The Remoter

Jinx

The old old old version of the reprap extruder was driven by a flex cable:



More info here.

I don't remember exactly but I think when I calculated it a flexible drive shaft for a extruder would need just about as much retraction as a bowden tube and adds about the same amount of friction. The only advantage I am aware of over a typical bowden would be the ability to do flexible filaments. (Surely someone has already done this. Does anyone know?)

Hey guys, that's a great idea!

I have a Dremel myself and didn't think about it. I'm sure those cables can transmit enough power but, what about the backlash?
I believe it would have to be used just like a normal bowden extruder because the backlash would cause some oozing. So, where is the advantage? If we could make one of those with no backlash that would be the perfect flexible transmision for Reprap. One application, would be sinchronizing both sides of a gantry style machine, so the movement is smoother, relying less on the smooth rods.


EDIT: I responded to this ddseeker post:

"a cheep Flexible Drive Shaft for a Dremal could possibly be used for the extruder"

without noticing the two other responses from Matt and Nicholas. So, it's very interesting that design was used before. Also interesting that I coincided with Nicholas.


Bottom line: I'll continue the development of the remoter with the original concept.

Edited 3 time(s). Last edit at 11/13/2013 03:50PM by Guizmo.

@MattMoses,

What was the specific function of the cable? Was it used just to avoid touching the filament with the gear?

Was not the original reprap a direct drive worm geared extruder... I was thinking that the flexible drive shaft would simply be replacing the motor's location on a direct drive extruder, still using the hobbed filament driver gear and the tensioned roller directly above the hot end.

Note sure how much torque loss and backlash you would get with using it like a dremel drive shaft.

(pardon my ignorance and incorrect nomenclature of things, still a newb)

Torque loss is not a big issue, you can always use a more powerful motor. Backlash is an issue, because many times the extruder has to retract the filament to avoid a fenomenon calle oozing, where the plastic keeps flowing after the motor has stopped pushing, due to the accumulated pressure on the melting chamber. So, when retracting, you aleviate that pressure, but if you cannot retract quickly enough, then you will have less than perfect prints. Bowden extruders suffer from that effect because over long distances the filament acts like a spring and it does not respond immediately to the motor movement, so the retractions have to be longer. Same effect would happen here, because there is always a delay of time between the input movement in on extreme to the output extreme, caused by the axial flexibility of the flexible shaft and the inertia of the parts. On resume, when turn to only one direction there is no problem, but when you have to change direction frequently, backlash becomes very apparent. So, for very precise machines, we always try to find a way to eliminate backlash or compensate for it, like bowden extruder's extra retraction programmed on some firmwares.

I'm very far from being an expert, but that's what I've learned

As promised, I have been working on a zero-backlash torque multiplier for the remoter. Nicholas.seward was kind enough to draw it up for me, since he's a lot better with solid CAD than me.


A quick explanation of what we're looking at: The two gears on the left are attached to cable pulleys, much like Guizmo's original design. Here's where things differ. The two pulleys are not directly connected. Neither are two reduction gears.

The hole that's visible in the gearbox brace in the foreground is the mounting point for one of the bowden cables. There is a similar one on the other side.

The point of this design is that there are two symmetrical gearboxes attached to the same drive gear. That means that the two pulleys move almost in sync, except for backlash. But when you tighten the bowden cables, all the slack comes out of the gearbox as the tension absorbs any play in the gears. If there is any static tension in the bowden cables, all the backlash is gone from the gearbox. With a gear reducer like this, very little friction is added; you could design the same zero-backlash reducer with worm gears, but the tensioning force would introduce a lot of friction between the worm and the gear.

So there it is, a zero-backlash torque multiplier for bowden drives.

nicholas.seward points out that you could mount a torsion spring in one of the pulleys to allow for stretching in the bowden cables without constant need for tightening.

Edited 1 time(s). Last edit at 11/13/2013 06:12PM by Annirak.

Good, good good . So one gear set works for one turning direction and the other set works on the opposite, with the torsion spring keeping the contact all the time, eliminating backlash, if I understood. I like it, much better than the worm design I did. I will incorporate a version on the design of the remoter, as a module to add.

Thanks!

I would highly recommend a stiff compression spring at one of the bowden sockets over a torsion spring. You would need a torsion spring with max torque of about 40in-lbf or a compression spring with a max force of about 30lbf. This is for a drop in replacement for GUS Simpson

The compression spring is much easier to source.

@Guizmo, you've got the idea exactly!

@Nicholas.Seward, I hadn't realised that the compression spring was the preferred alternative, but now that I think about it, it would be much easier to mount!

Yes, you're right.

Thanks Annirak and Nicholas. Any plan to use this system?

Well, a rough estimate of the additional costs put this at about $50 for GUS Simpson. That doesn't sound like much but that is more than the original cost for all the mechanicals.

I will definitely play with the design for my printers. However, it will be for medium to high range printers. It is too expensive for the low end.

When I have time (ha) I plan on making a version for GUS. It would also be amazing for a 5D robot arm printer.

Great!

