1 Topic by tealvince

  • tealvince
  • Member
  • Offline
  • Registered: 2013-01-16
  • Posts: 319

Topic: Expermenting with new filament winder concept

For the last two days, I've been experimenting with a simple filament winder concept and gotten some promising early results. 

The existing Filawinder seems fine, of course, but I just don't have any space to hang it even if I could justify buying it for the relatively small quantities of filament I go through.  However, I've grown too frustrated to use my Filastruder without a winder, as it always seems to find a way to kink up on itself or be in the way just when I need it not to be.  I had some new ideas bouncing in my head and thought I'd try them out using parts I have lying around from a previous failed experiment.

Based on a prototype I saw at Maker Faire, the idea is to extrude through an over-drilled nozzle, but then stretch the resulting filament as it emerges so that it shrinks in diameter to the desired size.  Has anybody here tried this sort of setup with a Filastruder?  With the emerging filament under constant tension, it comes out really straight (no kinks!), and the filament doesn't need a long unblocked breeze-free path to hang.  As a bonus, the larger nozzle bore puts less stress on the Filastruder, and might even increase its output.

For now, I rigged up a spool to a simple dc gear motor and pwm controller, then put together a mechanism with printed worm and pinion gears that would move the filament back and forth across the spool.  The spool simply winds the filament faster than it's extruded, and that stretches it out.  I drilled out a spare nozzle to 3mm.

I've been surprised at the results I'm getting even with this crude fixed-speed setup.  A testament to the Filastruder, the filament is pretty consistent in diameter, at least until the increasing diameter on the spool starts to make it smaller.  I'm also getting better throughput than before at 360mm/minute, though my gear motor is running too fast even at just above stall speed, so the filament is only 1.35mm in diameter.  Once I replace the gearbox with a slower one, that should equate to 200mm/minute for 1.75mm filament.

http://soliforum.com/i/?8l45fXD.jpg

http://soliforum.com/i/?dBTyMmP.jpg

Here are a few pics of the setup.  Note that the winder and extruder sit right next to each other, so I can set them up on a tabletop anywhere and then put it away when I'm done.  Possible next steps are to add a filament diameter sensor and Arduino for speed control, or perhaps a hobbed gear to pull the filament, so its diameter on the spool won't be a factor.  I already have a slower gear motor on order.

I'll keep experimenting and post results as I get them.  In the meantime, I welcome feedback and suggestions.

Post's attachments

IMAG0996.jpg
IMAG0996.jpg 2.77 mb, file has never been downloaded. 

IMAG0998.jpg
IMAG0998.jpg 2.46 mb, file has never been downloaded. 

You don't have the permssions to download the attachments of this post.

2 Reply by elmoret

  • elmoret
  • Filastruder
  • Offline
  • Registered: 2012-10-11
  • Posts: 6,048

Re: Expermenting with new filament winder concept

Taking a step back for a moment, it might be easier to resolve the kinking issue first. Generally kinking comes from one of the following:

Filament guide too far from the end of the nozzle.
Filament guide material too sticky (copper to PTFE work well).
Not having a clear, uninterrupted path to the floor.

It is definitely possible (and should be expected) to extrude tens of kilograms without a kink. Here's one example from a customer in your Filastruder's era:

http://cdn.shopify.com/s/files/1/0239/9287/files/IMG_20130605_161548_2_1_large.jpg?190

Also current era motors result in extrusion speeds closer to 500mm/min.

Moving on...

The trouble with direct pull is that even if you control for the effective spool diameter (which changes as the spool fills up in a fashion that is impossible to know with certainty without a sensor), you can't control for pellet packing density. This can influence output rate by up to 20%, which if you're pulling at a constant speed results in diameter variation of the same 20%.

Ian and myself have worked with direct pull plus an optical filament width sensor based on a $10 linear CCD, but it turns out that you're already screwed once you measure a increase or decrease in diameter. Because pellet packing density is random, it is difficult to predict a trend and control it with a PID loop. Using the sensor sometimes made filament diameter worse, actually. Lyman explored this with his v5 filament extruder too. On v6 he changed back to a loop-based design instead of direct pull.

