Retraction

[destroyer2012 0]

I am curious if anyone has had good success with retraction. I have tried (at Daid's suggestion in some thread) retracting 2 mm at 50 mm/s and priming 2mm after but it seemed I was net losing plastic and oozing was not really going away. I noticed that he had done that at 200C whereas I am printing at 230C, which is considerably higher and my PLA is considerably more like goo.

We've all seen the crazy stringy prints people have been getting. Does anyone mind posting pictures of how retraction improves print quality? Anyone got any good examples of prints without strings as a result of proper retraction settings?

[alaris2 1]

sorry I don't have a before and after print but using Daid's settings for retraction made a huge difference to my prints. There's virtually no cleanup required now, any remaining strings are like fine hairs and are easily removed.

in short, I can recommend those settings. I'm printing at 210C.

[gael 0]

Hi Alaris,

The Cura's retraction function gives very good results.

Since I replaced the bowden tube with a tube with ann inner diameter of 4.0mm instead of 3.2mm, it's even better than before!

++

Gaël

[approx 0]

I have used retraction with both Cura and Slic3r... My exprience is that retraction in general works good with both of them for small objects..but for large objects - for example, where the printhead needs to travel from one end of bed to another, I still get stringing - the idea there i believe is to travel quicker... I usually retract 2.1mm at 70-80mm/second...

Oh another thing, I am not using the stock extruder.. I am using Taylor's extruder (mk6 gear)...

[Daid 306]

Note: I think the acceleration of Marlin limits the retraction speed, so I don't think 50mm/s or 70mm/s makes a difference, but I've not checked it to be sure.

[destroyer2012 0]

I have recently succeeded in getting decent retraction. I am extruding blue UM PLA at 180C, and retract 3mm at 20mm/s, building at ~40 mm/s

As you can see all i get are tiny strings the size of hairs, which is a major improvement.

Daid's Build me marlin sets the E acceleration rate to 10,000mm/s^2. We can calculate how long the retraction takes using calculus:

Se = speed of extruder, mm/s

Xe = distance extruder has traveled, mm in our case it's 1.5, since we accelerate half the distance and decellerate the rest.

Ae = 10,000 acceleration of extruder

t = time extruder has spent accelerating, sec

Te = total time spent accelerating, sec

Xe = int(0,Te)[se(t)*dt]

distance we have traveled

Se(t) = Ae*t

speed we are at, given time we have spent accelerating

Algebra:

Xe = int(0,Te)[Ae*t*dt]

Xe = (Ae*Te^2)/2

Te = sqrt(2*Xe/Ae) = 0.01732 sec

Assuming acceleration is constant we could reach (and the above calc. assumes we do) 173.2 mm/sec. Or seen another way, it would take 0.005 sec to reach 50 mm/s, roughtly a quarter of the time it takes to reach the middle of our acceleration path (and an eighth of the total path), thus it's a miniscule time compared to how long it takes to actually retract 3 mm. However there's also a jerk, saying how fast the acceleration increases, and I don't know what that is.

With these back of the envelope calculations it seems to me that reaching 80 or 90mm/s is entirely possible, and there _would_ be a difference between specifying 20mm/s retraction and 80mm/s. But again depending on the jerk, maybe I'm wrong. It would be nice to know EXACTLY how fast the retraction is, so we can have the retraction work for everyone.

[floush 0]

Hi Destroyer2012,

I've just saw that you're basically working on the same stuff as I do currently Good to see so much work happening on the retraction front.

It's really bad style to lure somebody from the forum into the nasty mailing list, but could you have a quick read through this thread and give a quick opinion on it. Possibly you have similar/totally different findings. We should just take care that we're reaching our goal as soon as possible.

Here's the link:

https://groups.google.com/forum/?fromgr ... gDZLDdpcoU

Thanks

Flo

PS: Add me on Google+ if you want to have a chat about this. Possibly tomorrow?

[floush 0]

One more: @ Gaël: Could you go in more depth on the tube you're now using? How did it help with retraction to give the filament even more slack? Would you say that you're having so much less friction that it's better?

