Welcome! Log In Create A New Profile


Retraction optimization

Posted by Nate523 
Retraction optimization
April 10, 2016 11:18PM
Just curious to as what you guys do regarding retraction in a delta bowden tube style printer. I have it set at like 90 mm/s at 6mm retraction in order to get perfect cubes with no stringing. The problem is that this works when doing prints that don't involve too many retraction points, but when it starts retraction like crazy, say with a model with a lot of holes and gaps, the filament gets squished because it is getting retracted and unrectracted over and over again, which makes the friction in the tube increase and sometimes it will jam up the extruder. Is there a rule of thumb to look at a model and decide whether retraction is a good idea or not because I cannot find a happy medium.

Edited 1 time(s). Last edit at 04/10/2016 11:18PM by Nate523.
Re: Retraction optimization
April 10, 2016 11:39PM
My findings were you can only retract up to the distance of your transition gap. After I started, I moved to using a hexagon hot end with a bowden and if retracted more than 3 mm I would jam the hot end. Confusing at first and figured this out later. This is also what Eric from DisTech Automation would say to. He sells a hot end with customizable distances.

I tended to get blobs if I spent too much time at one point moving the exturder and not the hot end in x,y,z. Other advice I read is that strings can often be cleaned up with a utility knife. Blobs are pretty difficult to fix.

If you do a z retract as well ideally the extrusion and z-retract match in time. Now I am running a direct extruder which is much better for strings and blobs.

Edited 1 time(s). Last edit at 04/11/2016 01:20AM by CTCHunter1.
Re: Retraction optimization
April 11, 2016 12:34AM
Thanks for the info give it a look for sure!

What I have been finding is that it almost doesn't matter what speed I am goin either, because the retract will still occur as frequently and at the same distance, I know at slow speeds I can reduce the retract, but who wants to constantly print at 20 mm/s?!

I am trying to think of in my head a way to use friction wheels to extrude material instead of gears, that way it wont grind and deform the filament when retracting and such.
Re: Retraction optimization
April 11, 2016 01:23AM
I don't know of a hot end with a 6mm transition zone. If one retracts longer than the transition zone they bring melted filament into the cold part of their hot end where it instantly cools and jams it.

Edited 1 time(s). Last edit at 04/11/2016 01:23AM by CTCHunter1.
Re: Retraction optimization
April 11, 2016 02:40AM
Yah I figured 6mm might be a bit much, but thats what I had to make it not ooze. However, the hotend is not what is getting jammed at this point, it is the filament in the tube that gets so tough to push that the filament extruder itself either grinds the filament, or just starts stalling altogether.
Re: Retraction optimization
April 11, 2016 03:38AM
Usual bowden systems use a teflon tube with an inner diameter of around 4-5mm.
Some filament like PLA is quite stiff while most ABS types are more flexible.
This causes an issue in some bowden systems, especially if they are a bit longer.
Despite the obviously worse friction I found that the ridig air hose tubing works better than teflon here.
If you don't mind long retractions give the blue air hose a try and you can try to find something small enough in diameter to prevent excessive play of the filament.
For my 3mm filament back in the day I used 5mm air hose, had an inner diameter of 3.3mm - perfect.
4mm air hose is about 2.5mm inner diameter making it almost right for 1.75mm filament, but you need to change the tube connectors for the 4mm ones if yours are different.
Re: Retraction optimization
April 11, 2016 12:03PM
Many slicers have options to control the minimum length of filament extruded before a retract will occur. This is specifically to prevent going over the same area of filament with a drive gear. Perhaps have a look into if the slicer you use has this feature. A possible way to lower retract settings in a Bowden system is to use a "coast" or similar slicer function that will finish perimeter moves with the extruder motor off, letting some of the pressure in the hot end bleed off before the retract. Some firmware also has a pressure advance function which can do similar things, both can reduce the amount of retract you need to use.
Re: Retraction optimization
April 11, 2016 08:54PM
Air hose is an interesting idea I have not thought about that! I really don't want to mess with air hoses and new fittings right now, but I will definitely take a look into that int he future. As far as the minimum length of filament extruded before retract, that sounds great, except doesn't that mean that it wont retract then? which then leads me to the stringers which I am trying to get rid of in the first place? I might be asking too much for my printer to continuously retract and give clean extrusions.

I have looked at a rather expensive option:


Which looks great, except that in contacting them, they don't part out the gears and components yet, so even though you could print all the plastic parts, it seems you will have to fork out the 90 USD for a full assembly. Shame because I just wanted to buy the hardware and print the parts myself, would be a fun project and it looks pretty awesome as far as design point of view goes.

I noticed that my tube can handle the bent out of shape filament in lengths that are much shorter than my meter long tube. My build is like 550 high. So I actually have just modified with this idea:


It shte hover extruder, and I have seen some building these with success. I have not tested with lots of retraction, however, now to get no ooze, I only have to do just under 2mm of retraction versus the six previously mentioned. It also seems to have a lot more control of the filament because it is direct drive from the stepper motor. Definietly will update as I test, I am a little worried about vibration issues and wobbling now with the extruder mass being moved around a litte. Luckily the wat I have it up with bungee cords and counter weight, it doesn't seem to move from its centralized position at all. So here is hoping it will help smiling smiley
Re: Retraction optimization
April 11, 2016 11:56PM
It sounds like your tube might be too tight. Even you get PTFE tubing that's right in spec. your filament can be thick in diameter and cause seizing. I've ordered some PTFE tubing from amazon.com back when I was doing bowden. I looked up those orders and put them bellow. I've heard looser leads to more strings while tighter can lead to the problems your having.

Here is 1.8 mm inner diameter (ID) PTFE (ordered June 2015).

and here is 2mm inner diameter PTFE. (ordered June 2015)

I got this 2mm ID as well (ordered August 2015), don't remember why
Re: Retraction optimization
April 12, 2016 12:21PM
Thanks CTCHunter1,

I was wondering if it was possible to order thicker ones. The only issue being that it will take more retraction with those tubes because of the backlash? Still, definitely couldnt hurt to try. I have done some of the prints I was having trouble with aones with a lot ofretraction with this Hover Extruder idea, and I haven't had any problems with extrusion. So thats good smiling smiley
Re: Retraction optimization
April 12, 2016 02:51PM

What were your settings when you were using Hexagon and Bowden? I'm actually just finishing up with the hardware install of the same thing.
Re: Retraction optimization
April 12, 2016 06:47PM
To keep things relative, it would be interesting for those of us running bowden setups to post how long our bowden tube is and what our retract settings are for respective materials.
Re: Retraction optimization
April 12, 2016 11:51PM

Retraction 1.8mm. with .2mm advanced after retract. I had terrible stringing that I would clean up with a knife. I used have blobbing but choose settings for strings over blobs. The was on a HE3D extruder which doesn't even have herringbone gears. I also was often printing nylon @265 degrees which flat out leaks out.

I couldn't retract much more than 2mm without seizing the hot end. I messed up a couple nozzles too and end up using these MK8s by the end. I'm not sure that the hexagon nozzle is tappered right compared to the MK8. The CAD modals show it with an abrupt orifice with barely any tapper. My original hexagon nozzle has been siting in acetone for months. I did clear it once early on with guitar string.

I came close to destroying the hexagon a few times. I had to heat gun it twice to get clogged nozzles off the heater block with a wrench in a vice. The second time I had already striped the MK8 which has much shorter flats for the wrench than the stock nozzle. That MK8 went in the trash.

I think the hexagon is a good hot end. It's also priced nice. This is all within my first year of 3D printing. Now I'm using a Prometheus V2 which has also been good.

Firmware retraction compiled into Marlin. @ starup
M208 S-0.2 F180 ; Firmware retraction restart setting
M207 S1.8 F1600; Firmware Retract Speed

People say strings are also fixed by going faster in moves. My HE3D couldn't go very fast in moves.