I hope that helps!

Filastruder - the robust, inexpensive filament extruder
E3Dv6 All Metal Hotend - the premier 3D printing hotend

3 Reply by tealvince

  • tealvince
  • Member
  • Offline
  • Registered: 2013-01-16
  • Posts: 319

Re: Expermenting with new filament winder concept

Thanks Tim.

Yes, unfortunately the problem for me is practical;  I don't have the floor space anywhere where I can safely run a Filastruder for hours without being disturbed (by munchkins, wind, etc).  When I used to run it in my workshop, (in a space now occupied by my laser cutter) I found that I didn't quite have enough room which is why it would kink up.  Even when it did work, I found it too much of a hassle not being able to use my workshop (more of a work room) while extruding and then having to spool the results up afterwards since the loops didn't fall neatly in the limited space and tended to tangle up instead.

Regarding dealing with the uneven extrusion rates, I anticipated that this would likely be the biggest challenge, which is why I'm thinking of it as an experiment for now.  It's good to know that you have tried direct pull already, but IMHO it might not necessarily mean that the problem is unsolvable, just challenging.  I'm sure you had many people try to shoot you down when you proposed the Filastruder, and fortunately you ignored them and persevered.  Besides, as a former ME Controls major, I figure it's about time I finally put my degrees in this area to work.  wink

I'm not sure I understand what you mean about the randomness making the system difficult to control, since randomness should be a given.   The main factor I see is how quickly the rate of extrusion changes in response to those perturbations, and insuring that the controls system can respond fast enough to the change so that any allowed fluctuations in filament size due to the lag in response are acceptable.  When in doubt, I like to measure.  So far, in my setup last night, (3mm nozzle, melt filter) I've measured a max fluctuation of 15% and any change seem to have appeared relatively slowly, so that makes me hopeful that it's possible to control it.

When you implemented it, where did you try placing your sensor?  Did you run into a problem that kept you from placing it close enough to the nozzle?  IMHO, its placement seems critical to minimize lag.  If necessary, I might try adding optics or using different types of sensors  to help.  If I can get a better idea of what solutions you tried any why each didn't work out, maybe I can find an alternative that wasn't tried or wasn't available to you at the time.

4 Reply by elmoret

  • elmoret
  • Filastruder
  • Offline
  • Registered: 2012-10-11
  • Posts: 6,048

Re: Expermenting with new filament winder concept

That makes sense about space. I just wanted to check the simple solution first.

I didn't mean to say the problem is unsolvable, but to get you up to speed on the challenges. I was a controls major too. big_smile

So the problem is that it all comes down to how the pellets pack into the auger, but you can't measure that. You measure the filament diameter coming out 10 minutes later, so you end up with a lot of lag in the system. We played with sensor placement. We didn't want to put it much closer than an inch or so from the nozzle due to heat from both the nozzle and the molten filament, but even then it was too late. Its just really difficult to control disturbances that are random. This guy used two sensors, one right at the die and one further down. The best he's done is +/-0.1mm:

https://youtu.be/eJP1tWNeoas?t=114

You can see that the diameter moves as much as 0.1mm in as little as 50mm (2 inches).

By comparison, here's a run with a Filawinder that did +/-0.01mm:

http://www.soliforum.com/topic/9818/or- … olerances/

Here's some more discussion and data:

http://forums.reprap.org/read.php?70,479153

http://www.thingiverse.com/thing:380987/#comments

In the end, if you down this route you'll want the best temperature control possible, since that is easy to fix/control and a contributor to filament diameter variance... so that's low hanging fruit. I think with enough time you can get a direct pull setup to +/-0.05mm, but I doubt it is possible to beat a Filawinder setup.

