@Nicholas_Seward
Originally shared by Sven Eric Nielsen
A few days ago I’ve asked for a marlin modification to drive a remote extruder on a CoreXY printer setup. So the firmware is still a question, but this is my approach for a CoreXYE driven printer. All 3 belts are in a seperate layer in this concept. So basically there is no real belt crossover.
CoreXYE means basically a more or less traditional CoreXY belt system with a remote driven extruder.
This is the result of a short discussion I had with Nicholas Seward who made the sketch of the first CorEXY system I’ve seen. (Thanks that I can use your original picture for the comparison!).
In this discussion he mentioned that we will probably see some side forces on the X carriage due to the E belt. I’ve tried to compensate this in this sketch.
The good news is, that we probably have the possibility to drive the filament from 2 sides at the same time. The bad news is that this will be a really complex setup 
Maybe this approach can equalize the forces on the extruder carriage. But unfortunately it will add some torque to the extruder carriage (less problematic I think) and the same tilting problem on the Y axis that the H-Bot concept had.
I also made a quick and dirty 3D sketch of how a dual drive for the filament could look like for this approach.
Any ideas/suggestion?
What happens to the E belt when the gantry bridge moves in Y?
Lol… I’ve spent so much time on this that I didn’t notice this. Good point… Lol
But I my brain it worked. I have to review my drawing. Something is wrong here 
But first I have to stop laughing about my self
How about a worm drive for the extruder instead of mitered bevel gears? may give a bit more torque for extrusion. Also (I’m pretty sure you know this), try to keep belt length to a minimum, as belt stretch/tensioning is the main weakness of any Core(E)XY system
If you want to have the counter torque and you don’t want to have racking caused by the E you will need two steppers for the E and a layer 3 and layer 4 that look a lot like layer 1 and 2 except the belt completes the loop in the middle and goes around a pulley.
My first stab would be to make it symmetric with motors in each corner. I would then move the two string layers to one side of the plates and the two belt loops to the other side. What you lose in extra complexity could be gained in symmetric simplifications.
Ok, I found the reason why this is the wrong drawing. I’ve overwritten/mixed 2 ideas. I also had one setup where the E motor is fixed on the Y gantry. So this is a mixture and my other drawing is lost. Have to make a new one. 
@Ishaan_Gov
I’ve planned to use dyneema for X and Y. Therefor there is almost not stretching. But E would be a problem that could bring us the same problem like a bowden setup.
The worm gear idea could work. But it’s probably more difficult to print.
And then there is the question if it’s necessary to get more torque. Basically we have already alittle bit more torque than a standard Nema 17 because in this setup I’ve used a ration of 1:1,5. And normally a Nema 17 is already enough.
But yes, could be a idea for certain requirements.
@Nicholas_Seward
Yes, I guess you are right.
hmmm
But you’ve some new ideas in my brain 
I think I won’t spent time on my original sketch, instead I’ll play around with something different.
@Nicholas_Seward
Are you sure 2 motors are needed for E? I think it should be possible with the same motor and a layer 3 and 4. I think we only need to mirror your E belt. It’s late here and I’m tired. So am I wrong? Do I miss something?
@Sven_Eric_Nielsen One motor is fine but I think you need two loops of belting. However, steppers are cheap and can be slaved together if it will add symmetry. Just an idea.
@shauki
I know. That’s basically the reason why I’ve planned to revolve it 4-6 times.
In my setup there will be no axial movement
I’ve planned to use the same system like in cranes. This will cost some efficiency and is a little bit more complex, but brings accuracy and deletes the axial movement.
@Sven_Eric_Nielsen could you provide a bit more detail on your winding system? Most cranes use a spiral-groove spooling drum for the fast line.
@Ryan_Carlyle
It’s basically this system :
After I’ve seen it on a crane here in our street I thought : “there must be a person in the 3d community who already thought about this…”
And of course I found someone :
There are also some further investigations/discussions in a Google group. But I don’t have the link right now.
As I said, there are some disadvantages and it’s a little bit tricky to setup. But nobody said that 3D print and R&D work is easy 
The main problem I want to solve is that you can’t print the rolls. Every dimensioning or roundness failure would directly lead to a multiplied failure in your print. But I think there are ways to solve this.
And Btw, you will also get failures in your print when the dyneema or fishing line moves the roll up and down
Oh yeah, I’ve tried that before with spectra. The big issue is line slippage. The coefficient of friction of aramid-type fibers on aluminum/steel is pretty low. You need a LOT of figure-8 wraps and a LOT of line tension to apply any significant drive force.
You’ll also need to consider how to make the line loop closed/endless for E-drive. Might be fine to run a knot through the “traction winch” if you use the right kind of knot, but I don’t know.
Looped/endless timing belts are obviously a better choice for this kind of extruder drive, but then you’re greatly constrained by available loop length. And then the belt path has to be planar so you introduce miter gears and such. I think the best option is something like synchromesh cable that doesn’t require planar routing, but in a looped form factor. SDP/SI has some interesting options.
A round belt would work for the looser restriction of the E.
Won’t rubber change effective diameter due to normal force from line tension? Might be fine, I dunno
I’ve planned to use some kind of plastic pulleys. Nylon or something in this direction. That should help to improve the coefficient of friction. But ( @Ryan_Carlyle ) basically you are right, this is a “problem” that needs to be solved
For the E belt I don’t have the plan to use dyneema. Even though there impressive technics to make knot free endless strings, I think this is one step too far for the first try.
I’ll have a look on this synchro mesh idea. Even though it sounds expensive
Regarding rubber pulley… I guess in a setup with very low accelerations the rubber shouldn’t see some significant deformation. But you’ll definitely change your accuracy with the acceleration. But this is the theory. It’s a question of how big the influence is in reality.
I also thought about to put some kind of self gluing shrink tubing over a printed pulley.
I guess the merlin hotend is the best to check the quality of your xy accuracy because it doesn’t squeeze your extruded filament under the nozzle. ( @shauki )
But 0.2 is already definitely a step in the right direction.
Btw, I guess it’s entirely possible to adapt the merlin hotend idea to your hotend design. The only problematic thing is to get a M1.75 tap.
@Nicholas_Seward @Ryan_Carlyle
I’ve uploaded an updated version of Level 3. Now it should work.
https://plus.google.com/108869334922916902032/posts/9hx54PhXbPr
Yes, there is still the “problem” that the belt is incredible long. But this is just a concept idea.