I'm working on a 3 DOF robot arm printer and RD (remote drive) is a must. Just imagine the problems that would be faced if moving the motors with the arm, a total nightmare.

So you plan to make a version for GUS in about 5 years? I'm just kidding

I want to do a 6-DOF GUS with this. Not sure if it's manageable or not.

Wow! I'd like to see the concept Annirak. Mine is... strange. I'll post about it once it takes more shape.

Edited 1 time(s). Last edit at 11/15/2013 09:39AM by Guizmo.

I really like this idea.

Assuming this could be made to have continous rotation and no backlash, I'm thinking it could eliminate one of the Z-motors on a traditional mendel-style reprap.
I know the original Mendel had belt drive for the secondary, but I think it was abandoned due to backlash? (Maybe GT2 belts would be a simpler/cheaper fix for this)
Having synchronized motion would make calibration easier and enable manual turning of the screws without uppsetting bed/x-axis alignment.

On a tangential note, thinking about a bowden style drive (be it synchromesh/ball chain/speedometer/flex shaft) for an extruder, what if you combined it with a harmonic drive at the extruder?
I figure having a drive that is not easily driven backwards from the output would be nice at the extruder.
Given reasonably low friction in the bowden arrangement you could drive the cable at speed and still get plenty of torque through the harmonic drive.
Intuitively it feels like retraction calibration would be less messy and less dependant on the springiness of the filament than a traditional bowden extruder, but I may be wrong.
I wonder how durable a printed harmonic drive would be though...
The tendency for wind-up sounds like it may be a hurdle too, but I don't quite understand how it builds up in the mechanism.
I've never used a harmonic drive myself, so I don't know how feasible this would be, maybe it's just crazy-talk.

Oh wait... The remoter would be perfect for using external motors with a heated build chamber, wouldn't it?

Just throwing out some ideas here, please pull them apart at your leisure

/Andreas

Thanks Andreas,


All of them good ideas. I just hope one day the remoter can work continuoustly and without backlash. Well, backlash problem has been solved with Annirak's design, but for contionuous motion we would probably need a complete redesign. Feel free to to incorporate to any design or better yet, improve it.

Is there any reason we can't run a plain old cross-link chain through a bowden tube? I know that beaded chains are out due to some of the patents quoted above, but what about plain cross-link chains? We can still keep them in tension. As long as the link size is reasonably precise, we should be pretty well set.

If you just stay in 2D then that would work. You could also use a special chain that has a 90 degree twist every link for 3D.

You also need to smooth the outside of the chain.

Isn't the ball chain patent expired?

I just looked at the 1988 patent which is expired for sure. Using it we can just put a bunch of cylindrical beads on the wire. Big beads alternating with small beads. We can join the cable at one of the beads. The beads will need a sperical interface between them so we don't lock them in a column when we tension. Additionally you will probably want to solder many of the beads in place to keep from getting stacking error.

You could also use one type of bead that has shoulders on each side.

Sourcing the housing is the next problem. I guess we could make our own. All we have to do is wrap thin gauge music wire around a core and then coat that. Surely we can source it though.

I'd like it to be 3d, so something similar a ball chain is needed.

Making this at home will be extremely dificult. It has to be super precise to be usable. I bet there is something comercial already that can replace the ball chain.

@Nicholas, what is that chain with 90° twist?

Quote
Guizmo
EDIT: SPAL patent was re-examined in 1999, so it's still enforced. Anyway, it's good to learn and improve existing technology. Let's make it better!

As near as I can tell, re-examination does not reset the clock on patents. The 1988 patent has expired. Anything claimed in it should be fair game now.

Well, then there it is! Anyone want to use it?

How good do you believe those mechanisms will be with backlash? May we need 2 of them working in parallel just like your gearbox to eliminate backlash?

Yes, if you wanted to run two chains in parallel, tensioned against eachother, you could elmininate the backlash AND keep a torque multiplier in the system, just like with my gearbox.

It's a little early still but it appears you have the cables twisted on the first picture. What I mean is they cross to the 2nd piece and this will conflict and not allow it to work. Is that correct?

@tjb1: Thanks for the comment. Yes, the cables are "twisted" but that does not affect the system. If you analyze carrefully, you will see that there is no correct direction of turning, becasuse you can mount the driven mechanism anyway you want, and it is supposed you can drive the motor to any direction you want, so that's no problem. In fact, crossing the cables allows them to flex more easily than running them in parallel.

Quote
Guizmo
@tjb1: Thanks for the comment. Yes, the cables are "twisted" but that does not affect the system. If you analyze carrefully, you will see that there is no correct direction of turning, becasuse you can mount the driven mechanism anyway you want, and it is supposed you can drive the motor to any direction you want, so that's no problem. In fact, crossing the cables allows them to flex more easily than running them in parallel.

Yes, I knew it was too early... I see now, I thought both terminated at the same point. I now see one wraps. All good now, thanks.

Is there nothing more to be said?

Ralph

I will say more, I'm just waiting to have a printer to test, but surely, there's more to come.