Notifications
Clear all

Wipe tower slowness  

  RSS
Chris Hill
(@chris-hill)
Estimable Member
Wipe tower slowness

I'm curious whether filament swaps can be speeded up for the MMU.  I'm printing a model with 5 colours of PETG.  At every tool change the extruder does a couple of passes over the wipe tower at very slow speed before unloading the filament.  I timed these two slow passes: ~30s.  For the model I'm printing, with 471 filament swaps, that's nearly 4 hours spent doing just those two slow passes, or quarter of the total print time!

I believe these two passes are the 'cooling moves' defined under Filament Settings -> Advanced -> Toolchange parameters.  I'm assuming this has something to do with the ramming procedure to ensure the filament tip is clean, and since PETG can be stringy these moves are slower than for PLA for example.

Has anyone played with these cooling moves to see what effect an increase in speed has on reliability?

Posted : 28/04/2024 9:42 am
Chris Hill
(@chris-hill)
Estimable Member
Topic starter answered:
RE:

The model I was printing completed without a single intervention, so the default settings seem very reliable.  There was an amount of debris in the body of the MMU after the 471 filament swaps - just ~20mm wisps of stretched PETG that the cutter has apparently removed during the loading process.

So just to see what bad things happen if you mess with the cooling moves, I did a small 5-colour test print where I turned off the cooling moves - the default setting is 1 cooling move and I changed it to 0.

Firstly the print time given by PrusaSlicer reduced from 2:26 to 1:36 - a 50 minute saving on the 100 filament swaps, so my stopwatch was correct.  Secondly, the print turned out fine.  I had one intervention, in which the filament didn't pull out of the Nextruder, but I was able to pull it out manually.  I'm struggling to see how that's due to the loss of the cooling moves.  I don't really understand what the cooling moves are for - I assume they give the filament time to cool a little, so that the ramming procedure produces a nice smooth filament tip, rather than a stretched out strand of string.  If that's the case, it's not clear to me how NOT doing that would cause the filament to jam in the Nextruder - if the filament was still hot wouldn't it pull out more easily?  Maybe a longer stringy tip caught in the Nextruder gears?

Anyway, after the 100 filament swaps the amount of debris was not obviously more than one fifth of what I experienced with the 471 swaps in the previous model, but I think the wisps were probably a little bit longer than the 20mm of the previous print - it's difficult to be sure.  

So my impression is that removing the cooling moves didn't make an obvious difference, but that's very odd, because there must be a good reason for including them and slowing down the print so much.  Can anyone shed any light on the purpose of the cooling moves and therefore the risk associated with removing them?

This post was modified 2 months ago by Chris Hill
Posted : 29/04/2024 10:42 am
TwinForce liked
muddymaker
(@muddymaker)
Estimable Member
RE: Wipe tower slowness

That's very interesting and I'm intrigued. That's a massive time saving with no considerable downside if any at all.

 

May I ask what brand of filament/nozzle temperature and color you used in your experiment?

Posted : 30/04/2024 6:02 am
Chris Hill
(@chris-hill)
Estimable Member
Topic starter answered:
RE: Wipe tower slowness

Yes, I'm intrigued too, but I'm a long way from feeling safe removing the cooling moves.  That filament jam is a worry, and could well be connected somehow.  And I've just popped the front off the Nextruder to find a fair bit of debris behind the main plate - although that was also the case after an earlier print.  They look like the stringly tails of the filament tip that have either been snapped of as the filament was withdrawn, or sliced off by the cutter and then pushed back in by the next filament.

Four of the filaments I used were Voxelab PETG (white, grey, yellow and blue), and the fifth was a 400g roll of black Prusa Polymer PETG that came with the MMU kit.  No prizes for guessing what I was printing 🙂  I used the generic PETG profile for all of them, except that I increased the nozzle temperature by 15 degrees on account of the ObXidian nozzle (E3D Mk4 version) - I find the temperature bump gives me the best results with the ObXidian nozzle.

Posted : 30/04/2024 9:49 am
TwinForce
(@twinforce)
Active Member
RE:

Extreme interesting what u find out here! I was assembling my MMU3 today and downloaded the example "Sheep" and was wondering how prusa reduce the purge tower time from 54 min if I sliced it to the 31 min in the presliced file. After changing the “cooling moves” from 4  to 1 moves, I got an equal time of 35 min. I created a ticket via mail to get the information, how they're done it. Maybe you're got it for me. 🙂 

If i got an response from prusa i will reply it here!

This post was modified 3 weeks ago by TwinForce
Posted : 28/05/2024 10:18 pm
Share: