Identical or mirrored feature on opposite ends of part - inconsistent results

[marshaul]

I've been tuning and dialing in our Monoprice Maker Select Plus with ABS for a while now, and for most objects I can get usable prints now without much further tweaking. However, there is one issue which persists (it's also one of the first issues I observed, and one of the last to be resolved). It occurs with both PLA and ABS, but my focus here must be on ABS.

We are using the latest version of Cura to do our slicing, and using Octoprint to execute the print job. Short of dumping every parameter, I'm not sure what specific settings might be of relevance, but I'm happy to share specifics once I know where the focus is going to be.

I have attached photos of the two mounting holes, one of which printed acceptably (though not perfect), the other one of which did not. I can't post links, but the part is on thingiverse, thing 2153701.

First of all, note that these effects are extremely consistent from slice to slice, print to print. It doesn't just exhibit the same type of issue, but the pattern of malformed extrusions will be almost precisely the same from print to print. Whatever the cause, from the printer's perspective it is "intentional". I suspect it may be related to the persist presence of gaps between outer walls and the infill or ceiling wall (also visible near the tops of the photos), but I'm not certain.

It is most likely not a first layer adhesion issue; I'm printing on PEI and have had no issues with first layer adhesion since making the switch.

Note that neither of these holes is the start point (at least not for the first few layers). My first immediate thought would be "under extrusion after a retract", but the issue most recently resolved was precisely the opposite: too much extrusion after a retract, causing bulging/bloating of isolated features. Starting at 1.75mm retract distance/1.75mm retraction extra prime amount, we dialed it down to 1.0/0.25mm respectively, which resolved the excess extrusion issue completely (it was still slightly present at e.g. 1.0/0.5). Most notably, the problem of the mounting holes did not change substantively for any change in these parameters, leading me to conclude that "extra prime after retract" is not implicated, contrary to my reasoning.

Much earlier I also considered nozzle height variance across the print surface as a likely culprit, but this particular print does a good job of ruling that out, I think. Since I have to print it diagonally to fit it on the board anyway, the two mounting holes are very near the extreme top left and lower right corners. And, of course, the corners are the easiest points to set/verify nozzle height (since they are closest to their respective adjustment screws). I can confirm that, for the print pictured here, both corners were set via feeler gauge to 0.1mm nozzle height off the print bed. So, again, this would seem to rule out simple disparity in nozzle heights for the two holes.

And with those two ideas seemingly precluded, I'm out of speculation. I'm hoping that some of the experienced folks here can identify what the printer is doing wrong here.