Please provide the source where you bought it and share your retraction settings. This is really valuable info for the whole community and Ultimaker.

Thanks!

Flo

[destroyer2012 0]

I'd just like to update that I figured out another mod which makes retraction work way better: Putting duct tape around the filament holding screw and lever. Like this:

My extruder was rattling around a LOT on retractions before. After I did this, smooth & quiet, not to mention, zero strings.

So after I did that I could do 40mm/s retraction speed just fine, and I realized also that it may have been working fine before, just my code was screwed up causing it not to re-prime the extruder as much as it sucks up the filament. SO now that I fixed that AND the duct tape, retraction seems to be working well. Look at this:

THe top is messy because the head kept hitting the curling overhangs, but the bottom is string free!

[alaris2 1]

are you still priming with 2mm after? I went back and re-read your original post and realized you were doing that, but I was repriming with 0mm.

I tried a prime of 0.5mm just to see what would happen - and the strings came back.

I don't think I understand that observation since I expected the prime to occur after it reached the destination - not in anticipation of it!

anyway your pyramid looks much better. someone else was saying vibration was affecting the quality of their prints too..

[destroyer2012 0]

I have now changed it to suck 3.5 mm and prime 3.5 mm, at 40mm/s. So just to clarify it means it goes back 3.5mm and then forward 3.5mm, so no net gain or loss of filament. KISSlicer lets you set suck and prime seperately, which is why I am reporting it like this. Other slicers (slic3r for example) have like a "retraction" distance and "extra filament" distance, meaning it normally sucks and primes by "retraction" distance, and primes extra by "extra filament" distance, if you set it to not zero.

Anyway I have noticed I have a bit of play in my bowden clamp on the extruder side probably responsible for 1 mm of retraction distance. So if you have the stock quick-fit coupling with shim (resulting in zero play) you could probably get away with 2mm.

I imagine since you are extruding extra filament (suck 2, prime 2.5 right?) what happens is it leaves a big blob at the start of an outline, and then when its done with the outline the head touches and melts that blob again, causing strings.

[alaris2 1]

I see what you mean - it does say *extra* length in Cura but I missed that subtlety. yes, I would have suck 2, blow 2.5 I guess, at 50mm/s.

and yes, the blob at the start of the outline was my guess too - since the strings start from the opposite side from where you'd normally expect.

I had observed (in my earlier photo) that z layers vary in quality. it seems the layer after a layer involving retraction isn't quite primed enough, leading to loss of material along edges in particular.

I'll try the same print with your 3.5mm settings tonight and see if there's any noticeable difference for me. you really think the shim on the quick coupling makes a big difference? easy enough to print one and try I guess.

cheers, nik

[futurejames 0]

I'm running the 4mm ID / 6mm OD bowden tube setup from here

http://www.thingiverse.com/thing:21674

. And I stole destroyer2012's idea about duct tape but I used a twist tie.

I started playing a bit with retraction. I set it to 2mm and it made some difference. I'm currently running at 3.5mm right now and its making things much better, there's almost no hairs. I'm pretty certain with this setup I can risk putting that value even higher because of how little friction there is.

[alaris2 1]

my observation was that 2mm worked OK, but the problem is in re-priming the nozzle - I can get no hair prints, but just after retraction any fine details are lost - this is really bad news on the first layer where sometimes the PLA won't stick to the bed.

I tried re-priming with an extra 0.5mm but that leads to a different sort of stringing.

I intend to get some 4mm ID tube as soon as possible - where did you get yours from fj?

[futurejames 0]

I first tried to get it at McMasters but they won't ship outside the US.

I found another place, flurotherm.com that carried some, and were willing to ship outside US.

In the thread where we were discussing it, I recall that gael had published a link to a place in France that was selling it.

[ian 32]

i admit im still a newby to ultimaker, so here goes my next blond question

Im running Ultimaker PLA today at 230 degrees and my ultimaker is running like a charm at the moment.... and yes Florian im starting to fall asleep with the sweet sound of it printing !

Ok question, what settings or profiles do i have to use or where do i find them, to bring down retraction. Its not the biggest problem for me but would always be nice to make it better.

