gwhite
Posts: 211
Joined: Wed Jul 19, 2017 9:37 am

Re: Gaps in Perimeters (Again)

Mon Jul 26, 2021 2:46 pm

I got all three processes into Excel, combined them all on one sheet, and then thinned it down to the essential stuff. I ditched things like support & raft settings, which aren't involved.

I've attached a PDF that has the differences in settings hilited, using Airscapes's gap-free process as my "gold standard."

A lot of the differences are minor, or should have nothing to do with the gap problem, like my fan speeds.

Here's the really condensed version of the differences I think may be involved
Gap Process Critical Diffs.jpg
As time & printer availability permit, I'll start modifying my process to match Airscapes and see if I can find the critical setting(s) that cause the gaps. I think I can speed up the process by using a binary search approach. I'll try the top half of the settings, and then the bottom half. That should narrow it down a lot quickly.
Perimeter Gap Analysis Process Comparison.pdf
(69.7 KiB) Downloaded 511 times

gwhite
Posts: 211
Joined: Wed Jul 19, 2017 9:37 am

Re: Gaps in Perimeters (Again)

Mon Jul 26, 2021 8:58 pm

Curiosity got the better of me, and I ran a test over dinner. I started with Airscape's gapless process, and created two new processes. "Test 1A" had my parameters inserted from the top half of the table posted above, and "Test 1B" used my settings from the bottom half. The break was between "outlineUnderspeed" and "retractBetweenLayers".

Here is the result for Test 1A:
Gap Test 1 A.jpg
and here is what I got from Test 1B:
Gap Test 1B.jpg
So, whatever settings of mine that are causing the gaps are contained in this list:
Gap Test 1A Params.jpg
The ghosting I observed in the original gapless print has disappeared in the gapless Test 1B print.

The next step will be to split the Test 1A parameter list in half again, and see what that does.

gwhite
Posts: 211
Joined: Wed Jul 19, 2017 9:37 am

Re: Gaps in Perimeters (Again)

Tue Jul 27, 2021 5:14 pm

More test results, and things are beginning to get interesting...

Test 2A was Airscapes's gapless process, with three settings imported from my gappy process:

extraRestartDistance = -0.05
retractionZLift = 0.2
retractionSpeed = 2400
Test 2A.jpg
There is one section of the inner perimeters that shows a small narrow gap. Viewed from the bottom, the start of all three perimeter passes looks consistent in width. Basically, something in these three settings is causing a gap, but it's much better than before.

Test 2B started with Airscapes's process again, but with the four remaining settings imported from my gappy process:

printPerimetersInsideOut = 0
minInfillLength = 5
defaultSpeed = 4000
outlineUnderspeed = 0.3
Test 2B.jpg
This configuration did NOT have any gaps, but if you look on the bottom, the start of the 2nd perimeter pass looks starved for filament.

It appears that the large multiple gaps I was seeing are not the result of a single setting. I think it's a combination of whatever is producing the small gaps in Test 2A, and whatever setting(s) are causing the underextrusion at the start of the 2nd perimeter in Test 2B. Put those two together, and you have a problem.

I will continue to investigate.

gwhite
Posts: 211
Joined: Wed Jul 19, 2017 9:37 am

Re: Gaps in Perimeters (Again)

Wed Jul 28, 2021 9:05 pm

I ran another set of tests with just the Extra Restart Distance and Z-Lift parameters in play. Both state with Airscapes's gapless process, but with a single parameter change to the values from my process that creates gaps.

Test 3A: ERD changed to -0.05
Test 3A.jpg
There is a single tiny gap that you can only see with strong backlight. It's nothing compared to the original problem. On the bottom, you can see three fairly even perimeters, with no obvious under extrusion. There is a small bump of extra material at the end of the final (inner) pass. The extruder is going counter clockwise, from inside to outside, so the inner most pass is the last one. The bump occurs at the very end, so it shouldn't be related to ERD .

Test 3B: Z-Lift = 0.2
Test 3B.jpg
This had another miniscule gap, and on the bottom the extrusion looks pretty even between the three passes. The bump seen in the test with ERD = -0.05 is absent.

Because I'm now dealing with increasingly small gaps, which require looking very closely, I also went back & double checked Airscapes's process. I couldn't find any voids or even thin spots where the back light shines through.

Extra Restart Distance appears to be involved, but the defects it caused by itself are very minor compared to the original problem. I probably never would have noticed them if I wasn't looking VERY closely. I haven't identified (yet) what caused the large underextrusion that I think is a major part of the problem.

More tests to follow...

gwhite
Posts: 211
Joined: Wed Jul 19, 2017 9:37 am

Re: Gaps in Perimeters (Again)

Wed Aug 04, 2021 9:21 pm

After lots of experiments, I think I have things pinned down. The gaps were the the result of a combination of settings. The biggest contributor was the Default Speed, followed by Extra Restart Distance, with a tiny effect from the Minimum Infil Length. In the course of chasing this all down, I discovered a few other things. The "ghosting" of internal features on the exterior walls were caused by a combination of the print order (Outside-In is better), and the ThinWallOverlap. If the Overlap is set below ~ 25%, it requires extra passes & retractions to do some of the walls (at least in this model), and that contributes to the ghosting.

