Only the extruder needs to be accurate

Maybe this is the wrong assumption but if the moving parts to direct the extruder nozzle was feather light it would allow more speed without affecting accuracy.

If you used a core XY set up and just have the extruder nozzle, belts, and rails move would you have really accurate prints with no artifacts from vibration or at least to a minimum and print at super speeds... If...

And here is the idea I didn't see anyone ever try, have the fans, blower, feeder, and their supports, etc. follow the print head but without any contact to the extruder nozzle and on their own motors and rails to minimize artifacts appearing on the prints caused by snap back motion and vibration. Because the other parts don't need to be moving so accurately they could be on a suspension and the rails could be on rubber mounts, they could even have a counterweight to neutralize the vibration.

It would be a like a print head assistant.

Thoughts?

Its a novel approach and theoretically it would help. This is where things like flying extruders and piped air/water for cooling things are being used and developed to keep the mass of moving parts as low as possible. The component of the moving mass on a corexy that needs to be lightened IMO is the x rods/rail which are a very heavy part of the system. This is where deltas come out smiling, the rods are carbon fibre, the total moving mass is very low, get rid of the fans and try pumped air and you have a very low mass/inertia system. But you run into the limit of how fast you can extrude, my big delta can move faster than the extruder/hotend can keep up with. If you want to print fast then lighten everything up as much as you can, but you will make the system less rigid eventually and there is your limit.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Vibration is only part of the problem, and the majority of well built machines are stiff enough to not have such vibration. Flex shaft extruders can help on cube style machines and deltas, where flying extruders are also an option. The amount of weight in these arragements is not enough on its own to cause the acceleration ringing. It is fine acceleration mismatches and actual integrity of stepper control pulses.

---

Flex3Drive.com

To move the assembly that follows the nozzle you'd need a separate XY gantry. Maybe Z could be coupled.
There is the "tripteron" thread that shows a possible light and easy possibility to implement that. It lacks of accuracy and tends to wobble, but in this case we don't care.



All you need is separate carriers on the rails which are coupled with the nozzle gantry in a vibration damping way. Maybe the arms are dampening enough already?

Edited 3 time(s). Last edit at 02/04/2017 02:32AM by o_lampe.

Quote
o_lampe
To move the assembly that follows the nozzle you'd need a separate XY gantry. Maybe Z could be coupled.
There is the "tripteron" thread that shows a possible light and easy possibility to implement that. It lacks of accuracy and tends to wobble, but in this case we don't care.

xQTTP9q.png

All you need is separate carriers on the rails which are coupled with the nozzle gantry in a vibration damping way. Maybe the arms are dampening enough already?

Ideally you would want an identical layout so that both systems would have the same control circuits.

Quote
aardvark
Maybe this is the wrong assumption but if the moving parts to direct the extruder nozzle was feather light it would allow more speed without affecting accuracy.

This is actualy a fundamental rule when designing printers. There is however always a balance of how much effort/money you are willing to spend.
It is just good practice, and way to often simply ignored, to not make moving parts more massiv than is realy neccesary. Beside mass levers are one of the most often overlook issues when it comes to vibrations and imprecision due to acceleration.
Setups like the one you have proposed, flying extruder, have been build and they certainly help, but they are expensive since they are a lot more complicated and need more parts.
Stuff like bowden extruders were invented to remove mass from the printhead, however, they come with other disadvantages.
Personaly i try to reduce the mass of the moving extruder as much as possible, keep levers short, frames stiff and most important i try to not over-define the bearings and bushings, one of the most often made mistakes that lead to z wobble, loss of steps and to hot motors.
All in all it is just very basic mechanical engineering.

---

[www.bonkers.de]
[merlin-hotend.de]
[www.hackerspace-ffm.de]

A flying extruder as implemented on many deltas, is very cheap and easy to do, and extremely successful. I'm about to try a version with RC car type shock absorbers instead of elastic, to dampen the movement when printing at high speed. In reality its a bracket, some elastic and the extruder and bowden cable that you were using already. As for the system proposed above, its an interesting idea, but what about the principle of keeping it to as simple as possible with as few moving parts? A complete second set of mechanicals paralleling the hotend is adding layers of complexity, to solve a problem which by much simpler methods (flying extruder, cable driven extruder) has been solved already.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

One problem you may have with two gantries, one for the hot end and one for the assistant, is that if one does not follow the other, for whatever reason at all, you've just messed up two gantries. This is a great example of KISS. Why have an overly complex dual gantry system to lower weight, when you could have just used well designed lightweight components to begin with?

Quote
aardvark

Quote
o_lampe
To move the assembly that follows the nozzle you'd need a separate XY gantry. Maybe Z could be coupled.
There is the "tripteron" thread that shows a possible light and easy possibility to implement that. It lacks of accuracy and tends to wobble, but in this case we don't care.

xQTTP9q.png

All you need is separate carriers on the rails which are coupled with the nozzle gantry in a vibration damping way. Maybe the arms are dampening enough already?

Ideally you would want an identical layout so that both systems would have the same control circuits.

In this case, you wouldn't need a second control circuit, just make it a passive follower. The Tripteron kinematic is linear, so once the effector is aligned with the nozzle, it will stay there ( plus/minus above mentioned wobbling and shaking )
Of course the overall weight the stepper has to move grows, but it is decoupled from the nozzle carrier. That was the goal, right?

The idea I have could be as simple as a center suspended assembly having the assembly's weight eliminated that way.
The print head has a small cap that has a sponge inside to eliminate vibration transfer.


Edited 1 time(s). Last edit at 02/05/2017 08:44AM by aardvark.

With that you eliminate the weight, but not the mass! It is still the printhead that would have to accelerate and decelerate the cooling assembly, the sponge will dampen the action, but it does not eliminate the forces needed. You will need to give the cooling assembly an own drive system.

---

[www.bonkers.de]
[merlin-hotend.de]
[www.hackerspace-ffm.de]

Quote
Srek
With that you eliminate the weight, but not the mass! It is still the printhead that would have to accelerate and decelerate the cooling assembly, the sponge will dampen the action, but it does not eliminate the forces needed. You will need to give the cooling assembly an own drive system.

It can be pulled by belts to match the motion but if it fails the printhead will take over as a redundancy to prevent total failure.

I can see the interest in this as an academic discussion. But surely if the idea is to get the ancillary items (fans,sensors extruder) off the moving mass of the print head to improve quality (and speed?) if you then put them on a secondary mechanical system, which has to follow the print head, albeit with some physical connection in case of a failure, you still need to accelerate this system and move it about. So instead of having one set of motors with x amount of torque, you now have 2 sets of motors possibly with more total torque say 1.5-2x compared to before (you might need less powerful motors for your secondary mechanics).

This solves the problem as you now have some disconnection between hotend and ancillaries and more motor power, but at the expense of almost doubling the complexity. So why not just use larger motors, and better stepper drivers in the original system which has all the ancillary gear on the printhead, Nema 23's instead of nema 17's would enable you to move that mass faster and more accurately with less motion artefacts.

Edited 1 time(s). Last edit at 02/06/2017 09:39AM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

More torque means more vibration when reversing direction or accelerating requiring more weight of the frame or go slower to counteract.

A second less accurate dampened and padded from noise and vibration non accurate motion makes the rig quieter and the print head more nimble. I'm saving up for a core XY 3D printer and I intend to add this device and report the results.

I'm still investigating which machine or kit would work best for my needs.

Another option that you might not have considered is the way that lasers inside CD ROM drives work. Basically you have a big, heavy, inaccurate drive system with motors, worm drives etc, that moves the carriage around. Obviously this is not nearly accurate enough to position a laser right on a row of pits as the CD is spinning. So what they do is they have a secondary movement system that moves the lens back and forth, with an extremely fast and accurate feedback loop, which takes all the inaccuracy of the larger control system out of the equation.

Back to 3D printing land, you could put a small, fast X/Y positioning system on the print head, which would compensate for any overshoot that the main print head carriage has. Then you attach all your heavy fans and extruder to the main carriage, and just have the nozzle / heatsink on the secondary accurate positioning stage.


It might work.

@nebbian
Interesting point! It sounds like it needs a closed loop PID system to tell the nozzle where it is ( and where it should be )
It reminds me of multicopter controllers with the infamous MPU-6050 6DOF sensors that are used to keep the drone stable.
I have one laying around, I always wanted to put on the effector of my delta to read out any tilt or wobbling/overshooting.

I was considering a smaller more accurate positioning on the printhead but wasn't sure if there would be software available to support such a setup.

Additional problem, if it had the range of printing 25mm by 25mm on the sub-gantry and your model is 30mm how would it motion smoothly?

The way I'd do it is to have the normal gantry connected via the normal steppers, controlled by the normal controller. Then the sub-gantry would be independently controlled by a small microprocessor (arduino nano or smaller) located on the sub-gantry, and position itself via small, fast servos/geared steppers. The only purpose of that gantry would be to take out any overshoot/ringing. It would only have to have a range of a mm or so. You could do it using a gyro/acceleration board with some tuning required, or you could somehow get some accurate position information and use that instead. I'm not sure how to get accurate position information. Maybe some encoder strips?

