Notifications
Clear all

[Closed] Software Improvements for faster print speed.  

  RSS
tony.g5
(@tony-g5)
New Member
Software Improvements for faster print speed.

Just read about a new algorithm developed engineering researchers at the University of Michigan allows desktop 3D printers to boost speed without impacting quality. They did a print in half the time that was better than the slow print. They tested it on an HICTOP Prusa i3 and a LulzBot Taz 6 printers. Anyway something to keep an eye on.
Here is a link to the article,
https://all3dp.com/algorithm-3d-printing-time-michigan/

Posted : 04/11/2017 1:16 am
trond-eirik.k
(@trond-eirik-k)
Active Member
Re: Software Improvements for faster print speed.

Just read the same article, and I'm wondering if this algorithm may be included in a future software upgrade for the prusa firmware.
Will it? 😯

Posted : 04/11/2017 8:22 am
3d-gussner
(@3d-gussner)
Reputable Member Prusa-Translations
Re: Software Improvements for faster print speed.

Question is if this FBS algorithm will be open-source. But i looks promising.
The Trinamic drivers have stealthChop, spreadCycle, stallGuard, etc. and the question is how much faster the prints get using a combination of these features.

Posted : 04/11/2017 11:22 am
Thomas Drafting Inc.
(@thomas-drafting-inc)
Active Member
Re: Software Improvements for faster print speed.

Heres the video; super interesting.

John || MK3, MMU2, Creality CR-10S

Posted : 20/02/2018 7:14 am
devilhunter
(@devilhunter)
Reputable Member
Re: Software Improvements for faster print speed.

I doubt this will come to consumer printers at all.

The Trinamic 21xx Series is already pretty fast, getting up to 200mm/sec for infill via the the MK3.
For outer layers, you need to slow down to a crawl anyway, because you need time to lay down plastic for the outer layer neatly, and time to cool them off again.

Also we're already at the limit of the 30Watt heater cartridge in the hotend.

This could be useful for very quick and very sloppy prints with the Vulcano.

Posted : 20/02/2018 12:36 pm
Nikolai
(@nikolai)
Noble Member
Re: Software Improvements for faster print speed.

This article is misleading. The developed algorithm makes a poorly maintained printer perform better again. It doesn't make our great assembled MK3 print twice a speed.
Check hackaday article instead. They tested this algorithm.

Often linked posts:
Going small with MMU2
Real Multi Material
My prints on Instagram

Posted : 27/02/2018 1:08 am
metacollin
(@metacollin)
Eminent Member
Re: Software Improvements for faster print speed.


we're already at the limit of the 30Watt heater cartridge in the hotend.

Sure, except for when printing outer layers amongst other things, during which:

you need to slow down to a crawl anyway, because you need time to lay down plastic for the outer layer neatly

Basically, we're at the the volumetric limit for the hotend when printing infill or solid infill at 200m/s at 0.2mm layer height, and no other times. For all other situations the printer prints in, including every other action besides infill and solid infill at 0.2mm layer height and thinner layers, the hotend limit is not being hit and is not relevant. So I don't really understand why this is even being brought up. Yes, 200mm/s 0.2mm infill and solid infill can't be sped up due to the hotend extrusion rate. Everything besides those two exact situations is not at any such hotend limit, so can we please only mention said limit when it is...well...actually the limiting factor?

A very significant amount of a print's print time is spent on printing perimeters. And cooling is really only an issue when a layer takes less than a certain amount of time (which is filament dependent and filled in under 'auto cooling'). That leaves the whole idea of laying down the filament neatly. And that's totally true - if you want nice, accurate, high quality perimeters (which we all do - they pretty much determine the print quality all by themselves), you have to slow down. And the reason we have to slow down is due to the dynamic behavior of the printer. Things like vibration, the rigidity of the printer, inertial slop, overshoot, undershoot, ringing, those things combined are almost entirely responsible for why we have to slow down to get nice perimeters.

And all of those things are dynamic behavior, and anything that can compensate or mitigate them will yield and immediate increase in print speed (except when other things like the hotend are the limiting factor) while maintaining the same print quality.

This algorithm will improve any printer that must be slowed down during certain operations to preserve print quality. Which I believe is most of them, and definitely applies to any of the Prusa printers. Last I checked, the frame was aluminum, not solid diamond, none the less all-welded steel, so the idea of it only improving quality on crappy, poorly constructed printers is nonsense.

In terms of mechanical rigidity, every 3D printer that isn't printing at the hotend's limit at all times for reasons involving print quality is one that will be able to have those times sped up if it can either be made more rigid, or compensate for the lack of rigidity. This algorithm does the latter.

Unless your 3D printer looks similar to this (that is solid steel construction, not hollow by the way. Thing weighs around 1000KG), then it is not so rigid as to not see improved speed in any situation not limited by the hotend as a result of this algorithm.

The downside is that this algorithm likely requires significant manual turning both for a certain printer model, as well as per-device tuning similar to the 'dialing in' we all know and love/hate with our own printers, but that much more so. That said, it's legit and could definitely decrease print times significantly. The hackaday article inexplicably misses the point:

They printed a model at 0.1mm layer height twice as fast without any real loss in quality. Yes, they didn't double the print speed at 0.2mm layer height. I think this algorithm probably would speed it up a bit, though definitely not doubling it (due to hotend extrusion rate being the limit during some of the print time).

Personally, I would love to be able to print 0.1mm layer heights with the print time of the same object at 0.2mm layer height. Frankly, the hackaday article seems more witch hunt and nit-picking than substance. Hackaday says 0.1mm isn't standard, and 0.2mm is. Our own prusa printers claim 0.15mm is standard. The simple fact is, there ISN'T a standard universally agreed upon, so this paper and video is guilty of nothing that hackaday isn't also guilty of, which is dictating what is 'standard' print quality. If they're wrong for choosing 0.1mm, then so is hackaday for claiming 0.2mm is standard, or that 0.1mm isn't standard.

Basically, the algorithm does indeed have a very real and beneficial impact on print quality. The paper chose the situation where this impact would be maximized. And of course it did. They are trying to demonstrate that the algorithm actually has an impact and does something. What kind of idiot would expect anyone to chose anything but the situation that demonstrates this most effectively?

Well, I guess the author of that article for one =P.

Posted : 15/03/2018 1:37 pm
Share: