Even higher speed printing?

Share any mods you have made and the awesome results.

Moderator: Elijah

ccrome
Posts: 410
Joined: Thu Jul 25, 2013 5:37 pm

Even higher speed printing?

Postby ccrome » Tue Aug 06, 2013 12:51 pm

Hi all,
I'm waiting for my Type A Machine, so in the meantime I'm thinking about even higher speed printing, perhaps for production purposes. I have a few questions about what the current limiting factors are, and what can be done to speed up the printing process.

I'm a DSP engineer by trade and I can't help but think that things can be improved by using modern control theory to drive all the axes (including the hot end). I have not looked to see what is currently used, but by guess is that it's simply stepper motors with no feedback.

Here are my thoughts anyway:

Hot end: I imagine that if you send filament through the hot end too fast, it will cool down, and the thermocouple feedback loop will always lag the filament. Perhaps this could be improved with a feed-forward component to the control equation so as to maintain a constant plastic exit temperature regardless of extruder speed. This would require a little science to figure out the exact parameters, but may not be too difficult.

Steppers: Why does everybody use steppers as opposed to DC motors with linear encoders and feedback? I've seen threads on this question before, but I've never seen a cogent answer other than, "it's hard". I've done control systems before and know that they can perform very well, but I haven't done one just like this. My suspicion is that a feedback control system with a linear encoder feedback will allow a smaller & cheaper motor to drive the system at higher speeds, at the expense of adding a linear codestrip and linear encoder (just like every $50 inkjet printer in the world -- they can't be too expensive). Of course, there will be some software too ;-). Are the steppers a limiting factor in the system design performance?

Stiffness: I assume that machine stiffness would be a factor when accelerating the print head at very high rates. Accelerating the print head with very high rates will surely cause vibrations that show up in the print. This can be mitigated by lowering the maximum acceleration, but on parts with lots of turns, the print head may never hit the maximum speed. In principle, modern control theory can help here too, by predicting how the vibrations will happen, and compensating for them. However, this is probably a much more difficult control problem than the ones above, and likely more easily solved by making the machine stiffer & heavier. But it's worth thinking about anyway.


Anyway, I'd love to experiment with really increasing printing speed, but I need to understand what the current limitations are to know where to start attacking the problem :-)

Any thoughts?

Thanks,
-Caleb

LaserGnomes
Posts: 1058
Joined: Thu Dec 06, 2012 10:15 pm

Re: Even higher speed printing?

Postby LaserGnomes » Tue Aug 06, 2013 1:10 pm

Having raced our printer to breaking point... there are few factors.

Hot End.
You can compensate the hotend through better variables and "tuning" (search "is your printer into swinging")to heat the material at greater feed rates. There is a top limit.. have yet to reach it. It could be overcame by a longer shaft allowing for more room to apply heat.

Motors.
No idea vs DC motors. Cheap etc.. not a real concern. The motors now are not expensive. And if you are trying to build a race car you don't put a cheap engine in it. On that note.. I would look for good DC motors vs cheap ones.

Case.
Speed becomes an issue due to the weight of the extruder assembly. If you could find a way to move the weight off the assembly higher speeds would be no problem. We've added weights to the bottom of our case to reduce vibrations in general.


Real World.
Moving too fast is going to cause other problems (which may or may not be addressable).
- Whipping the filament off the roll
- Breaking the filament due to rapid back/forth motions in succession
- Vibrations of assembly hardware (screws, bolts, belts)
- Pulling laid lines away from placement (rounding corners etc)
- Excessive wear on the wiring harness with high speed small movements


Gl!

ccrome
Posts: 410
Joined: Thu Jul 25, 2013 5:37 pm

Re: Even higher speed printing?

Postby ccrome » Tue Aug 06, 2013 2:11 pm

Hi, thanks for the quick reply!

LaserGnomes wrote:
Hot End.
You can compensate the hotend through better variables and "tuning" (search "is your printer into swinging")to heat the material at greater feed rates. There is a top limit.. have yet to reach it. It could be overcame by a longer shaft allowing for more room to apply heat.

Good to know that it's currently not a problem.

LaserGnomes wrote:Motors.
No idea vs DC motors. Cheap etc.. not a real concern. The motors now are not expensive. And if you are trying to build a race car you don't put a cheap engine in it. On that note.. I would look for good DC motors vs cheap ones.

So, the point is not just saving money. The point is that with a DC motor, you can use a whole separate control system, that should be able to accelerate faster than an equivalent stepper. So, you can either make the system lighter by moving to a lighter motor, or alternatively, make the system faster by putting in a bigger motor. The power/weight ratio of a DC motor should be better than a stepper.

LaserGnomes wrote:Case.
Speed becomes an issue due to the weight of the extruder assembly. If you could find a way to move the weight off the assembly higher speeds would be no problem. We've added weights to the bottom of our case to reduce vibrations in general.

Heh, I guess there are some machines that hard mount the extruder to the case, and only carry the hot-end on the gantry. I guess that has other issues :-)

LaserGnomes wrote:Real World.
Moving too fast is going to cause other problems (which may or may not be addressable).
- Whipping the filament off the roll
- Breaking the filament due to rapid back/forth motions in succession
- Vibrations of assembly hardware (screws, bolts, belts)
- Pulling laid lines away from placement (rounding corners etc)
- Excessive wear on the wiring harness with high speed small movements

Interesting. Certainly sounds like some could be addressed -- I look forward to pushing the limits and finding out what the issues are ;-)

Thanks,
-Caleb

Elijah
Site Admin
Posts: 767
Joined: Wed Oct 10, 2012 6:33 pm

Re: Even higher speed printing?

