Welcome! Log In Create A New Profile

Advanced

algorithm to speed up prints?

Posted by shadowphile 
algorithm to speed up prints?
June 15, 2018 06:44PM
This works by modeling the dynamics of your printer to compensate for deflections and vibrations. (which I've always wondered how feasible)
Is this something of interest for the future of the Duet firmware? The article mentions it could done in the firmware (probably HAS to be at that level given the bandwidth requirements for physical vibrations can be awfully high)
link

Edited 1 time(s). Last edit at 06/15/2018 07:21PM by shadowphile.
Re: algorithm to speed up prints?
June 15, 2018 11:43PM
I read the Article and the comments below Noting hackaday comments.

Watching the Video from University of Michigan.
Low and behold I saw and heard my 3D Printer's1. They sounded like this before I learned how to properly align and tune them.
And even then with all the chatter from loose belts and squalling lead screws my printers turned out good prints on their worst days with original firmware.

Now that I am using a properly configured Marlin 1.1.X and have corrected all mechanical alignments. My tuned 3D Prints appear to be comparable to prints in the Article.
I have not calculated my Max filament extrusion rate but by test printing can state that it is close to it's max limit now.
To 3D Print any faster the Printers Hot-end heater cartridge needs more wattage or replaced with a modern E3D’s Volcano Hot-end and Bowden extruder.
For some reason 3D Printers can not print faster than the Hot-end can heat filament!!!


*1 --- I mean the 3D printer in their Test video is the same Printer I have.
Re: algorithm to speed up prints?
June 16, 2018 07:33PM
"For some reason 3D Printers can not print faster than the Hot-end can heat filament!!!"

Isn't this rather obvious? If the filament isn't molten, how can it keep extruding? Or was that sarcasm? smiling smiley
(edit: the actual limitation is the low thermal conductivity of the plastic itself. The Volcano doesn't have any higher wattage, it just extends the time the plastic is given to melt before it is pushed out the nozzle. Higher wattage heaters are more likely to burn the plastic before the temp sensor can tell the heater to back off, or at best minimize temp variations in the heater block.)

Anyway, thanks for the hackaday reference. I read it and the following endless argument. I did not question the results because I know enough about control theory to understand that any system can be controlled better (without feedback) by a more accurate dynamic model of the system. How that model is created (and I'm sure it requires measurements of an actual system) and how practical the algorithm is to implement is the HARD part that seemed toot-worthy.

Based on what I read from the hackaday article I would retract any immediate request for dc42 to jump in and make it happen. I'm used to a similar relationship between scientific papers about improved graphics algorithms (speed and realism) and software developers who CAN read those papers and immediately implement the improvements, so maybe I posted a dumb request out of excitement!

Edited 1 time(s). Last edit at 06/16/2018 08:16PM by shadowphile.
Re: algorithm to speed up prints?
June 17, 2018 10:08AM
Quote
shadowphile
"For some reason 3D Printers can not print faster than the Hot-end can heat filament!!!"

was that sarcasm? smiling smiley

This was not sarcasm!

Was try to grasp the ideal and put my head around how to add an algorithm to speed up my prints when I am printing near max speed for my 3D Printer now.

I even pondered the fact that my hot-end heating block is horizontal thereby giving me a vertical heating distance of 10mm.

If I was to change the hot-end heating block with a vertical block that would give me a vertical heating distance of 20mm.


Though I still have only a 40watt heater to heat the hot-end block, would this increase the filament melting and allow for a faster print speed.
Thought about adding dual heaters with a larger hot-end heater block, would this increase the filament melting and allow for a faster print speed.

Quote
shadowphile
Higher wattage heaters are more likely to burn the plastic before the temp sensor can tell the heater to back off, or at best minimize temp variations in the heater block.)

Continuing on the dual heater ideal.
Even though if I have dual heaters. Still only need to heat the hot-end to 200C for PLA 230C for ABS ect... using larger heat block to compensate for the dual 40watt heaters.

You see no sarcasm!
Most of us who really wanted to speed up the Printing process have adjusted everything to allow for near max speed now, the real question is can we go even faster?

Edited 2 time(s). Last edit at 06/17/2018 10:38AM by Roberts_Clif.
Re: algorithm to speed up prints?
June 17, 2018 05:56PM
You're reply is partially correct but still does not address WHY there a limitations to the current approach of flowing plastic through a heater block.
Please be patient while I try to explain some of the physics that causes these limitations. I will exaggerate some things to make the issues more clear.
This is pretty long because I am trying to explain the thermal physics without using any equations. Describing how heat-flow works using the fairly simple equations would make this email very very short!



-It's common to get heat and temperature and watts confused so let me describe them exactly but using metaphors instead of equations:

Start with one gram of water (metric heat properties are usually based on water)
HEAT is a quantity. You can put a certain amount of heat into something. A quantity of Heat is described in JOULES.
TEMPERATURE is a value only, you can't collect it.
(a cup of 50 deg water holds more heat than a teaspoon of 50 deg water, even though they are at the same temperature.)
More heat inside something will make it hotter. How MUCH hotter depends totally on what the material is.)
CONDUCTIVITY describes how easily heat travels through a material. This depends on both the shape of the material and the property of the material itself.
POWER describes how FAST heat is moving from one place to another, not how much heat you have, or the temperature.
If the material is highly conductive, then heating one end of a cube of material will allow that heat to flow faster from one end to the other. Since the heat won't collect in one spot inside the cube too much, the temperature of that spot inside the cube of material won't get a lot higher hotter than other parts of the cube. You can heat the cube up faster with more even temperature throughout the material (because as described above, the temperature of a location in the material depends on how much heat it contains). Thermal power is measured in WATTS, which is actually JOULES per SECOND.
There is another important thermal property of a material that is called the heat-capacity, or how much heat you have to push into a material to make it's temperature rise. I will try to avoid using this property to simplify the discussion. The heat capacity differs a lot between materials.)

Plastic has LOW thermal conductivity, aluminum has HIGH conductivity. You could call the plastic an insulator also.

Why the plastic filament is the bottleneck and not the power of the heater:
-Lets say you want to melt a kilo of plastic as fast as you can then pour it down a funnel.
-You put a cylinder of that amount of plastic into a steel pot and put it on the stove.
-You turn on the stove burner and wait.
-The bottom of the plastic will start to melt first because the heat can't flow through the plastic fast enough to melt the top of the plastic at the same time.
-So you turn up the power of the stove burner to try to make it melt faster.
-More power from the burner makes it hotter. Now the bottom of the plastic melts faster but it also gets hotter, perhaps TOO hot, albeit the top of the plastic block will eventually melt faster. But you cannot heat the plastic higher than a certain temperature without destroying it. So turning up the stove burner to a higher power is not allowed. That is why more heaters won't help you melt the plastic faster because the temperature sensor and control electronics makes sure the heater block's average power output is forced to remain the same, even though you put a higher power heater in. You can actually have too powerful of a heater in your block: Put a 100W heater in your block (not really, that would be dangerous!) then do a heater calibration in your firmware, and it will tell you that your heater rating is TOO high. Why too high? More physics again:

-The conductivity of the aluminum block is high, MUCH higher than the plastic. That means the whole heater block reaches the same temperature fast and why you don't see popular designs with two heaters on each side of the heater block. The heater block already wraps all the way around the melt zone inside the heater block at the same temperature (because it is so conductive to heat) so there is nowhere to push more heat into the plastic from different sides. Since you can't turn up the temperature of the block, then there is no way physically to force more heat into the plastic.
(also, a really powerful heater WILL cause some parts of the aluminum block to get hotter than others, which could burn parts of the plastic even though the average power from the heater stays the same. Remember that the heater is never on continuously except when it is warming up.

-Back to plastic on the stove. When you turn up the stove burner, it basically CAN push more heat into the plastic faster but it can only do that by raising the temperature, which you are not allowed to do. If the stove burner is too weak, then at maximum power it can't even get the temperature high enough to melt the plastic at all because the heat that is going into the plastic has time to flow further into the plastic and escape out the sides of the pot. If the heat can't collect somewhere, the temperature won't rise.

A steak might be a better metaphor to describe trying to specifically melt a filament. The reason you can't cook a steak in one minute on a super hot pan is because you will burn the outside of the steak but it will still be raw in the middle. (although some people like that kind of steak). To cook the whole steak more evenly you have to slow down how fast the heat is going into the steak, to allow it to collect more in the middle so the middle of the steak is closer to the same temperature as the outside. So you need a LOWER POWER burner to cook the whole steak more evenly without burning it, but it WILL take longer.
-There is only one way to melt more plastic (or cook 8 oz of steak) faster, and that is by finding a way to push more heat IN PARALLEL into the steak. You can slice the steak into a bunch of thin slices and spread those slices around the surface of the pan. What you did was expose more SURFACE AREA of the same amount of steak so you have more ways to get more heat into the steak faster. But a block of metal all at the same temperature surrounding a tube containing plastic has no other way to touch more of the surface of the plastic.
Even though the Volcano has four times the flowrate maximum than the standard V6 heater block, it still only comes with a 30 watt heater! The longer block only works because the plastic has more time for the heat to soak into the core of the filament and melt it completely before it reaches the nozzle.



Based on the above, here are some interesting facts:
-Some plastics have less resistance to the flow of heat than others, so you can push more heat into ALL of the plastic faster without any of the plastic getting too hot. In this situation you could push the plastic through faster, for same size heater.

-Some filaments have metal powder, so those should conduct heat better than pure plastic and theoretically allow a faster flowrate for the same size heater.

-In the opposite direction, any filament filled with insulating materials, wood for example, may only allow a slower flowrate. Again it depends on the specific thermal properties of the wood vs the plastic to know how the wood changes the maximum allowed flowrate.

-To make life more fun, you can't just use a longer block for all printing because the longer a plastic is at a high temperature, the more it degrades and looses strength. So you can ONLY do fast flow-rates. Plus, wood-filled plastics can burn the wood if it stays in the heater too long, and cause bad clogging.

