Perimeter Underlap

Is anyone else having Perimeter Underlap problems? As you can see from the Photo, the start and end of the perimeter do not meet. The size of the gap varies depending on the object and where it is on the build platform. The results are the same with Sprinter, Marlin .9 and Marlin 1.0 firmware. By the way Marlin seems to produce much smoother and faster prints. The results are also the same using Sfact and Slic3r.

Have tried two different RepRap Prusa SAE printers and three different computers and the results are the same. Have not tried different electronics, as the only electronics I have are the Sanquinololu 1.3. Although I saw somewhere that there might be some kind of math error problem with the Sanguinololu have not been able to find any information on it.

Any information or insight into the problem is greatly appreciated.

Thanks,

Jim K.

Could you attach the STL for that part ?

The "overlap" could be ooze, did you set up retraction ?

---

Most of my technical comments should be correct, but is THIS one ?
Anyway, as a rule of thumb, always double check what people write.

Attached are the stl file and the scad file that created it. It is a repair part for a sliding door. I picked it because it is small and shows the problem. It occurs on all prints. Somtimes it is not noticeable.

Jim K.

Edited 1 time(s). Last edit at 03/02/2012 08:03PM by jimcking.

Forgot to mention that yes I do have retract turned on at 7000 mm/sec/sec (the default). Don't think it is a retract issue.

Jim K.

Same result with different firmwares seems to rule out math error in the IC code.

Does changing the number of perimeters or the layer height in slic3r affects the result ?

I suggest you post your slicer config.ini and the firmware configuration too - often other people can put their finger on a bad value you couldn't see. Oh, and a version of the resulting gcode would be nice too.

---

Most of my technical comments should be correct, but is THIS one ?
Anyway, as a rule of thumb, always double check what people write.

I never changed anything in the slic3r. It is all default. Just downloaded it and ran it. Marlin 1.0 is also default for the SAE Prusa. Just selected SAE Prusa and sanguinololu in the configuration.h

Jim K.

Perhaps you have clip set to a large value?

---

[www.hydraraptor.blogspot.com]

Or perhaps the speed or acceleration is too high for your extruder and it stalls at the start of a line, so misses the first segment of a circle?

---

[www.hydraraptor.blogspot.com]

nophead

Thanks for the information. Clip like all the other variables is set to the default. As yo can see from the sample print, it has both underlap and overlap depending on where it is. It is also very consistant as is shown by the even spacing on the top half. Have never seen the extruder stall except when the temperature is too low. It really doen't stall then either, the hobbed bolt just strips plastic off the filament.

Would you please print the stl file attached above and send me a picture of how it comes out on your machine? Have heard other people say your prints are great. The file is very small. Also, What electronics are you using? Have heard that their might be some kind of math error in the Sanquinololu processor like the old pentium chips had.

Thanks,

Jim K.

I will print it as soon as one of my machines comes free. They are all in the middle of long builds at the moment.

If the retraction speed is too fast for your extruder it will stall without stripping filament. Since the retraction only lasts small fraction of a second you might not notice it. The effect at a line end would be no reversal, so overlap. At line start it would be the fast forward that is missing, so it would be short of plastic.

The defaults in any of these softwares are just what works for the author, so may not work on your machine. You might have different motors.

---

[www.hydraraptor.blogspot.com]

nophead

Thanks for the info. Have tried 3000 and 7000 for retract and have not seen any difference. Will try others.

What electronics are you using?

Thanks,

Jim K.

DeuxVis,

Attached is the gcode file from SFact and the stl file that was used for the print that is in the picture. Used Pronterface to print it.

Maybe you could print it and take a picture of your results so I can get an idea of where to go?

Thanks,

Jim K.

This is what it looks like printed on my Mendel90



All it proves is there is nothing wrong with the STL. I don't use the same hardware or firmware or host. I sliced it with a version of Skienforge from nearly 2 years ago. I only use Carve, Export, Inset, Multiply, Preface and Raft. I deleted all the other modules.

Sorry I can't print g-code. I use the same simple hardware and software I wrote 5 years ago, which uses Ethernet instead of USB and sends binary commands in machine units instead of gcode in real world units.

I suggest retract settings of 1.5mm at 10mm/s. That is what I have always used, with no extruder acceleration.

Edited 1 time(s). Last edit at 03/02/2012 06:42PM by nophead.

---

[www.hydraraptor.blogspot.com]

nophead,


Thank You! Thank You! Thank You! I really appreciate you doing this for me. Now I know what it can look like and will not rest until I get the settings right to do it. Will try those retract settings.

Thanks Again,


Jim K.

Jim, I have tried generating gcode from your stl, and compared it with yours using Repetier host (nice gcode preview feature, see attached pictures).
It looks like you have some problems in your gcode : missing a part of the skirt, and... something that looks like your underlap (but less large).
I have generated my version of the gcode from your stl with skeinforge, and it does not have those problems.

I don't know what is causing this but it seems to be partially a slicer problem. You said it did it with both sfact and slic3r ? Haven't tried skeinforge ?

---

Most of my technical comments should be correct, but is THIS one ?
Anyway, as a rule of thumb, always double check what people write.

The difference is the effect of clip in SF. Without it you get complete loops for the outlines. With it it makes a gap and connects inner outlines.

---

[www.hydraraptor.blogspot.com]

I confirm I have clip totally disabled.

---

Most of my technical comments should be correct, but is THIS one ?
Anyway, as a rule of thumb, always double check what people write.

Thanks to both of you for all the information on this. I am starting to understand what is going on a lot more now.

nophead,

I really feel dumb about this. Have not been able to find where to adjust the amount of retract. Marlin 1.0 has an adjustment for retract acceleration. It is set for 3000 mm/sec/sec now. Did not see any option on how to turn it off.

In SFact Dimention I found retract speed. It is set for 15 mm/sec now I will change it to 10 mm/sec per your suggestion. The other setting is oozrate mm/min and is currently set to 75 mm/min. It appears to be doing some type of dynamic retract amount. Looking at it print, it sometimes retracts a very small amount and other times it retracts a very large amount. It does not appear to be doing it correctly either. It sometimes retracts a very large amount when I don't think it should retract at all.

DeuxVis,

Yes I did try it with Slic3r also and it appeared to do the same thing. In Slic3r it is set for 1mm retraction and retraction speed of 30 mm/sec. The part is very small and I run cooldown in SFact with a fan to get the part to come out. Did not see a cooldown setting in Slic3r to slow down the print on the small diameter top part. The result is that with Slic3r the print gets too hot and deforms even with the fan on. Have only tried Slic3r once.

Thanks again for all your help guys.

Jim K.

The latest version of Slic3r (0.7.0) now has a new "Cooling" tab which looks useful

Thanks jezmy,

Tried the new version Slic3r (0.7.0) and as you said it works with cool. Also, it does not have the underlap problem that Sfact has for me.

I turned off clip and nearly everthing else on Sfact and the underlap problem still occurred. The underlap or overlap changes depending on what the print is and it's orientation.

Guess I will just switch to Slic3r. Anyone know if it can be directly interfaced to Pronterface like Sfact it?

Thanks,

Jim K.

jimcking Wrote:
-------------------------------------------------------
> Guess I will just switch to Slic3r. Anyone know
> if it can be directly interfaced to Pronterface
> like Sfact it?


Yes, see the bottom of the readme here [github.com]