzle and thin layers.
Another thing is printing ninjaflex with retractions and precision.

The design is far from finished, I know how I would like to modify it but I simply did not have time yet.
I avoided gears to make the design as simple as possible, but the plan was that the next version should have a gear.
Instead I used a 3.5 mm knurled shaft which I inserted in the motor shaft after drilling the motor shaft through and fixed with threads and a nut.

The other components are a ball bearing with a track, two printed components and some screws.

I will try to make videos and provide more extensive information later, but for now I will just post some pictures:

And here is what happens if you run the motor at 1250 mA instead of 600 mA.

FLIR0319.jpg (159.99 KiB) Viewed 19663 times

Both the motor and the plastic (ABS) parts survived, but I would not recommend testing this..

Tinkergnome's firmware has some nice features for adjusting current and selecting extruder, but I have not had time to test all of them yet.

[Neotko]

Ohh now I get it. So no spring? That clip marks are a great idea. And you drilled it :O well that sure makes it much smaller to add the bolt.

And withthat long distance through a ptfe to the head, no problems... That's just great. Maybe I should just install it over the head and get rid of the space problem. Also easier to keep two hotends like your idea.

[Anders Olsson]

The arm that holds the bearing acts as a spring, it has three settings, not optimized yet though.
You can print the parts and have a look
An even shorter tube would be even better, but going from 3 mm and 60 cm bowden to 1.75 mm and 8 cm tube and no bends probably is enough for what I do.
Drilling through the shaft is not that difficult if you take apart the motor and put the shaft part in a lathe.

I would still consider a geared version though, the 3.5 mm knurled shaft is not optimal, going for gears and a larger knurled wheel is probably the best option.

[reibuehl]

After seeing Anders' design, I started to build a version for my UM2 but since I lack the mechanical workshop, I thought to try it without modifying the motor shaft and using a more of the shelves knurled wheel. Unfortunately I do not just lack a mechanical workshop but also apparently the needed 3D modeling skills so I am currently stuck.
My idea was to use a knurled wheel like this directly on the motor shaft.

[Neotko]

The 4 holes screw on the motor are on the same place as other nema 17, so if you don't want to design it you could use the IRobertI feeder. He also has a guide to keep the filament inside for flexibles.

[reibuehl]

Interesting idea, but I am not sure that the space on top of the printhead would be large enough to operate a IRobertI feeder. I think it needs to be a slimmer design like the one from Anders.

[Neotko]

Yes indeed. But to run some test it might work. Then you can see if the wheel works or needs other system.

[LePaul]

Being the noob I am, it really never occurred to me that the heat of the steppers cause some of the filament issues

Wouldn't a cooling fan help? Or is it WHERE the stepper generates the heat?

[Anders Olsson]

I did some tests tonight.
At 600 mA, using the 3.5 mm knurled wheel, the motor skips at about 1.5 kg of feeding force:

The original UM2 feeder should skip at about 5 kg.

Larger diameter of the filament means that more feeding force is needed to create the same hydraulic pressure inside the nozzle though.
So, to compare these numbers you have to take the area of the filament into account. A 2.85 mm filament has 2.75 times the area of a 1.75 mm filament.

This means the direct feeder gives an extrusion pressure comparable to 1.5 x 2.85 = 4.1 kg for the original feeder.
Of course, the lack if a bowden tube will mean less friction, so I guess is that my direct feeder has roughly the same extrusion performance as the original bowden feeder.

If you use a larger diameter knurled wheel mounted directly on the motor, the feeding force will decrease with the diameter.
A 7 mm diameter knurled wheel would have half the feeding force at 600mA.

I also did some tests at 1000 mA and it seems like I can reach about 2.5 kg then, but grinding is limiting.

So one could consider a larger knurled wheel and a fan to cool the motor running at 1000 mA.

Larger knurled wheel will mean fewer steps per mm of filament though, meaning lower feeding precision, possibly limiting accuracy when using small nozzles and fine layers.
So, a geared feeder is interesting for several reasons.
The lack of heat transfer from motor to filament is another clear advantage of a geared version.

[Neotko]

Wow nice test! I will have my misumi gears this week, and hopefully a beta design finished.

[Anders Olsson]

Nice, looking forward to see your beta design!

One thing that I had in mind by the way is to take advantage of the UM2+ firmware for the direct feeder (in the UM2 case).
The new geared feeder on the UM2 has a 200 step motor and a total gear ration that might be possible to match with a geared direct feeder.

That way, one would only need to load the UM2+ firmware and set 1.75 mm diameter in the settings and lower the current to 600mA.
I haven't the calculations yet though, and I forgot what the gear ratio of the UM2+ feeder is.

[Neotko]

But also, to avoid having to tinker or install other firmware, to change the esteps or motor current it's quite easy with gcode.

Also I think tinkergnome said that latest marlin (not the um branch) have on their rc1 the posibility to have different amps for each extruder and even different pid. That's one of the problems of um, they haven't update the core for year and a half.

Anyway, with a gcode it's easy to di, the user should just print 'directfeeder.gcode' to change settings or 'bowden.gcode' (for example). Also tinkergnome firmware (only um2) allows to swap extruder0/1 on the machine.

[Anders Olsson]

I have been using that tinkergnome firmware for some time and is certainly has some nice features
The individual motor current is not 100% working though. I haven't had time to sort it out yet though, but one thing I noticed is that the second extruder current setting seems to need power cycling to really update the current.

The UM2+ geared feeder seems to have a 11:36 gear ratio and the knurled wheel appears to be about 10 mm in diameter by the way: https://all3dp.com/ultimaker-2-plus-review/
(I know Meduza told me these numbers some time ago, but I failed to write them down somewhere)
It might be unnecessary to care about details like matching gear ratio with the UM2+ firmware, but I like to simplify things for the end user if I can

[Neotko]

Indeed sounds like a great idea to keep it as much as hassle free as posible.

Maybe printing the extruder gear to mach that, and using a high impact/temp filament with a very small nozzle? I though about printing my gears with a small nozzle, but also I don't think it would last too much if it's too small gears. I don't really know a think about gears

[Anders Olsson]

My initial idea was actually to use a timing belt (like the short belts) instead, those gears could be a bit easier to print if you need small ones.
I prefer timing belts instead of gears, but I realize that it means one more component compared to gears (the belt).

We even considered to rebuild the delta printer at work because the gears in the geared feeder of that one sounds awful upon retraction.
But I think that for small gear sets like this direct feeder it is probably fine.