Why am I having to run with an extrusion multiplier of 60%?
-
Do not underestimate the infill overlap effect. For me, 5% is good, 10% is puffy and 15% is ridiculous overextrusion (with S3D and a 0.4mm line width/extrusion width). An interesting way to dial this in is to use M572 (pressure advance) to reduce extrusion during speed changes, IE where infill meets the sides. You can (and must) run a much higher infill overlap value when using high values of M572 (depending on your actual extrusion hysteresis), or risk infill not touching perimeter. I once "perfected" my top surface finish by balancing acceleration, pressure advance, feedrate and infill overlap. I've since stopped using pressure advance until I can get caught up with the firmware that fixed some bugs, so I use a much lower infill overlap setting than I used when I had pressure advance activated.
-
@bot:
Do not underestimate the infill overlap effect. For me, 5% is good, 10% is puffy and 15% is ridiculous overextrusion (with S3D and a 0.4mm line width/extrusion width). An interesting way to dial this in is to use M572 (pressure advance) to reduce extrusion during speed changes, IE where infill meets the sides. You can (and must) run a much higher infill overlap value when using high values of M572 (depending on your actual extrusion hysteresis), or risk infill not touching perimeter. I once "perfected" my top surface finish by balancing acceleration, pressure advance, feedrate and infill overlap. I've since stopped using pressure advance until I can get caught up with the firmware that fixed some bugs, so I use a much lower infill overlap setting than I used when I had pressure advance activated.
I've dropped from the automatic 0.48 nozzle width down to 0.40, and am seeing better results, but I'm still heading down around the 85-90% extrusion multiplier before it gets neater.
Never thought about the overlap, so I'll toy with that as well.
What I am seeing are protruded corners in all extrusion settings, that I suspect my be retraction/acceleration related.
My retraction is set at 1mm@60mm/sec, and I understand that a direct drive extruder should run extraction between 0.5-2.5mm. Any suggestions?
What can be done to find best acceleration settings?
-
I don't understand why S3D uses oversize automatic sizing. It's true, that you can vary that size of the extrusion a bit and still get good results, but I like to stick to the default orifice size when possible.
I didn't pit much thought into the overlap originally, either, but I found that even small adjustments to it make a huge difference (when you are as close to dialing it in as you are). Definitely go overboard, too. Exaggerating things is often the best way to understand their effects (thank Don Bailey for that tip :P). When you figure out what happens when you go too far one way or the other, dial it in to where you need it.
If you are seeing protruded corners, then it sounds like pressure advance is needed. M572 0.1 is a good place to start, but probably too much for your setup, since it is direct drive. It is definitely correlated with acceleration and instantaneous speed change values, as the "gentler" you have the acceleration, the more the pressure varies in the nozzle, the more the pressure advance is needed. I had to play with a huge range of values of acceleration, feedrate, "jerk" (instantaneous speed change) on the same print and see what I liked best. I spent probably a week printing I bet nearly 50 little tests that had both straight lines, hard corners, large and small radii and text embossed into flat surfaces. You just have to try all the possible combinations and see which you like best. I find the best results are found by simply slowing down the outer perimeters in S3D and using moderate (not super conservative/slow) acceleration settings.
As for retraction, I personally find that between 0.2 mm and 0.4mm length at 5 mm/s speed is optimal for my setup, which is a nema 11 5:1 reduction direct drive through an e3d v6.
I wouldn't stress too much, your results seem pretty respectable already. Just pick one and roll with it. I tend to avoid "top surfaces" as much as possible by orienting prints at an angle and using support for stabilization. Usually the "top layers" of my prints are one or two perimeter extrusion lines, or even a tiny little point/blob.
-
My retraction is set at 1mm@60mm/sec, and I understand that a direct drive extruder should run extraction between 0.5-2.5mm. Any suggestions?
The best way I've found is to use firmware retraction - even if it's just temporarily. DC42's firmware uses G10 to retract and G11 to unretract. The retraction amount and feedrate are set by using M207 Sn.n Fnnnn where S is the amount in mm and F is the feedrate (speed). I believe David has just added extra parameters so that you can change the unretract speed and amount too. Anyway, for example I have M207 S2.0 F3000 in my config.g file but the great thing is that you can set a print going, then play around with the retraction by entering M207 commands through the web interface. Print 2 objects ( I use thin cuboids 3mm in X and 20mm in Y by about 30mm tall) spaced about 50 to 100 mm apart. Start with a low retraction and increase it until you lose all signs of stringing between the two. If you want, you can try it different temperatures and/or different filaments. This'll tell you what to use, even if you go back to "normal" retraction after you've set it up.
To use firmware retraction you need to tell your slicer that's what you want to do. In Slic3r it's easy - just tick the box under "Printer Settings" - "General" and it'll insert G10 and G11 commands instead of "normal" retraction commands. In S3D, you have to jump through a few hoops. I managed to get it working and posted my method here https://www.duet3d.com/forum/thread.php?id=396 before I gave up with S3D and started the process of getting my money back.
What can be done to find best acceleration settings?
My way is a bit unconventional. Basically, I use acceleration to limit speeds on short moves without it having to much of an adverse effect on overall print time. In theory, Slic3r has a setting for what it calls "gap fill" which it defines as being short zigzag moves. What I can't find out is how short a move needs to be and there doesn't see to be a way of setting the threshold. Anyway, imagine printing a largish square, say 100mm or more with infill at 45 degrees to the sides. I can print this quite happily at 90mm/sec or more and it's fine on the longer moves but as it gets into the corner, it appears to go faster and faster. In fact it's not, it's just that the moves are getting shorter. So, if acceleration was infinitely large and the speed is 90mm/sec, when the distance moved is 30mm, the time taken will be 0.33 of a second, at 20mm it's 0.22 of a second and at 10mm it's 0.1 seconds etc and it gets a bit alarming to watch. Also, you may be able to throw the carriages around like that, but the filament is soft a squishy and won't respond to those sort of changes in speed even if the extruder can, so the print quality will suffer. So, if we now limit the acceleration, on short moves, the axes will never get up to 90mm/sec because there isn't time and the shorter the move, the slower the speed will be. I've been playing around with it a bit and am currently using max acceleration of 1200mm/s^2 but I'm throwing around a great big heavy diamond hot end so you might well find a higher setting works better for you.
Anyway, that's just my twopence worth. It works for me, No doubt other people have their own preferred way of doing it.
HTH
Ian -
@bot:
If you are seeing protruded corners, then it sounds like pressure advance is needed. M572 0.1 is a good place to start, but probably too much for your setup, since it is direct drive. It is definitely correlated with acceleration and instantaneous speed change values, as the "gentler" you have the acceleration, the more the pressure varies in the nozzle, the more the pressure advance is needed. I had to play with a huge range of values of acceleration, feedrate, "jerk" (instantaneous speed change) on the same print and see what I liked best. I spent probably a week printing I bet nearly 50 little tests that had both straight lines, hard corners, large and small radii and text embossed into flat surfaces. You just have to try all the possible combinations and see which you like best. I find the best results are found by simply slowing down the outer perimeters in S3D and using moderate (not super conservative/slow) acceleration settings.
As for retraction, I personally find that between 0.2 mm and 0.4mm length at 5 mm/s speed is optimal for my setup, which is a nema 11 5:1 reduction direct drive through an e3d v6.
Where would this "M572 0.1" go? In the config.g file?
I'm a bit confused about retraction, as I read it's more about speed and you should set it as fast as it'll go without shredding the filament.
I've got mine at 60mm/s_1mm, and all seems well enough, but I see yours is a geared extruder of some sort….mine is a nema 17 with a 9mm OD drive wheel/gear.
-
The command to put into config.g would actually be M572 D0 S0.1 (D being the number of the extruder). Do a google search for "dc42 M572 reprap" and you'll probably come up with some of the posts where David has detailed the usage of the command.
I used to be in the same camp as you, and was printing fine with very fast retraction. Then, I tried 25mm/s and thought wow this is great. Then I realized that 5 mm/s was even better. I have no idea whether the gearing factor has much affect on the results. For me the benefit of doing a retract small and slowly is that it produces a very consistent sure movement. I noticed at times that the fast retractions weren't consistent, but that is likely due to the fact that I'm using a nema 11 and it might have been stalling with too quick of retraction. It's worth trying out very low speeds, you'll only waste a bit of time.
-
Time for an update because the problem I was having has mostly gone away. What's really frustrating is that I don't know why.
I had a minor disaster when trying out a new print surface which forced me to make several changes at once, some of which are irreversible so I can't go back to isolate which one was the cause. There is no reason that I can think of why any of the things I changed would have any effect on the apparent over extrusion issue though. Anyway, to cut a long story short, the print started to lift off the bed which destroyed the mini height sensor and ripped it's mount from the carriage.
So the first change was that I beefed up the height sensor mount and printed a new part for the X carriage. I managed to fit this without disturbing any of the belts or pulleys on the X axis - just a new plastic part which forms the undercarriage.
The second change was that I fitted a new DC42 mini height sensor - this one is the later design.
The third change was that while this was going on, my shiny new Duex5 expansion board arrived so I fitted that to replace the add on stepper driver that I had been using. Note that this is just to control the 3rd extruder for the Diamond hot end but that I had been having over extrusion like problems using just one extruder connected to the main Duet wifi board.
Obviously I had to recalibrate Z homing but this is something I do on an almost daily basis because I regularly swap between 3 sheets of glass, each with different print surfaces attached.
The other thing is that the machine must have sustained a fair old whack to tear the height sensor mount off the rest of the carriage but as far as I can tell, nothing is bent or twisted, everything moves freely with no binding or "slop" and prints are all coming out well - very well in fact.
I'm dammed if I can see any reason why any of those changes should have any effect but since making them, I've had to up my extrusion multiplier from between 0.7 and 0.8 to around 0.9 to 0.95 which is much more sensible.
-
All I can say is 'how bizarre' not very helpful I know but 0.9 tallies with what I am seeing so far with an e3dtitan and v6 combo…
-
This whole thing has been bizarre from the start. However, as further update, I still have problems with certain objects and it's definitely Slic3R that is at fault.
If anyone wants to try it, model a sphere about 35mm diameter. Take about 5mm off the bottom to give a flat surface for bed adherence. Slice with slic3R and print. The outer perimeters which are sort of overhanging get vastly over extruded. As an extreme test, use 5 perimeters. This will make the problem much much worse. Be warned, if you try this, keep the mouse hovering over the emergency stop button because you'll be hitting it about layer 4 if not before. Using 0.3mm layer height, with 3 solid bottom layers, after the first infill (layer 4) the outside gooey mess of plastic measured around 4mm in height! I'd say it was over extruding these perimeters by about 300 to 400 %
Which makes me wonder, what else is slic3R doing wrong?
Now if I could only find another slicer that can easily support 3 extruders, be made to do firmware retraction and ideally support wipe/priming towers…..............
-
Hi Ian
Have you tried raising an issue on slic3rs github page, they might be able to fix it as a bug or explain why it happens.
Cheers
Tony
-
If I recall correctly, I believe you said at one point you're using Slic3r version 1.2.9, correct?
You may want to try a more recent "development" version. They've made quite a bit of changes since 1.2.9 and I've been pretty happy with it.
Unfortunately you'd need to build it myself, but the instructions they provide aren't difficult to follow.
https://github.com/alexrj/Slic3r/wiki/Running-Slic3r-from-git-on-Windows
You could also try a pre-compiled build from Prusa Research
-
Many thanks for that. Yes I am using 1.2.9. I'll give it a go but it may be some time before I get chance now.
-
It appears as if Prusa Research has really picked up the Slic3r torch and ran with it! I wasn't aware of all the improvements they've been implementing into Slic3r until now, so if I were you I'd not even bother with building the development release and just try the Prusa fork.
I'm looking forward to trying it out myself.
-
Thanks again. On another thread, Tony pointed out that Joseph Prusa was doing some stuff with active priming towers - I'd love to get my hands on that for use with my Diamond hot end. Unfortunately, this whole 3d printing thing may have to go on back burner for a few weeks after tomorrow.
-
Just a quicky - which file do I download - is it the win64 full zip (I have 64bit windows)?
-
Yep! That would be the one…
Just a quicky - which file do I download - is it the win64 full zip (I have 64bit windows)?
-
Just wanted to put an update in here. I recently wrote a review (http://forums.reprap.org/read.php?1,755161,756827,page=2#msg-756827) on reprap forum of cloned titan extruders. In the process I switched from my genuine (but early release) titan to the Trianglelabs clone. The clone has a hobbed drive gear that is 7.35mm in diameter, my original has a 7.75mm drive gear. However I have had to increase my steps/mm from 333 with the original to 470 with the clone, with a 1.8 deg/200steps motor. Oddly both setups print really well with their respective steps/mm.
I cannot fathom why a 5% change in drive gear diameter has resulted in a 40% increase in steps/mm. Its probably coincidence but the 40% corresponds to Ian's having to run at 60% extrusion multiplier rather too well. The only factor that might be relevant, and I mention it in the review, the teeth on my genuine titan's drive gear are very sharp, whereas the teeth on the clone are much more rounded. Maybe the effect of the sharp teeth was that the original (possibly faulty as there was an issue with titan drive gears on early models) bit into the filament much more reducing their effective diameter, what seems odd is that it was not in the order of 40% smaller effective diameter, that would be very noticeable.
I'm not sure the answer will be very apparent, but if anyone knows, or has had the same issue I'd love to hear about it.
-
I was one of the ones that got a bad one from the first batch and that measure 7.75mm Dim on the hob the replacement had the teeth cut deeper so may well be around the 7.35 mark had loads of grief with the original bad one and even the new one struggled with a long Bowden Tube but is fine in a flying extruder config
Doug
ps I may still be able to get at the pics of them
https://forum.e3d-online.com/index.php?threads/titan-not-extruding-correctly.1214/page-2 post #30
-
So it sounds as though my genuine one had the correct drive gear, and that the clone has a copy of the problematic drive gear. Still doesn't explain the huge change in steps/mm.
-
Possibly got the number of teeth on the 2 gears wrong?
and the bad one is 7.75 co sounds like you got a bad one as well but it is very obvious which is which the one with the wider teeth is the good one