Postby Elijah » Tue Aug 06, 2013 2:51 pm

First of all, thank you so much for your feedback! I love it when conversations like this start.

ccrome wrote:Stiffness: I assume that machine stiffness would be a factor when accelerating the print head at very high rates. Accelerating the print head with very high rates will surely cause vibrations that show up in the print. This can be mitigated by lowering the maximum acceleration, but on parts with lots of turns, the print head may never hit the maximum speed. In principle, modern control theory can help here too, by predicting how the vibrations will happen, and compensating for them. However, this is probably a much more difficult control problem than the ones above, and likely more easily solved by making the machine stiffer & heavier. But it's worth thinking about anyway.


Adjusting the acceleration profile is a good idea, though I will point out that the printer very rarely hits it's top speed in the current acceleration profile. Too steep an acceleration profile can cause prints to appear with wavy surfaces due to inertia. (Of course, in truth there's more to arcs and curves then that)

ccrome wrote:So, the point is not just saving money. The point is that with a DC motor, you can use a whole separate control system, that should be able to accelerate faster than an equivalent stepper. So, you can either make the system lighter by moving to a lighter motor, or alternatively, make the system faster by putting in a bigger motor. The power/weight ratio of a DC motor should be better than a stepper.
-Caleb


*Begin stream of consciousness*

Cost effective DC motors aren't nearly as precise from my understanding. At the moment, the Series 1 has a positional accuracy of a few cubic microns per microstep. I imagine you could add an encoder strip so that in the event of a skip, there might be a bit of failure protection. But it would be expensive to implement an entirely different control system. But the existing open-loop system is only troublesome for the reason of failure due to mechanical reliability issues. Also, anybody that has been involved with 3D printing for more than a year and a half may cringe at the thought of putting a DC motor on the extruder... (Not saying that you suggested that, but only that it's a less than good idea for reasons of precision and retraction.)

Also if you want more power, double the voltage going into the motors. Currently it's 12V. We're testing 24V. That's a 4x(?) increase in torque - which, apart from being less susceptible to skipping, means that you'll be able to print faster.

So now you're faster, but your hot end is cooling. We're working on hot ends that will be more capable of high speeds and have better heating systems.

So now your hot end is up to the task but the machine isn't stiff enough. I really like the idea of values to compensate for the rigidity of the machine. The next machine will be made of sheet metal, and that may help...
Elijah Post
Type A Machines

ccrome
Posts: 410
Joined: Thu Jul 25, 2013 5:37 pm

Re: Even higher speed printing?

Postby ccrome » Tue Aug 06, 2013 4:30 pm

Elijah wrote:Adjusting the acceleration profile is a good idea, though I will point out that the printer very rarely hits it's top speed in the current acceleration profile. Too steep an acceleration profile can cause prints to appear with wavy surfaces due to inertia. (Of course, in truth there's more to arcs and curves then that)

This is exactly the type of problem that a modern control system can, in principle, solve. I've seen this on the Cube that we have at work, and it's not beautiful. And of course with the Cube, there's no way to fix it. :evil:

Elijah wrote:Cost effective DC motors aren't nearly as precise from my understanding. At the moment, the Series 1 has a positional accuracy of a few cubic microns per microstep.

That's assuming no load conditions. As soon as a load is applied (static or dynamic), the stepper will be pulled out of it's nominal position, no? Not necessarily a full step out of position, but away from the nominal location. Also, that assumes that there is no stretch/flex/lash in the belt system. The assumption that position of the head == the motor step position is not necessarily true. This is where a codestrip/feedback system will help.

The codestrip will be assembled in in such a way as to measure the position of the head, not the position of the motor. That way, you can make sure the head gets to the position you're asking for, not just that the motor gets there.


Elijah wrote: I imagine you could add an encoder strip so that in the event of a skip, there might be a bit of failure protection. But it would be expensive to implement an entirely different control system.

An open-loop DC motor will definitely not be precise. In fact, it would be totally useless. The key is to couple a DC motor with an encoder. I'm thinking that the combination of DC motor+encoder should be as accurate, or even more accurate, than the steppers. About the expense, I'm not so sure. inkjet printers had to move away from steppers long ago because of all their problems. why shouldn't 3d printers do the same?

Elijah wrote:But the existing open-loop system is only troublesome for the reason of failure due to mechanical reliability issues.

Hmm, I wonder how much of the 'wavy lines at high speed' is due to the motor inertia bouncing against the spring constant of the stepper motor, vs. spring constants in the rest of the drive system. Either way, I think this can be reduced by a proper control system.

Elijah wrote:Also, anybody that has been involved with 3D printing for more than a year and a half may cringe at the thought of putting a DC motor on the extruder... (Not saying that you suggested that, but only that it's a less than good idea for reasons of precision and retraction.)

:-) Yes, that's exactly what I'm suggesting. Of course, not an open loop DC motor. Again, that would be completely useless. It would be a DC motor, coupled with a rotary encoder on the hobbed nut (or equivalent). If there is any gearing in the system, this would largely eliminate any issues with gear lash because the control system works on the *actual* position of the nut, not the assumed position of the nut based on the position of the motor.

Perhaps this has all been tried before and rejected. I just haven't seen it :-)

Elijah wrote:So now you're faster, but your hot end is cooling. We're working on hot ends that will be more capable of high speeds and have better heating systems.

Ah, a feed-forward control system may give your current hot end design some more life. You can model the heat transfer of the hot end using FEA, then when you know you'll be feeding more, you can drive a lot more power into the hot end to keep it hot enough. There may be a limit to what you can do here though. Keeping the hot-end mass low will be the key I think.

I wonder.....

Anyway, I look forward to getting a machine a playing around.
Thanks,
-Caleb


Return to “Mods”

Who is online

Users browsing this forum: No registered users and 1 guest