Mass Production Mendel Style Printer Questions

Hello to all!

My name is Ivan Kozlov, and after several months of prototypes, rework and calculations on the subject of commercial viability, I have decided the best, easiest and most reliable way to get my diverse and partially customizable airframes to the market is to hybridize FFF and aerospace compositing. To this end, I have designed an extremely large 3d printer in the style of the original mendel (z axis on bed, xy axis on gantry). I would greatly appreciate if any of you could look over the design, and offer criticism;

Firstly, specifications are as follows:

Build area: 1500mm x, 2500mm y, 500mm z

Core construction materials: Openbuilds C-beam for core supports, and the normal extrusions for most of the gantries, 8mm ACME leadscrews, steel wheels, custom routed aluminum plates, metal/composite sandwich board with sequential PCB heaters and PEI sheet.

The printer is designed to support a production line; therefore, there shall be 6 extruders at regular intervals, locked to the same leadscrew; therefore, in effect the printer is a thin rectangular build surface repeated 6 times. The extruders themselves are going to be bowden extruders with filament being supplied from a gantry atop the printer which moves approximately with the gantry on the Y axis. I intend to move the whole gantry on the Y axis with two leadscrews, but is one leadscrew and a beefy nema 17 stepper sufficient to move 700 grams or so quickly and precisely with a leadscrew on the X Axis, assuming that I utilize the anti backlash nuts?

The Z Axis is controlled with 4 stepper motors and leadscrews driving four vertical gantries which form the corners of the build plate; is this in any way feasible, or is there some simpler solution?

The second question pertains to the extruders; I wish to utilize a thick extruder such as the E3d volcano to form the infill, but a small 0.015mm extruder head to create fine lines; how do I prevent the heads from impacting the print?

According to my calculations, I'm going to need to run 2 magnetic probes, 21 stepper motors, 17 thermistors, 12 heaters, 5 pcb heaters, and 12 magnetic endstops; should I just give up and construct the hardware myself out of a Rasberry Pi V2, or is there some board that I do not know about that is this extensible?

I intend to print with modified XT_Copolymer, possibly impregnated with composite fibers; off the top of your various hats, what print speed would you estimate optimistically given the hardware and assuming that the filament poses no speed reductions? I'm aiming for 50mm/s for highly optimized prints, but that is probably quite silly.

What are your thoughts on this design? Should I pursue some other printer architecture?

Finally, my apologies if this has been asked or if I am not posting to the correct forum.

Edit; Whoops, I had wished to post this in reprappers; how could I move the thread?

Kind Regards,
Ivan Kozlov

Edited 1 time(s). Last edit at 09/20/2015 03:30AM by Charon.

1) 4 screws at the corners of the build plate are a bad idea. 3 points define a plane, 4 points define a potato chip. Any sort of minor misalignment will cause the whole thing to bind. I assume that you'll be using 4 motors to drive those 4 screws. Bad idea. Put pulleys on them and use a belt to drive 3 screws with one motor.

2) filaments with composite fibers are abrasive and will chew up the nozzles very quickly. Look into hardened nozzles or be prepared to swap nozzles a lot.

3) Have you done any calculations on how much this big machine is going to weigh and how much flexure to expect? 8mm lead screws seems small to me for lifting a bed as large and heavy as you going to use.

4) How are you going to switch between large and small nozzle extruders? Is this machine going to have six extruders or twelve?

5) It's very hard to look over the design without a drawing.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Priviet Ivan,

I agree with the_digital_dentist about the 3 points leveling bed,
but you can lift the bed from two points only.
IMHO do your machine with 1 motor by axis/function.
Anyways, I'm always very dubitative about ultra large 3D printers,
especialy with high quality expectations.
Just think about buying a 1500x2500mm buid plate
with an overall flatness within 0.1mm !
The machine's chassis will be extremely heavy,
and the machine very expensive.

Maybe a single robot arm will be more pragmatic for that dimensions.
What about to start from a second hand industrial robot arm ?.

++JM

0.015 mm nozzle? Might be hard to get enough pleasure to get anything out of it with an ordinary extruder. Also any reenforcement-fibers in the filament have to be smaller then the nozzle. Even if it is a 0.15 mm nozzle, do some testing before you start the build.
For a larger printer the XY-carriage system might be lighter than the build plate.
I would also recommend you to do some calculations what build times you will get. A square meter at 0.1mm layer hight will take close to 28 hours at 100mm/s (if my quick check is correct). The real print time will be longer with accelerations, decelerations and movements.
This is an industrial size machine so use industrial controllers.
Servo motors will do a much better job on a large machine, they are faster, stronger and more reliable then steppers.

Thank you for your responses; here is some more clarification:

@ The_digital_dentist:

Thank you for the correction in terms of the number of drive screws. You are entirely correct, especially at tolerances which are precise to this degree. However, I question the need to put a belt on the screws and drive them with a stepper: I was going to utilize the separate motors to change the orientation of the build plate, as part of the autonomous levelling sequence.

I shall likely utilize steel nozzles, and long reinforcement fibers.

The bed is going to be anchored to four gantries, which would form a corner of the bed; if the screws are vertical, they would not be likely to flex too much. The estimated weight of the bed is 5kg.

The machine is likely to have 12 extruders, which is where my question stems.