Filastruder - the robust, inexpensive filament extruder
E3Dv6 All Metal Hotend - the premier 3D printing hotend

5 Reply by tealvince

  • tealvince
  • Member
  • Offline
  • Registered: 2013-01-16
  • Posts: 319

Re: Expermenting with new filament winder concept

Thanks for the extra info!

Yeah, I agree that I don't think I can beat a filawinder for maintaining accurate filament diameter, and that's not really my goal.  As I said, a filawinder is a perfectly fine solution except the space and cost aren't ideal for me.  Even if I can get +/-.05mm that would be an improvement for me, since the alternative right now is to leave my Filastruder up on a shelf.

With the primative pulling winder I can now make my own bad filament which at least gets me back into the game.  I'll also see if I can come up with a compact layout for a non-pulling winder that might work better for my space needs than the existing filawinder.

6 Reply by wahada

  • wahada
  • Member
  • Offline
  • Registered: 2014-02-03
  • Posts: 60

Re: Expermenting with new filament winder concept

Tealvince,
Since you have no floor space have you considered mounting the filastruder and winder vertically on a wall?  I have such a setup that produces nice filament and takes up essentially zero floor space.

7 Reply by AZERATE

  • AZERATE
  • AZERATE
  • PRINTiT
  • Offline
  • From: Chicago
  • Registered: 2013-11-28
  • Posts: 2,077

Re: Expermenting with new filament winder concept

Laundry rooms are overrated big_smile

Printit Mason and Printit Horizon printers
Multiple SD2s- Bulldog XL, E3D v5/v6/Lite6, Volcano, Hobb Goblin, Titan, .9 motor, Lawsy carriages, direct Y drive, fishing line...the list goes on
Filawinder and Filastruder #1870.....worth every penny!

8 Reply by tealvince

  • tealvince
  • Member
  • Offline
  • Registered: 2013-01-16
  • Posts: 319

Re: Expermenting with new filament winder concept

wahada wrote:

Tealvince,
Since you have no floor space have you considered mounting the filastruder and winder vertically on a wall?  I have such a setup that produces nice filament and takes up essentially zero floor space.

Yeah, I have even less wall space than floor space.  Also, having an extruded setup mounted on the wall seems like it would be the ugliest thing in the world.  I'd probably only make filament occasionally, so I'm thinking of building a winder into a self contained box that holds the filastruder in some way and folds out or otherwise would let me quickly open it up, extrude a roll, then just as quickly put it away.

9 Reply by tealvince

  • tealvince
  • Member
  • Offline
  • Registered: 2013-01-16
  • Posts: 319

Re: Expermenting with new filament winder concept

AZERATE wrote:

Laundry rooms are overrated big_smile

My wife prefers that what's made in the man cave MUST stay in the man cave smile

10 Reply by tealvince

  • tealvince
  • Member
  • Offline
  • Registered: 2013-01-16
  • Posts: 319

Re: Expermenting with new filament winder concept

Success!

I went back to the drawing board, and built something mirroring existing Filawinder layouts, but built mostly using simple parts I had lying around and inside a portable frame that will become a box that can hold both the winder and Filastruder when not in use.  The Filastruder sits on top of the box when extruding, and will fit inside when not.  The spool rotates on a spindle that can also detach for storage.

http://soliforum.com/i/?tOADpKR.jpg

Instead of an Arduino and electronic sensors, it uses simple bang-bang on/off control triggered by a lightweight printed rocker mechanism and an optical switch I cobbled together from an LED, CDS cell, and solid state relay.  An off-the-shelf motor speed control unit lets me fine tune the motor running speed, though it worked great for me on the first try on a random initial speed with no additional adjustment.

http://soliforum.com/i/?pPEPs1J.jpg

I still have a lot to do, but this proves the concept for me.  To the right is an unused hobbed gear drive mechanism from a previous experiment.  I plan to take its motor and use it to move the filament back and forth in front of the spool (the mount for the feed tube is already hinged in prep for this).  And then I need to enclose and finish the box so it all looks nice.  But it's already creating the best filament I've ever gotten from my Filastruder.  No more coils on the floor ever again!