Dual Filament adapter with an E3D v6 Hotend

[kraegar]

Gallery of a dual color treefrog: http://imgur.com/a/hssVb

SeeMeCNC asked if I'd take a crack at getting the dual filament adapter working with the E3D v6 on my Rostock Max v2, and now that I'm getting reliable results I wanted to share so others can try it, too. So far I've only been testing with PLA, but a quick try with PETG showed similar results, so I think it will be doable, too. (I don't print with ABS, so can't speak to that. I haven't yet tested semiflex or any other flexibles)

I started with printing a simple Yin-Yang, then once I had it down took a shot at the tree frog. Happy to say I made it through all 124 layers, 124 filament changes, 10,640 retracts, and nearly 5 hours of printing, and had zero jams. Yeah, the print is a little stringy, but that'll buff right out.

At first I got the constant jams others have run into. What I found was the key to not jamming on the v6 was to purge out a few mm of filament quickly, then retract very fast. My early experiments were doing just that - extruding 5mm @ 25mm/s, followed immediately by a 150mm retract @ 100mm/s. This worked, but left piles of filament everywhere. After thinking about it for a while, I realized I could get a similar result by doing a long-ish retraction (3mm), followed by a pause (2 seconds), then re-extrude that retraction (again, 3mm) and immediately retract the 150mm @ 100mm/s. On the other side of it, you can't extrude the last little bit (3 - 5mm at full speed. There's an air bubble trapped in the heatbreak, so you have to slow down. I find 25mm/s is good. This works very reliably. To tell how well it's working, look through your bowden and see if you're getting a good, clean break on the end of your filament. If not, play with the pause - I'm also doing a move to a "purge zone" so any ooze scrapes off on my prime tower. That move is after the 3mm retract, so adds to those 2 seconds.

I'm doing all of this in RepRap Firmware on a Duet controller, which allows me to use a "free" and "post" gcode macro for the toolchanges. These occur when "T0/T1" are called during a print. Someone clever can adapt the necessary bits to the Rambo. I've tested with Cura & Simplify3d. Since I'm doing the toolchange in firmware, both work just fine. I find a prime pillar from 20 - 25mm works.

Here's my "free" script:
M83 ; relative extruder moves
G91 ; relative moves
G1 E-3 F1500 ; retract 3mm
G1 Z10 F9000 ; move up 10mm
G90 ; absolute moves
G1 X0 Y130.00 F9000 ; move to back left side of the bed
G4 S2 ; Pause 2 seconds
G1 E3 ; Extrude 3mm
G1 E-150 F6000 ;retract 150mm @ 100mm/s

And here's my "post" script:
M83 ; relative extruder moves
G1 E145 F6000 ; Feed 145mm @ 100mm/s
G1 E3 F1500 ; Feed 3mm @ 25mm/s
G91 ; relative moves
G1 Z-10 F9000 ; move down 10mm

[gchristopher]

... I could get a similar result by doing a long-ish retraction (3mm), followed by a pause (2 seconds), then re-extrude that retraction (again, 3mm) and immediately retract the 150mm @ 100mm/s. On the other side of it, you can't extrude the last little bit (3 - 5mm at full speed.

Whoa! That's a really fascinating idea! I noticed the simpler "extrude a bit, then retract fast" method worked for getting the filament, but that way doesn't extrude any extra material. The size of the purge tower is a little discouraging, but that's about what I have to extrude when doing a normal filament change, so I don't see any way to improve that aspect with an E3D V6.

The dinosaurs are running E3D V6's and have a RAMBo already set up for dual extruders. I'll order the SeeMeCNC adapter and see if I can replicate your results with their setup. They're LONG overdue for some real testing of their beautiful 713maker effector platform, too.

Shouldn't most of the slicers support tool change gcode, too, so you don't need a firmware change? For as simply as you have it worked out, I guess you could insert that gcode as a pre-print step, too...

(edit: Ordered the adapter!)