And yes I know I am the professional for asking stupid questions ! :lol:

Best wishes guys.

Ian

[Daid 306]

Florians retraction settings, which work pretty good, are:

Retraction

Minimal travel: 5.0mm

Speed: 70 mm/s

Distance: 4.5 mm

Extra length: 0 mm

Dwindle (in expert settings)

Enable dwindle

Pent up volume: 1.0 mm3

Slow down volume: 6.0 mm3

[alaris2 1]

ooh, this should be interesting to try. I've had dwindle switched off until now. That retraction looks pretty aggressive compared with the default Cura settings!

[Daid 306]

Cura's defaults are usually quite conservative. But I already had pretty nice results with 2mm retraction at 50mm/s without dwindle. These new settings seem to work better. However, I had a print failing today which I think was caused by the dwindle. (I was printing loads of support and there was too little plastic ending up in the print)

[joergen 2]

However, I had a print failing today which I think was caused by the dwindle. (I was printing loads of support and there was too little plastic ending up in the print)

that was my main experience with dwindle, even printing without support, but similar things (bunch of small support pins, 3x6mm). with dwindle on, the outside was completely incomplete, which is why I gave up on dwindle, and just use retraction, since that is working great for me.

[alaris2 1]

I tried printing Daid's Ultimaker feet last night with dwindle.

without dwindle (retraction only) the first 10 layers or so are just a mess.

with dwindle on, it comes out perfect. really impressed with these settings - however the printer makes the weirdest noises so you have to prepare yourself for that!

I generally don't print things with support, since I've never had any success with support ;(

Joergen's comment is interesting, since I've observed exactly this effect on similar parts, but it was always due to aggressive retraction i found.

I do think the 4.5mm retraction might be too much - I have also had perfectly good results from 2mm. it would be interesting to hear the logic behind the numbers, or is it just trial and error?

[floush 0]

The values I came up with are based on experiments with the hollow calibration pyramid.

I increased the retraction from 2mm up to 5mm in 0.5 steps and found 4.5mm to be best.

After that I played with dwindle and found it to be very confusing. At some point I was sure it isn't doing any good to the print, but when comparing a hollow 4.5mm retraction pyramid with one PLUS dwindle it still looked better (most likely just due to lower speeds...).

So don't take these settings as a scientific fact, it's just a first try. Everybody should play with these settings and share the results.

4.5mm is quite a lot, so with "retraction-heavy" prints you end up with destroyed filament... therefore we need to get this number down by other means and improve the filament feeder to be less destructive to the filament.

Cheers

Flo

[destroyer2012 0]

I can't understand how you guys manage to retract at 70mm/s because anything faster than 40mm/s for me results in skipped extruder steps and failed prints. THat said I am now retracting 4.5mm but because I've switched to printing ABS my stringing has gotten a lot worse. Well not because I switched to ABS but because I changed my bowden cable to be 4mm ID, which is a lot stretchier, but also ABS itself is more elastic than PLA. So I think my setup is now requiring more retraction because the filament/bowden system is stretching more than it was before.

Another thing is the ABS tends to make little blobs that hang off the side of prints rather than strings. SO for example the print head oozes while traveling so all the oozate ends up wiped off the part of the print it's arriving at, and this leaves a nice row of little blobs on the smooth surface of the part, signifying where a travel move ended. THe blobs are pretty easy to remove as there's almost nothing holding them on (and ABS is less sticky than PLA) but I also get blobs at the start of travel moves because 40mm/s is not fast enough.

I wish we had some way of confirming what current my motor is actually getting without fancy equipment so I can tune my system to be more like yours (as I suspect not enough current is the reason why I can't get to 70mm/s).

[Daid 306]

Most likely your extruder skips because the motor power is wrong. My extruder is so strong that I find if very hard to turn it by hand if the motor is powered.

[joergen 2]

Most likely your extruder skips because the motor power is wrong. My extruder is so strong that I find if very hard to turn it by hand if the motor is powered.

The FW (marlin) has it capped at 45mm/s, and setting it faster won't do anything. neither faster retraction, nor skipped steps.