A REAL fix for blobbing and support failure: Pre-retraction!

[626Pilot]

First, thanks for coming back to us. KISS is one of the best slicers, certainly better than Cura or slic3r in many ways.

I have a Rostock MAX delta printer with a Bowden-fed E3D v6 extruder, and even though it's a great hot end, I've had problems with stringing and blobbing from the start. Lowering retraction speeds (25mm/sec) seems to have fixed the stringing, but the blobbing? Still a serious problem! I sat down and stared at a print in progress so that I could figure out what was happening. It all happens so fast, so I had to watch for awhile, but eventually I figured out why it blobs - and a method that will probably fix blobbing better than any existing solution, other than "switch to a direct-drive extruder."

Even a hot end that's not actively feeding filament will ooze a few millimeters, just from the stuff in the melt chamber expanding and having nowhere else to go. When it has been extruding, the hot end temp is higher to compensate for heat loss from melting the filament, so the oozing is made worse. When KISSlicer gets to the end of a path, it stops, retracts, wipes, lifts, moves, descends, and then primes. The trouble is that it stops and retracts right at the end of the path. The combination of static oozing and higher hot-end temp means that the ooze will come out faster than the filament can be retracted. (You can try retracting at Ludicrous Speed, but it still won't be fast enough to beat the ooze, and it will only worsen the stringing.) The faster you print, and the taller your layers, the worse it gets.

There is a suggestion to slow down close to the end of the path here, and I think it's very much needed to help address this problem, but it's not enough. It can be augmented with another feature, which I call pre-retraction.

In addition to a setting that lets the printer slow down near the end of a path, there should be a setting that specifies pre-retraction distance. Rather than letting it ooze a blob while it stops and waits for the retraction to complete, it starts the retraction just before it gets there, perhaps around 1-2mm depending on speed and layer height. That way, feed pressure + static ooze = correct extrusion until end-of-path. At that point, the retraction, having already been in progress, will rewind enough filament that the ooze can't make it out of the nozzle until after it wipes and lifts.

It would take a few calibration prints to get it dialed in, but after that, it would work in concert with slowing down near the end of a path to produce much better results across a wide range of speeds and layer heights. By itself it should be enough to clean up slow, thin-layered prints. With the speed gradient, it should scale up to faster and fatter-layered prints. It should also help out some with supports, which work OK on smaller models, but have a tendency to get knocked around and eventually fall over on tall prints. The nozzle impacts a blob left from the last layer, and because the support is just a single line wide, it can yank it off the interface plastic. No blobs would mean support that works better!

I really think this could be the answer to the quality difference between Bowden and direct-drive extruders! The physics make sense. I just hope you have some time to try implementing it. If there is anything I can help you with, please let me know. I can get other Bowden printer owners to help test if need be.

[Isaac_alaska]

sounds like a good idea to me, this would probably help a lot with seam-hiding, too

[626Pilot]

Another thing occurred to me today. What if, instead of stop-and-retract at path end, the tool is jogged inwards to the infill space (or to the inmost line, if it's a thin part) and THEN retracted? If it oozes there, it isn't visible.

[626Pilot]

Another thing occurred to me today. What if, instead of stop-and-retract at path end, the tool is jogged inwards to the infill space (or to the inmost line, if it's a thin part) and THEN retracted? If it oozes there, it isn't visible.

Looks like the Depth and Gap settings handle this, but Depth only goes to a maximum of 1 loop. It might be better if it was 2 or 3. With Depth=1 and Gap=2 (maxed settings) I still get some protruding filament at path end. I think the Depth strategy might be better if it did a 90 degree turn and headed straight inward, rather than heading in on a 45. The way it is now gives the filament more exposure to ooze out. If it did a left/right turn right at path start and headed inward, it would "close its own escape route" and the filament would not be able to ooze outward!

Still, would be better if the retraction could start before the end of the path. It really needs to if we're going to get near-perfect seam hiding, especially on printers that use Bowden tubes. My setup has a much shorter Bowden tube than stock, but it remains an issue.

[mhackney]

Hey 626Pilot, you've seen my fly reels and other items on the SeeMeCNC forum. I am getting as near perfect seam hiding as I could imagine with KISS. I'm using 1 & 1 for Depth and Gap, no wipe, De-string yes, and use corners yes (not really applicable on around part though!) with 45° jitter. My Bowden is probably the same length as stock but I am using a Kraken with modified short-bore nozzles. They really minimize back pressure and result in crisper retracts. That might be the difference with my setup.

It would be great to have more control over the start and end path strategy as you describe though. My concern on retract before path end would be that it would be highly dependent on the printer's mechanics, Suck speed, print speed and probably a number of other factors. At one point I used wipe to clean things up and that worked pretty well. But after using the modded nozzles I found I did not have to use it. I also print my perimeters slooooowwww (20-25mm/s) to get the best finish and print from inside out.