Edited 2 time(s). Last edit at 04/13/2016 12:05AM by CTCHunter1.
Re: Retraction optimization
April 13, 2016 01:45PM
Hi Nate,
There are several issues at play here.

The primary one - you've hit on it - is filament deformation with multiple retracts.
This is unavoidable, and should be compensated for by using a small value in Slic3r's Printer Settings - Extruder 1 - Extra length on restart (if you are using Slic3r). If you have none, then the nozzle looses a bit of pressure with each retract/unretract cycle, if you have too much then there will be a pressure buildup and it will over-extrude.

If the drive gear is shredding the filament then this can be entirely mitigated with a drive gear with teeth that are not too sharp. This may not be an option with your present extruder but this is what I did: I started out with a hobbed grear that I made myself - it clogged a lot, so I purchased several Mk7 drive gears and found one that was well finished and had teeth that were not too sharp, and bored it out for the Nema23, .9 degree direct drive extruder I built - problem solved.
If you don't have the option to change the drive gear with your present extruder then you may want to consider an upgrade.

The extruder should have a tension bearing that has a semi-circular profile so it can provide high enough pressure without flattening the filament. You can make one very easily.
These steps will eliminate a major point of failure that can occur on long prints, and make extrusion more even. The issue of clogged drive gears is something that the Reprap community really should address...

Another thing that has been mentioned is the filament being pulled up into the cold area. Some hot-ends are obviously more susceptible to jamming when this happens, especially crappy Chinese clones. PLA sticks and deforms more as well so that could be a factor. The way to mitigate this is to reduce the retraction amount.

But there is one other thing at play here - filament reheat time. When the filament is pulled back into the cooler area, and then extruded - there is a delay in the remelt (I did a ton of experiments with this). The more the retraction distance the worse this delay is. The best way to mitigate this is to add a slow acceleration to the first mm of movement after the unretract using a post-processor, but that method is not available to anyone else yet... so what to do. Make the timeout as short as possible by minimizing retraction to the distance needed to reduce stringing, and make retraction as fast as possible - with the fastest reliable extruder acceleration, and with the fastest reliable travel speed. This will also reduce blobbing.

I have a very long bowden tube - 850mm. My tuned settings are: 6.7mm of retraction at 130mm per sec, 900mm/sec extruder acceleration, and 95mm/sec unretract speed (it's easier to retract faster because of the spring tension of the filament). Extra length on Restart: .15mm. Retract time is about 0.12 seconds, unretract takes about 0.15.

A procedure for tuning retraction has been documented here.

Edited 1 time(s). Last edit at 04/13/2016 02:03PM by Paul Wanamaker.

My printer: Raptosaur - Large Format Delta - [www.paulwanamaker.wordpress.com]
Can you answer questions about Calibration, Printing issues, Mechanics? Write it up and improve the Wiki!
Re: Retraction optimization
April 14, 2016 12:26AM
Very good information Paul, maybe soemtime if I feel like going back to a long bowden tube I will try to apply some of what you have detailed. I wasn't sure if there were gears out there that would not eat at the filament like mine does, the teeth are pretty sharp.

So far, with the bowden tube now only like 140 to 200 mm, with the hover extruder setup, I have had no extrusion issues, and it seems like it is very consistent with a 2-3mm retraction at 90 mm/s and it doesn't seem to delay at all really when it unretracts. I have the copy of the E3D J6 head from amazon, and I have to say, it is much better than the China brand that came with this printer! That was a piece of junk really...

Overall, I like this hover extruder idea, because I really don't like a huge bowden tube really, I like that it is short. I have a model printing right now at average 135 mm/s, 0.2 nozzle, 0.05 height, and it seems to be doing really good surprisingly, which I thin is contributing to more accurate extrusion response now. Not sure if I should be going that fast with a small nozzle though, been trying to research on what typical speed are and such for 0.2 nozzles.
Sorry, only registered users may post in this forum.

Click here to login