Odd behavior with silk PLA

Post Reply
SixTwoSix
Posts: 3
Joined: 25 Feb 2024, 14:11

Odd behavior with silk PLA

Post by SixTwoSix »

This is on a custom delta printer. Have tried both E3D V6 and Revo Six hot ends; same behavior with both. Bed and extruder temps are both PID-tuned, and never vary from the set point by more than one degree C even with the fans on.

I keep having this issue where paths that begin after multiple retracts tend to under-extrude for the first 20mm or so, which makes the top surface finish look uneven near the start of the path. The same surfaces, when printed without any recent retracts, look just fine. The problem is specific to Silk PLA... there is no such issue when printing normal PLA.

"So lower your retract!" - I can't. It's already at the minimum amount possible that doesn't cause stringing: trigger 5, min jump 5, vPrime 20mm/sec, destring 1.2mm, velDstr 20mm/sec, wipe 10mm.

The stuff seems to act like a non-Newtonian fluid, and the advice is to try very low retract/prime speeds. I've tried many values between 1mm/sec and 60mm/sec. It seems to make no difference.

So I'm thinking, okay, maybe we need "linear advance" or whatever. AFAIK, that's called PreloadVE in KISSlicer... but that makes it even worse! Any PreloadVE above 0 causes more under-extrusion near start-of-path, and it gets worse the higher that variable is set to. The "Experimental" PreloadVE wizard works fine with normal PLA, but not at all with silk PLA. In fact, using the value it computes tends to result in delayed extrusion of a huge blob of the material.

The best results I have achieved so far, after MANY hours of tinkering, trying different temps, retract lengths, wipe lengths, vPrime, velDstr, changing diameter of Bowden tube, etc., have been to set PreloadVE to 0.

At this point, I'm stuck. There's apparently no setting for extra length after retract. I'm pretty sure that would fix the problem, as the underextrusion resolves after a few seconds and the rest of the top surface looks good. I assume the point of PreloadVE is to achieve the same effect, but it seems like it would need to accept negative values to work with silk PLA. (Which it doesn't. I checked!)
Post Reply