My apologies for not providing schematics, but I'm presently transferring the paper schematics to a 3d model, which I shall post here.

@ J-Max

Privet!

I was hoping to avoid having the machine be a cantilever on the Y axis , which is why I was going to utilize two motors.

I agree with your concerns; the core idea is that the printer is in essence 6 long rectangular printers, but without all the ancillary hardware. Which leads me to my question: assuming that I construct the build plate with a component of inductive metal, then would it not make sense to simply derive the Z height not only from known initialization endstops, but also continuously from two inductive probes at either end of the X axis, therefore correcting for any bed deviations while printing, in firmware?

The problem with robot arms would be that I would have to purchase 6 of them in order to have the production speed I was hoping for; is there some way to avoid this that I have missed?

@ SlowFoot

My apologies, I had wished to say 0.15mm nozzle!

I'm going to construct a small prototype based on the same concept before I construct the large machine, of course. In terms of build speed, the printer is likely to render small parts in large quantities rather than large parts in small quantities, so I'm not too worried about print speed; my calculations tell me that the limiting factor is going to be the later stages of the construction process.

Are there some industrial controllers which are going to be worth the money, and which are not over $250?

What servos did you have in mind in particular? A brushless motor, with a gearbox and an encoder?

Once again, thank you very much for your comments: I shall keep you all updated with pictures of the build.

Kind Regards,
Ivan Kozlov

Quote
Charon

My apologies, I had wished to say 0.15mm nozzle!

No problem I kind of figured that out.

Quote
Charon
I'm going to construct a small prototype based on the same concept before I construct the large machine, of course. In terms of build speed, the printer is likely to render small parts in large quantities rather than large parts in small quantities, so I'm not too worried about print speed; my calculations tell me that the limiting factor is going to be the later stages of the construction process.

Are there some industrial controllers which are going to be worth the money, and which are not over $250?

No don't think so, add another zero and you are in the ballpark.

Quote
Charon
What servos did you have in mind in particular? A brushless motor, with a gearbox and an encoder?

Frankly I think you will be much better of with a cluster of smaller (normal) printers then one gigantic. 20 kits of a good make will be less expensive, faster to set up, easier to maintain and give much better redundancy. Probably higher production too but a lot less impressive.

Hi again,

Quote
SlowFoot
Frankly I think you will be much better of with a cluster of smaller (normal) printers then one gigantic. 20 kits of a good make will be less expensive, faster to set up, easier to maintain and give much better redundancy. Probably higher production too but a lot less impressive.

I strongly agree !
OTOH, to have multiple printers alows you to
split mutliple works into different groups of machines.
And if a machine have a beakdown,
the production will continue on the others.
Multiple heads on a gantry means you cannot
have a different leveling for each head.

In my humble opinion, you must think again
about the whole production process.
Try to get the most flexible tool,
easy to maintain, easy to upgrade, easy to slod,
because evolutions goes quick in the actual world.

++JM

Very good points;

I was on the fence about the extremely large printer on purely kinematic grounds, but the other points are valid as well. What about a middle ground; a set of 750x750x500 printers with a small number of locked extruder heads? That way, the complexity need not increase terribly, but the individual system complexity would be substantially reduced, and redundancy would then be more feasible.

Another point I had was the print bed: would it be feasible to orient the bed to allow better overhang printing? Is there a slicer for software like this?

Another question I had was about the huge commercial low cost cartesian printers such as the Gigabot: can they print well, or are there severe issues?

Thank you very much for you advice: I'm probably going to scale the printers down, but increase their number.

Are there scalable or extendable reprap controllers on the market which are not worth more than the mechanical components of the printer?

As a side point, have any of you utilized colorfabb_XT? Have you got any thoughts or comparisons to other filaments?

And finally: what are your thoughts on SLA technology?

I shall certainly update this thread with pictures of the printers, when I get the parts!

Kind Regards,
Ivan Kozlov

Have you seen this project on kickstarter???

[www.kickstarter.com]

Thank you for that link; I shall certainly look it over.

One more question, if possible!

On the subject of CNC controllers, has anyone seen a controller which would be extendable to 15 stepper motors?

Kind Regards,
Ivan Kozlov

... for driving more motors in parallel you can multiply/multiplex the DIR and CLOCK pins ... for driving them separate you'll need all the pins, so a controller with 30 I/O-pins only for the motors (without ENABLE) and all the single pins for sensors, switches and other switching/sensing tasks ...

Count the available I/O-pins of an Arduino Mega or Due - should be enough

So - extending a CNC-controller for 15 motors is no big problem -- it's the software controlling them, what's the biggest hurdle

---

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

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

Thank you for the reply!

I concur on the subject of available pins; I've been looking over possible controllers, and I have two possibilities: the smoothieboard/Azteeg X5, or the beaglebone with the printer cape. It's difficult to say which is superior for the application, but the core issue pertains to the pins: does either of these controllers have enough GPIO pins to control two extra stepper motors, and how many pins would be left over? Would smoothieware be difficult to extend to 5-Axis operation, with the XY gantry and a printed bed utilizing simple hinges and 3 gantries to precisely tilt?

On a slightly different topic, does anyone have a source for aluminum foam sandwich panels? I've been looking for light, yet extremely rigid, 3d printer beds.

Kind Regards,
Ivan Kozlov