-How does 3mm filament perform vs 1.75mm? For one inch of filament it takes more heat to melt the 3mm than the 1.75mm, because there is more total plastic in one inch, AND the thinner filament can heat up faster inside the core of the filament.
However, you have to push the thinner filament faster into the heater block for given nozzle flowrate and that requires more power to melt the filament in time before it reaches the nozzle. The thicker filament, although it takes more heat to melt one inch, can push into the heater block more slowly for the same nozzle flowrate and THAT allows the heater more time to push more heat into the plastic until it melts. In the end the two tend to cancel out and that is why you still see both sizes of filaments sold. I haven't calculated the exact numbers to confirm that but the reasons for having thick or thin filaments has to more to do with breakage and other non-heat related issues. 1.75mm seems to be winning the size race but you will notice that the e3d V6 heater block specifies a maximum flowrate of about 11 mm^2 per second, WITHOUT caring which size filament is being used. I started with 3mm but changed when more and more suppliers only offered 1.75 filament. The thinner is easier to bend without breaking.

Thermal engineering can get much more complicated than this but the standard method of melting a tube of plastic on it's way to the nozzle is fairly simple to calculate well.



What CAN be done to increase flowrate other than a longer block?
-Maybe microwaves. Only some plastics would react though, but since microwaves can penetrate more deeply than regular heat, you could heat the whole filament inside at once. In fact, when you need to heat up a block of steel to white hot, you can't do it fast enough with any size of blow-torch, but an induction heater, which puts the steel inside a super strong electric field, can heat a wrench to white-hot in seconds, pretty cool to watch! (but in my example it required 5000 watts of electric power from the wall).
-Expose more plastic to the heater than just a length of filament. You could split the filament into multiple thin filaments and have them flow separately through the heater block, then collect again just before the nozzle. This would potentially require a bigger heater because know the plastic isn't the bottleneck, the maximum power available is. But imagine trying to clean THAT block, blegh.

This is already way too long and you already may know a lot of this, but I can't tell except based on your comments.
BTW, I took shortcuts to emphasize the principles so please somebody else don't jump in with critiques unless I'm totally wrong, or point out things I didn't mention more like heat-capacity or the energy required for phase-changes.

(how did we get here from my original post about an algorithm that basically models the shaking of you printer in order to stop the shaking of the nozzle?)

Edited 1 time(s). Last edit at 06/17/2018 06:05PM by shadowphile.
Re: algorithm to speed up prints?
June 18, 2018 05:02AM
Quote
shadowphile
You're reply is partially correct but still does not address WHY there a limitations to the current approach of flowing plastic through a heater block.
Please be patient while I try to explain some of the physics that causes these limitations. I will exaggerate some things to make the issues more clear.
This is pretty long because I am trying to explain the thermal physics without using any equations. Describing how heat-flow works using the fairly simple equations would make this email very very short!
...

That's a great explanation! Now I won't be able to stop thinking about 3D printing even when barbecuing steaks.

Quote
shadowphile
What CAN be done to increase flowrate other than a longer block?
...
-Expose more plastic to the heater than just a length of filament. You could split the filament into multiple thin filaments and have them flow separately through the heater block, then collect again just before the nozzle. This would potentially require a bigger heater because know the plastic isn't the bottleneck, the maximum power available is.

See [somei3deas.wordpress.com] for something similar.

Edited 1 time(s). Last edit at 06/18/2018 05:03AM by dc42.



Large delta printer [miscsolutions.wordpress.com], Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].
Re: algorithm to speed up prints?
October 01, 2018 05:58PM
Re: algorithm to speed up prints?
October 01, 2018 11:40PM
Quote
shadowphile
case in point!
super-fast printing PLA


Fake news. Is not PLA, but a special filament pla based
Re: algorithm to speed up prints?
October 02, 2018 03:40PM
'Fake news' is such a catch-all abused term and doesn't seem to me to apply here.
Anyway, my only point was as a follow up to my previous posts about the inherent limits of print speed caused by the thermal conductivity of the filament. Diamond is an excellent conductor of heat and so their invention is entirely reasonable as a (technical) solution to faster print speeds. And the product is for sale so it's not Kickstarter vaporware.
Re: algorithm to speed up prints?
October 02, 2018 05:12PM
Could be true but most of us will never pay 81 dollars for 1kg of high speed PLA filament, with no viable benefit from this new PLA filament.
Doubled down doubt that most hobby 3D Printers will reach this 500mm/s print speed with any viable 3D Prints.

Edited 4 time(s). Last edit at 10/02/2018 07:15PM by Roberts_Clif.


Electronics Technician / Computer Programmer
Re: algorithm to speed up prints?
October 02, 2018 09:35PM
I wish they gave some specs. They talk about improved material and printing properties but with no numbers, grrrr! Carbodeon actually makes the nano-diamond stuff; the filament is just one application. They are also looking at finishes, paints, lubricants, etc. If the improved properties are true, even marginally, and the price drops, it could attract a following. Everybody wants to be your go-to filament supplier.
Sorry, only registered users may post in this forum.

Click here to login