[kraegar]

I implemented it in firmware primarily so I could change it rapidly mid print. I did lots of quick little tests with switching just to see what jammed and what didn't, and I could adjust and re-test much faster in firmware. (On RRF the changes are just made in gcode files, no recompile, just edit the file and save. Can even be done mid-print)

[kraegar]

Some additional points:

The distance to retract may need to change depending on your bowden setup. Test this, don't just go with 3mm. You want as short as you can that generates a clean break. Test the pause duration as well.

If you do get a jam, leave the hotend at temp, pop the bottom of the Y adapter off, clean out any filament strands with tweezers, and then push some filament directly into the short bowden by hand. That'll clear the jam, and not just make it worse. If you cool it down, you could end up needing to pull your hotend apart to clear it.

[gchristopher]

Jams seem to be pretty filament and temperature sensitive, too. I bet the best retraction profile will be different from PLA to ABS.

It's interesting that you've started with PLA. For me, ABS has been much less likely to jam. (I didn't even know the E3D's would jam until I tried dual-extruder PLA prints and had to deal with longer retractions to fight oozing.)

This will also be interesting to try with the volcano hot end. I was printing with two regular V6's mounted side-by-side, but only have one volcano, so I never tried any large nozzle multicolor/filament prints. I wonder if the tendency to jam or best retraction profiles will be much different?

[kraegar]

I don't print ABS, so have no experience to share there. I did figure if it was going to jam, it'd be with PLA. I had some good jams early, but they're what lead me to figuring out how to do it. No experience on the volcano, either.

As for temperature, I've tested the above from 195 - 215, with 2 brands and 6 spools of PLA.

[kraegar]

Prusa posted this, indicating that some v6's have a burr in their heatbreaks that may be causing some of the jamming people are running into.

http://www.prusaprinters.org/update-mul ... l-upgrade/

[kraegar]

This is what the tips of my filament look like after my switch process: http://imgur.com/msU4hia

If you're getting any more of a string than the one on the far right, you need to tune your pause / retract until it at least looks like that.

[kraegar]

FYI - I cleaned up the amount of ooze during the swap considerably by changing the line:
G1 E145 F6000 ; Feed 145mm @ 100mm/s
to
G1 E142 F6000 ; Feed 142mm @ 100mm/s

I'm still playing with the value to get it "perfect" but it's going pretty well so far.

[kraegar]

If you're doing this with a duet / reprapfirmware, you can replace the short retract with your firmware retraction values if you have them configured. So I'm just doing a G10 / G11, and also got rid of the 10mm lift that way. More efficient and less likely to string, since it's just your "normal" retraction that you've probably tuned.

[lightninjay]

FYI - I cleaned up the amount of ooze during the swap considerably by changing the line:
G1 E145 F6000 ; Feed 145mm @ 100mm/s
to
G1 E1425 F6000 ; Feed 142mm @ 100mm/s

I'm still playing with the value to get it "perfect" but it's going pretty well so far.

I hope you meant to type "142" there! That would be a horrid thing for one's printer to do.

[kraegar]

haha, thanks for catching that, I'll edit to make sure people don't make that mistake.

[kraegar]

Here's an alternate method I'm playing with. Only about 100 filament swaps so far with it, but it's working. Not as clean of a break as the previous. I'm testing a different extruder setup (More torque) but it wasn't working because I lost max retraction speed (can only do 60mm/s).

Here's the new "free"

M83 ;relative extruder moves
G91 ;relative moves
G10
G1 Z10 F12000

G90 ;absolute moves
G1 X0 Y130.00 F12000 ; move to back left side of the bed

G4 S2
G11
G1 E-20 F1200
G1 E20 F3600
G1 E-150 F3600


And here's the new "post"
M83 ;relative extruder moves
G1 E142 F3600
G1 E3 F1500

G91 ;relative moves
G1 Z-10 F9000 ;move down 10mm


The addition of the E-20 / E20 lines "pack" the trail that happens, and get it clean enough to swap. It's far from ideal yet, but it's working so far for me. If your bit of PTFE below the Y is very long, you may need to do more than 150mm retracts with this method - any tail you get has to completely clear the top of the Y adapter.

