Welcome! Log In Create A New Profile

Advanced

Better slicer/controller

Posted by newbob 
Better slicer/controller
April 21, 2017 08:13AM
I'm new to 3D printing. I was hoping that I could print fast but that's far from reality.

I wonder if there's a slicer/controller that models optimal tool path, acceleration/deceleration and speed based on weight/stiffness of the gantry while limiting max seed/acceleration based plastic/nozzle/temperature used?

Right now everything seems so manual and I am beginning to understand why people own so many 3d printers - each one is probably tuned for specific job.

At the least, is there a controller that uses accelerometer/strain gauge attached to the frame to know when gantry is exceeding it's stiffness or to detect resonance and decease acceleration automatically?
Re: Better slicer/controller
April 21, 2017 08:26AM
Nope. not that I know of. You just have to manually dial the settings in for your printer and go from there.
Re: Better slicer/controller
April 21, 2017 12:59PM
Hi Newbob, that's not a bad idea that... It's quite doable, could be done with Piezo discs bonded to the frame. Might take a bit of calibration, we were also thinking of reading analogue data from piezo z probes to detect rough surfaces on previous layers and generate slow down in feedrate.

The counter argument is that since 3d printing isn't really done with dynamically varying settings within one print, once you know your machine's capabilities you just print with settings a little way below that point.

3d printers currently either lack (or don't need, depending on your point of view) any form of closed loop control of motion let alone print quality.

Edited 1 time(s). Last edit at 04/21/2017 01:02PM by DjDemonD.


Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions
Re: Better slicer/controller
April 21, 2017 04:21PM
Hi DjDemonD

I have a pretty lightweight printer so I see plenty of shake when printing at higher speeds. Every moving part has inertia. Every time hot-end accelerates/decelerates it's motion will be counteracted by the frame at different angles and frequency. Cartesian printers Y axis usually have higher innertia than X axis. So even the same print oriented differently will generate different forces depending which axis is used more. All forces could be calculated in slicer (if there was one) or measured realtime by sensors. Even if software cannot act on it real-time, logs could give idea why print failed and prevent one from doing same, bad, thing over and over again.

I also though that I wanted a way to measure filament pressure and federate together to counteract varying sizes and composition of filament. It looks like someone already did that, in part, in 2013 but it did not catch on (I think it's a great idea and will definitively do that one for my printer). From reading piezo thread I see that piezo sensors can sense a rate of change. Maybe combined with a feed rate that would be enough to detect filament issues and would be cheaper too.

Edited 2 time(s). Last edit at 04/21/2017 04:42PM by newbob.
Re: Better slicer/controller
April 21, 2017 04:36PM
I think this sort of fresh thinking is what can lead to change. Though my understanding is that measuring actual filament flow through a nozzle has not been done.

I did think of a way to autocalibrate an extruder and that is to extrude a new filament onto a glass portion of the bed under which an optical scanner type sensor would be placed that would measure the filament (presuming perfect z height by the next generation of z probes) and calculate the volume extruded and then adjust the esteps, but this is more along the lines of ways of making a zero config consumer printer which can take any filament not cartridges of expensive super high tolerance stuff.

But that's not what you are talking about. I think the challenge is that once you get more into it you are conditioned by it, and are more likely to increment change than go for a big paradigm shift.

Some sort of frame resonance sensor would be interesting to try, the data you get would surely be useful perhaps in unexpected ways. But (and here's the conditioning speaking) I'm not sure what you would use it for practically, certain moves could be slowed down, others sped ip as there would be feedback to determine which moves reduced print quality and which didn't?

Perhaps measuring pressure at the nozzle is possible, but I'm not sure if that alone would be enough or how that would translate into print quality.

Edited 1 time(s). Last edit at 04/21/2017 04:39PM by DjDemonD.


Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions
Re: Better slicer/controller
April 21, 2017 04:55PM
The cost/benefit tradeoff just doesn't make sense. There is no point in adding a bunch of wiring, sensors, and software when you could just spend $20 building a more rigid frame to begin with.

Trying to fix low-quality physical systems in software sounds smart, but is nearly impossible in practice. Cheap mechanical systems do not fail in predictable or linear ways. Sensor data is too noisy to be useful. You cannot generate a mathematical model of the system because it changes day to day as parts loosen or the ambient temperature/humidity shifts.

The clever stuff only gets used on systems that are already near-perfect because:
a) The extreme cost and complexity can be justified.
b) The system behaves well enough that mathematical models can effectively predict and control it.


Quote

So even the same print oriented differently will generate different forces depending which axis is used more.

Acceleration and jerk are set in the printer config file, the toolpath does not have any influence on the maximum forces your printer will experience.

Edited 1 time(s). Last edit at 04/21/2017 04:56PM by 691175002.
Re: Better slicer/controller
April 21, 2017 05:40PM
Quote
691175002
The cost/benefit tradeoff just doesn't make sense. There is no point in adding a bunch of wiring, sensors, and software when you could just spend $20 building a more rigid frame to begin with.

Trying to fix low-quality physical systems in software sounds smart, but is nearly impossible in practice. Cheap mechanical systems do not fail in predictable or linear ways. Sensor data is too noisy to be useful. You cannot generate a mathematical model of the system because it changes day to day as parts loosen or the ambient temperature/humidity shifts.

The clever stuff only gets used on systems that are already near-perfect because:
a) The extreme cost and complexity can be justified.
b) The system behaves well enough that mathematical models can effectively predict and control it.


Quote

So even the same print oriented differently will generate different forces depending which axis is used more.

Acceleration and jerk are set in the printer config file, the tool-path does not have any influence on the maximum forces your printer will experience.

Even if it was the case it does not change the fact that orientation of the model will yield different speed/quality. Btw, at least Craftware appears to be capable of generating it's own acceleration and deceleration g-codes (I have not tried but there's an option).

I think calculating 3d printer forces should be possible. What you say maybe true for CNC machines with hard to calculate forces resulting from substractive process. 3d printers on the other hand have to contend with occasional bump into half molten plastic - rest of 3d printer motion is well known as it's generated by xyze axes.

Edited 1 time(s). Last edit at 04/21/2017 05:41PM by newbob.
Re: Better slicer/controller
April 21, 2017 05:55PM
But then this relates to the thread on ringing and frames, build a solid, heavy frame and populate it with lightweight but stiff moving parts and the limiting factor to how fast you can go is how fast you can melt plastic not how much your frame bends or vibrates.

But that's not what most printers look like, the frames are too bendy and the moving parts are often a little too heavy.

Edited 1 time(s). Last edit at 04/21/2017 05:59PM by DjDemonD.


Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions
Re: Better slicer/controller
April 21, 2017 06:13PM
Quote
DjDemonD
But then this relates to the thread on ringing and frames, build a solid, heavy frame and populate it with lightweight but stiff moving parts and the limiting factor to how fast you can go is how fast you can melt plastic not how much your frame bends or vibrates.

Agreed to a certain point (I still think you can print overall faster when accounting for acceleration/deceleration forces). Right now extrusion is holding printers down more than their construction. This is why I think filament feed sensor and extruder force meter would help if not printing faster than at least to determine which setup/settings are the most optimal.
Sorry, only registered users may post in this forum.

Click here to login