I put together a table that lists the default settings for the MakerGear M2E, Airscapes's Gapless Process ("AGP"), my original process that had bad gaps, the effect of my settings vs Airscapes's, and finally, the process I settled on. It mostly uses the AGP settings, with the major exceptions being the print order and the Thin Wall Overlap to eliminate ghosting.
Final Table.jpg
One other thing I found is that the Zlift helps with the quality of the top surface because it avoids dragging the nozzle across features that have already been printed. When I started out with a 70% Thin Wall Overlap, it was over extruding, and the top layer had a lot of bumps that the nozzle dragged across. I originally found the 70% setting necessary to fill in around inset text on the bottom layer for other jobs, but it doesn't need to be that high for most of a print. For maximum top surface quality, you can always turn on Zlift with a process set just to run for the top layer.

In any event, I've learned a lot, and I now have a process that produces much nicer results that what I had before.

Thank you Airscapes!

gwhite
Posts: 211
Joined: Wed Jul 19, 2017 9:37 am

Re: Gaps in Perimeters (Again)

Wed Aug 11, 2021 2:26 pm

Well, there's more to the story. Once I went back to a larger version of my model, I discovered that without a small negative Extra Restart Distance (ERD), I was getting "zits" on the faces of some dovetails I'm using to built a larger overall structure. I could clean them up by hand, but that is going to be very tedious. I can get rid of most of he zits with an ERD = -0.07 mm, but as soon as I do that, I get gaps in the perimeters again, despite all of the other changes. ERD's around -0.01 or -0.02 don't have a noticeable effect, but by the time I get up to -0.07, the gaps are are quite evident, and appear to be the result of under extrusion.

I really think this is a bug inS3D. The ERD should only affect a very small effect at the beginning of a perimeter pass. I'm seeing under extrusion that lasts for many mm, and in some cases, almost a cm.

I tried the test models in PrusaSlicer, and without any tuning of the settings, I get prints with no zits & no gaps.

Here is a comparison of the S3D processes with ERD=0, ERD=-0.07, and Prusa Slicer ("PS"). The first shows the zits I was getting on the dovetails:
Dovetail Zit Comparison.jpg
And here is a comparison of the test file where I get the gaps:
ERD & Gap Comparison.jpg
PrusaSlicer doesn't have as many options for fine-tuning things, but their default algorithms seem to work remarkably well. The one thin arm where S3D tries to use 3 extrusions to make a 1.5 mm wide wall with a 0.40 mm extrusion width, PrusaSlicers sensibly does it with 4.

I'm going to report this to Tech Support, but for now, I have no choice but to switch to using PrusaSlicer. I don't like many aspects of the user interface and the way they store printing parameters in a bunch of separate places, but if it's the only way I can get quality prints, I will have to learn to live with it. Of course, if this is a bug, I can't expect S3D to ever fix it, given their track record on updates for the last 2 years...

gwhite
Posts: 211
Joined: Wed Jul 19, 2017 9:37 am

Re: Gaps in Perimeters (Again)

Thu Aug 12, 2021 3:26 pm

Just for completeness, here is my final factory file that was used to print the gap models in the previous post. Both are based on my "DGP" (Doug's Gapless Process). The first model is set up with ERD=0, and has no gaps. The second model is the same, only ERD = -0.07.
Gap Test DGP + -r07 ERD.factory
(85.89 KiB) Downloaded 501 times

Ret
Posts: 80
Joined: Wed Feb 07, 2018 4:26 pm

Re: Gaps in Perimeters (Again)

Sat Aug 21, 2021 5:47 pm

Have you made any changes to your M2's firmware or board?

Side note-
Scrap esun filament and move to a higher quality plastic.
Check your extruder, their plastic extruder housing is not great
Scrap S3D and move to Cura or Prusa Slicer

gwhite
Posts: 211
Joined: Wed Jul 19, 2017 9:37 am

Re: Gaps in Perimeters (Again)

Sat Aug 21, 2021 9:59 pm

I'm using Prusa now, due to this one issue. My MG2E is stock, with the factory firmware. The printed extruder housing does wear over time, but I've printed spares. The next time I replace one, I think I will drill it out and install a brass sleeve where the filament wears.

I've actually been very pleased with the eSun PLA+, and I've used dozens of rolls by now. I've seen reports of issues, but I've never had a problem I could trace to the filament. I've used HatchBox in the past, and I've noticed no significant reason to pay 15% more for it. I've also tried a couple other brands that were horrible. like Paramount. They had a specific color I needed, but it warps like crazy.

Telling someone to use a better filament isn't very helpful if you don't say what you consider "better". If you look at the reviews on Amazon, eSun has slightly higher scores than any of the PLA+ alternatives.

Ret
Posts: 80
Joined: Wed Feb 07, 2018 4:26 pm

Re: Gaps in Perimeters (Again)

Sun Aug 22, 2021 12:00 pm

I recommend switching your extruder for something like a bondtech if you ever plan to modify your extruder any bit. I recently switched mine out for a bondtech LGX and won’t be looking back at MG’s. It mainly shines with flexy material which the MG’s underperforms. Play with your jerk settings too-Simplify3d generates terrible gcode segments which often changes the flow of your extruder super rapidly which leads to inconsistencies, to what extent, it depends. Or just stay on Prusa’s slicer and avoid headaches.

Don’t forget- Majority of those reviews on amazon are from people with a sub $300 printer and wouldn’t benefit from a higher quality filament since their limitation is simply their printer and not the filament. I consider hatchbox to be in the same quality class as esun(in most cases).

Next time you feel like trying new filament, try a filament made of pure virgin ingeo resin(3D850 or 3D870). I personally use 3Dfuel since they seem to have the lowest price- made in USA as well. Both hatchbox and esun have used ingeo’s resin in some of their PLA’s but have been notorious for adding fillers. Both are also not directly on the official reseller list that NatureWorks provided.

Return to “Troubleshooting and Bug Reports”