Another approach would be to add some PID factors to the firmware. This would mean no extra hardware required at all, but again would require some tuning based on what the resonant frequency of the print head was (dependant on print head mass, belt tension, gantry mass etc). This might actually be the best approach. It would mean being able to use higher speeds/acceleration, but is more of a dead reckoning approach -- there's no feedback to take care of inaccuracy.

Why bother with any of this? It is already possible to build machines that can accurately and reliably move the print head faster than one can melt filament - even with a Volcano hot end. What is the point in building a machine capable of extreme print speeds when, because of the laws of physics you need to allow enough time for filament to melt? Just my twopence worth..........

Always the voice of reason!

Quote
deckingman
Why bother with any of this? It is already possible to build machines that can accurately and reliably move the print head faster than one can melt filament - even with a Volcano hot end. What is the point in building a machine capable of extreme print speeds when, because of the laws of physics you need to allow enough time for filament to melt? Just my twopence worth..........

What machine is the fastest and makes the most accurate prints?
How much does it cost VS modding a slower one to preform faster?

The fastest 3D printer The D3D.ONE-EVO @ 450mm/sec (reliably)
$1800

I'm just proposing a simpler cheaper way of getting to those speeds.

[youtu.be]

Quote
aardvark
The fastest 3D printer The D3D.ONE-EVO @ 450mm/sec (reliably)
$1800

I'm just proposing a simpler cheaper way of getting to those speeds.

[youtu.be]

Is 'reliable' your only criteria for a fast printer? I only ask as I doubt it makes high quality prints at that speed. It uses the same mechanism as the ultimaker but they do not make a point of printing at such high speeds.

So there is the next point, if you want a high speed ultimaker style mechanism, why not make one? No need for complex extra gantries following each other. I think you will find the budget for making 2 moving heads work well will need more unique parts and drive cost up much more then a nicely designed minimalist lightweight gantry will.

... did some "high-speed"-testing with one of my belt-driven machines -- it has a pretty sturdy frame and can move with up to 800mm/s, but will start to wobble because of belts elasticity.

Here a video of the "fast" moves - [vimeo.com]

And here another test with much higher accuracy and a spindel-driven machine - [vimeo.com]

With the belt driven type I'll expect best accuracy at around 0,05mm (50 microns) with speeds below 200mm/s -- while the spindel driven type will give something like 5 microns at speeds of 50mm/s ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

I think there is nothing wrong with the quest for speed, we do need FDM 3D printers to print faster (with good quality) if we want them to progress from a prototyping/hobby tool to a mainstream consumer device (or limited-run manufacturing tool). Retail 2D printers, print in amazing quality at high speed. I'd say this was done by using cheap components made accurate by clever firmware, and inks which dry/spray exactly as needed for fast printing. You can argue about the business model and cost of ink, but that's not really the debate here, they work very well for very low cost.

The issue with FDM is that the plastic we are laying down has to be melted and deposited accurately at speed, There are currently limits to how much you can melt in a given time, sure we can design heater blocks which can melt more plastic, but they radiate a lot of heat and spoil the print beneath them. We could use materials that are easier to melt at a lower temperature, but the objects we make from these materials will have poor qualities. Think PLA versus ABS.

I'd love to see if the system proposed here works or helps, or is even practical, it will be useful information if it fails, at least we would know its a route not worth pursuing. I've already made my opinion clear about whether I think it will work, but ultimately 50% of predictions are probably wrong and so not worth a damn.

Surely what we need is to be thinking like the 2D printer, multiple nozzles, even just for infill this would save a lot of time.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Quote
DjDemonD


.........................Surely what we need is to be thinking like the 2D printer, multiple nozzles, even just for infill this would save a lot of time.

Spray the molten plastic in tiny droplets through multiple nozzles - now your talking. Like an inkjet printer but with molten plastic instead of ink. Might need a new type of plastic though.........

That's what a polyjet is correct? Okay how do we make one of those?

Edited 1 time(s). Last edit at 02/13/2017 10:22AM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

... the polyjet/objet printers uses UV-curing polymers.

I'm developing around pastedispenser applications with single-nozzle heads - a pretty expensive thingy ... around 8k€ with controller

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

Quote
DjDemonD
That's what a polyjet is correct? Okay how do we make one of those?

Damm. You mean someone else has already thought of it?