I wish I had some clear insight to share regarding the results of the retraction tests. Instead, some observations...

Both PLA and ABS exhibited the same issues, though ABS to a significantly lesser degree.

In both cases, regardless of varying temp and retraction settings, the same effect was seen, which was the formation of branches on the start point of a perimeter. Retraction settings DID make a difference, but not a huge one.

As a result of earlier tests, I decided that adding a part cooling fan might be interesting. They were interesting all right, but in completely the opposite way I was thinking. More on that in a bit.

While I have a long series of photos taken with prints done with different settings, there really is not sufficient difference to to justify posting them all here. So I'll just post one photo the exhibits the general problem, as well as the "surprise" related to the part cooling fan.


The object on the left is rendered with the part cooling fan off. The object on the right (surprise!) was rendered with mostly the default auto-fan settings (which basically just result in the fan being cranked up to full speed for every layer after the first). All other settings were identical between the two tests. Material was PLA.

While I don't have any photos to illustrate, the fan was DEFINITELY a help when it came to bridging. Haven't experimented with that specifically yet, but I do notice that the fill layers rendered on top of the infill honeycomb exhibit virtually no sagging, i.e., the resulting surface is pretty dang flat.

Along the way, I did some reading. Naturally, ideas were expressed that were in complete contradiction with one another, so it was kinda hard to get a read on what was going on based on someone else' experience.

One of the things that seemed worth doing was checking my extruder calibration. It was dead on. (Heh..., actually a little surprised about that).

I tried a print using fan settings clough42 mentioned over in the itty bitty thread. Exact same results, though I kind of expected that.

Read some stuff that made it pretty clear that part cooling fan shroud geometry and the specifics of the airflow can have a huge effect. I.e., it ain't an all or nothing kind of thing.

Etc.

Conclusions:
  • The retraction test is possibly a degenerate case, or at least a corner case that may not be entirely relevant to what I want to do with this printer anyway. It's still a challenge for me, but it ain't the end all be all...
  • For future research: how to best manage cooling for layers with many perimeters, because I think that is a relevant special case, and the Slic3r auto-fan settings aren't really up to it.
    • Read about direct rendering options that bypass a slicer altogether. Wonder if this puts in place a better situation wrt automated analysis of cooling needs, since the print path can be altered in ways that aren't really possibly when slicing.

  • The results may suggest OVER cooling, even with the part cooling fan off (remember that stock MF HE fan shroud..., it blows down onto the part).
  • I think it is worth trying clough42's HE cooling shroud.