Dual bolt extruder drive

[jumpmobile 0]

Hey Guys

I have been searching for news about the Kraken hotend from E3D and occasionally stumbeled upon this thing. It has nothing to do with E3D (i think) it is just that the name is the same, but he (Ziggy from Brainydeep) got a thing to work that i was thinking about for the last few weeks/months.

 

He designed an extruder with two drive bolts and by that way doubling the area of friction between bolt and filament which in turn doubles the maximum appliable force onto the filament.

He did it by simply using the same mechanism twice (i didn't quite understand why he wanted to have the driving bolt mounted floating instead of the pressure wheel, but okay) and adding a second motor to it. Now this makes his design rather bulky, but it seems that he has made huge improvements to printing speed (350 mm/s ?!).

I was trying this for me a few days ago, when i saw that MoonCactus has designed a cute extruder in openSCAD (the only CAD-program i can use up to now):

http://www.thingiverse.com/thing:73082

Now all i wanted to do... ("all", yeah right) ... was to add the same bolt-bearring-tensioner-setup that MoonCactus used, mirror it and put it under the tension-spring (in a way, that the drive bolt is on the side with the cantilever and the bearring is on the stepper-side). By this way you could connect the two drive bolts with one pair of gears with a 1:1 ratio.

The sad thing is: MoonCactus seems to be a programmer and has a completely other approach to openSCAD than I have. I simply can't read his code. Also my computer freezes when I try to compile his design.

This is gonna be painful, so before i start on this i want to know if anyone has tried a similar thing? And if yes: why didn't it work? Or did it?

 

[illuminarti 18]

Nice project, for sure - Maybe I'm not understanding the design, but I'm not sure the two bolts part is particularly innovative, actually. For instance, the QU-BD extruders already pinch the filament between two contra-rotating rollers - the main difference may be these folks use two motors to do it, and so are able to get twice the torque - But maybe a single larger motor might be better, anyway.

The claimed throughput is impressive - a sustained 50mm³/second - the biggest challenge with that is cooling all that plastic fast enough, and also dealing with head-deceleration ooze at the corners; in the clip where they showed it printing a box, notice how the corners are piling up with plastic way above the print height? My RXL does that too, when printing quickly. The head slows down going into the corner, but the nozzle keeps oozing at the same rate.

 

[jumpmobile 0]

What i was thinking is: the part that slips as soon as the filament can't extrude anymore is not the motor itself but the drive bolt that starts grinding at that point. So there must be some "potential" left there.

When you have a driving system that is capable of blocking a NEMA17 as soon as the filament blocks then you have reached the end of what is possible for your system. Then you could go to adjust the spring tension on the filament so the drive-bolts will start grinding again, when filament is blocked.

Perhaps i will rewrite that tomorrow. It's late already....

But thanks for the heads up on QU-BD. I didn't know them by now.

THX F

 

[illuminarti 18]

Yeah, more contact area on the filament is a good thing to generate pressure while reducing risk of damaging the filament. But you also have to be able to manage the pressure at the other end.

Temperature management is another challenge when extruding fast. You need high temp (or big melt zone) to melt plastic fast enough, as it only spends a second or two in hot zone before being extruded. When you slow down the extrusion rate - corners, complicated paths, or lots of jumping around - suddenly the plastic is sitting at extremely high temps and can burn quickly.

 

[jumpmobile 0]

So a longer shaft on the nozzle and a thicker heatblock should help, right? I intend to turn and drill my own nozzles one day to see what a change in nozzle diameter does to my prints. So i could do this in one go.

On the material build up on corners: that is a problem i haven't been aware of yet, but it is obvious.

Do you have any idea how this could be fixed or have you tried to fix that?

Just thinking:

This could be fixed by changing the firmware. You should preprocess the next few steps, see if there is a corner coming and decrease the extrusion rate way before reaching the corner.

As far as i know there is a feature called "look ahead" that does a similar thing with the velocity. This also leads to less buildup on corners.

But when changing the flow there is much more down time than with changing velocity so you'll have to preprocess way more steps.

 

[illuminarti 18]

Yes, I personally think a longer melt zone is a good thing, on balance. It certainly allows for faster prints, but gives a larger volume of molten plastic ready to ooze, I guess - others, including UM in the UM2 have gone for a very short zone... I guess if you can instantaneously heat plastic right on demand as it's about to be extruded, that is the ideal solution. But, in practice you can't vary the temp that fast with current technology... so that you end up running the block hotter than you really want the plastic to be, so that the plastic quickly gets to the temp you do want it at... but then as soon as plastic starts to spend longer in the hot zone, it overheats.

As currently implemented, 'look ahead' is basically about managing the head jerk setting: basically, it checks to see if you need to slow down an earlier segment in order to be traveling at an acceptable speed by the time you get to a later one.

For instance... imagine a 150mm long line that has a slight kink at the end, and then a 1mm segment before a right-angle bend. The long segment is long enough that you can get to full speed. And you don't need to slow down at the kink, because it's virtually straight, and doesn't exceed the jerk setting. But then suddenly you're going full speed, and you've got 1mm to slow down to a virtual stop for that tight corner. So the look ahead works backwards, basically, figuring out what's the fastest that you could be going at the start of the 1mm bit. And then plans the longer line's acceleration and deceleration, to make sure it is going no faster than that by the end of the long bit.

Regarding the corner ooze... I think there's already some experimental code in Marlin (and I think in Repetier) that tries to manage the pressure variations as speed changes. I don't know if it looks within line segments (as you accelerate and decelerate) or just between lines of different speeds. But basically, you need to have some sort of internal model of the pressures, and adjust extrusion rates, or even do mini-retractions, to keep the pressure in step with actual speed, rather than ahead or behind.

 

[jumpmobile 0]

Hmmmm....

A model of internal pressure. That sounds difficult: You can't really measure it....

 

edit:

Except if you take a strain gauge ( DMS ) and measure the force with which the bowden tubing is pulled away from the print head. This should be equal to the force that is applied to the filament aka. pressure drop across the hotend.

Then you could model the hotend as a pipe of length a, diameter b that reduces to diameter c.

Environment pressure, moisture in the filament, friction of flowing filament are not included in that model.

Maybe this sensor will do:

https://www.sparkfun.com/products/8712

 

[johan 0]

Hi,

I made a double extruder for my Reprap two years ago. It is a combination of the Makerbot MK6 direct drive extruder, and a hinged Wades. It uses two high torque steppers rotating in oposit diections. I made this solution after having missed stepps on a single stepper direct drive extruder.It's like a tanks! :smile: It's not grinding filament, but I have had occational blokings in the hotend (j-head 0.35mm). I actually beleve that it is the unmolten filament that blocks the hotend at high speeds (FR 250)

https://dl.dropboxusercontent.com/u/55953222/Extruder%20closed.JPG

https://dl.dropboxusercontent.com/u/55953222/Extruder%20open.JPG

Johan