[Humanzee]

With lots of kreagar's help, some things we now know about implementing this using Repetier and MatterControl:

MatterControl does not support M83, and M83 will make crazy things happen if you try to use it in MC's custom gcode, so the M83 part has to go in the firmware. It doesn't support switching back via M82 at the end either, so all of this tweaking needs to be done in firmware as far as I can tell (short of using a slicer that supports M83).

What you're changing is in configuration.h in the dual extrusion firmware. You'll find:

Code: Select all

#define EXT0_SELECT_COMMANDS "M117 Extruder 0\nT0\nM84 P4\nG91\nG1E172F6000\nG1E5F500\nG90"
#define EXT0_DESELECT_COMMANDS "G91\nG1E-180F6000\nG90"

and a little farther down:

Code: Select all

#define EXT1_SELECT_COMMANDS "M117 Extruder 1\nT1\nM84 P4\nG91\nG1E172F6000\nG1E5F500\nG90"
#define EXT1_DESELECT_COMMANDS "G91\nG1E-180F6000\nG90"

Following kreagar's original versions of "free" and "release," those become:

Code: Select all

#define EXT0_SELECT_COMMANDS "M117 Extruder 0\nT0\nM84P4\nM83\nG1E145F6000\nG1E3F1500\nG91\nG1Z-10F9000\nG90\nm82"  // modified
#define EXT0_DESELECT_COMMANDS "M83\nG91\nG1E-3F1500\nG1Z10F9000\nG90\nG1X0Y130.00F9000\nG4S2\nG1E3F1500\nG1E-150F6000\nm82" // modified

and:

Code: Select all

#define EXT1_SELECT_COMMANDS "M117 Extruder 1\nT1\nM84P4\nM83\nG1E145F6000\nG1E3F1500\nG91\nG1Z-10F9000\nG90\nm82"  // modified
#define EXT1_DESELECT_COMMANDS "M83\nG91\nG1E-3F1500\nG1Z10F9000\nG90\nG1X0Y130.00F9000\nG4S2\nG1E3F1500\nG1E-150F6000\nm82" // modified

With any other value or code modifications changed accordingly. These are non-tweaked values to be used as a departure point.

Thanks again to kreagar for all of his work!

[clytle374]

I've been having great difficulty getting this to work right. First off is that pla refuses to reliably feed back into the the heat break. I often get a thick spot on the end that has no chance of going through the heat break. I've tried and sometimes it is impossible to even pull it back through a heat break. I've tried many different retraction distances and delays, even added extra delays.

I started to really look at the heat break and noticed that they seem to be drilled from both ends with a mismatch. I used some fulfillment and mashed it flat with pliers, put a bunch of ajax in the hole and lapped it out for hours. Didn't really help, sadly I realized later I lapped the smaller of the 2. Don't have anyway to measure the bores accurately. I've noticed that both heat breaks have a considerably larger hole on the nozzle end and seem to be tightest at the thin spot between the 6 and 7 mm threads. It appears that the pla swells up to this larger diameter and then gets stretched out when pulled through the heat break hot, then snaps back to the larger diameter. So I gave up on pla and decided to see if I could swap between other materials.

PETG and ABS seem to work much better but still can't seem to get the stringing issue solved. Just get a random string here and there, then they slowly build up until they produced a jam.

I'm starting to think that I have 2 unfortunate heat breaks. Trying to decide if I should get the proper tools and lap one out to a straight and smooth 2mm. One of them was bad enough that I could put a .078 drill bit shank a little more than 2/3 thru on one end and a little more than 1/3 thru from the other. I believe catching on the bottoms of the 2 drill bits mismatch. Of course I keep thinking that I'll hang up in the bowden tube next.

Any thoughts?
Thanks
Cory

[clytle374]

So I drilled one of the heat breaks out to .081 inch, it left a good finish. That mostly eliminated the issue with pla developing a thick stop on the end.

Still having horrible issues with strings. If you all hadn't gotten this working, I'd be sure it's impossible

The pulling out further seems to work better, and with a delay around 5 seconds, but somehow it then occasionally grabs material out of the nozzle.. like this