We are finally back with an insanely long Blog Update!

With the FABtotum Personal Fabricator we went through a series of experiences in designing a proper belt system, and we thought it might be cool to share this experience.
Some designs spawned several prototypes, some of those where a step forward, other a failure, but all helped us progress to today’s development stage.
Here a bit of the story.

When dealing with hybrid machining you face a serious limit: provide fast movements and high accelerations (especially the latter) for additive fabrication and, at the same time, be able to have a very high inertia or pure “strenght” during cutting and milling operations.
For compact sized hybrid devices the answer has been in introducing new belt designs to keep forces under control and provide fast movements if necessary.

The first attempt can be seen in the Alpha prototype.
The Alpha Belt system was a modified carthesian one.
This version was the starting point in trying to mix subtractive and additive fabrication with a belt design.
In this version the Y axis motor drives a long belt that goes around the building space.
The X axis motor moves the head carriage with a normal belt. The best thing about this design was inertia and stability, but acceleration and noise was pretty high, due to a custom double-sided T5 belt.
The belt tensioning had to be perfect and the circuit was split in 2, so you had to manually balance the tension and try it a couple of times before getting it right.
It wasn’t the fastest and the strongest belt driven system but it could get the job done.

The system itself can be simplified as follows:

With respect to a simple cartesian motion in which each motor has his own belt and axis, the above design is more balanced, so that any force on the gantry is supported by an opposing force each time, easpecially when moving and cutting near the borders of the working plane.
From a mathematical point of view, however, the forces can be calculated easily as any cartesian system (0.5 Nm Nema17 motors have been used to fill the variables)

 

Eventually, we moved to evaluate a variant, in wich the Y belt was connected to the other side of the Y bridge.
This was a intresting concept that, however, had a flaw: space consumption!
The Y axis (moved by the blue belt) runs around the structure,
The next major attempt in improving hybrid manufacturing was during early alpha, with the HBOT design.

The Hbot design, like the CoreXY (of wich we’ll talk in the next paragraph) uses 2 motors connected to a single belt.

The system works like a ship pulley (or “block”) where the force is split in half. In fact it works like a 2:1 reduction.
You don’t need to lift the W weight, but only half of that, because your movement is reduced by a factor of 2.

In an H-bot or CoreXY system the X or Y movements are basically a sum of 45 degrees vectors. We are effectively working on a tilted reference cartesian chart here.
Since vectors add to one another the resulting force in the normal XY plane is much higher. Sounds complex? here a scheme:

 

As you can see, with the same motors you get a different result than the basic cartesian system, with twice as much force in each direction (except in diagonal moves, ergo in the direction of the native tilted reference cartesian system). The price to pay is the distance you move, wich is effectively reduced to half, just as expected.

This alternative driving method was simple and effective in most movements and condition but one: X-axis movements. To fight the movement on the Carriage, forces have to be balanced properly. If the XY carriage faces resistance (like in any machining), there is a resulting force in the gantry that causes the gantry itself to flex, generating positioning errors.
This is pretty clear if you think of “what is holding what” and looking at the gantry.
The Hbot design can bring substantial advantages during additive manufacturing but it’s not suitable for subtractive machining due to this problem.

To balance the forces in a belt driven design, systems like the MIT’s Core XY have been developed, and we did some research too to see if it could fit our application.

 

From a movement point of view the COREXY can be summed up as a normal cartesian motion tilted on a 45 degrees angle, just like the simpler H-Bot (see http://corexy.com/theory.html).

Unlike the H-bot, however, the CoreXY introduced a crossed belt that balances the forces on the gantry. From this basic implementation the COREXY is a valid hybrid system that however has some flaws.
First and foremost, the belt is very long, introducing a “dampening effect” due to the belt extending under tension.

One solution is the one to increase the belt size, but this means more friction, noise and loss in force in a system with 8 idlers and 2 motor pulleys for a total of 8 90°angle turns and 2 180° angle turns.

Never cross the belts
Try to imagine all life as you know it stopping instantaneously and every molecule in your body exploding at the speed of light. Ok it’s not going to happen but that’s what you’d deserve for driving belts that way.
Belts in the coreXY are not running on the same plane because they have to cross on one side to balance forces. The result is that each tooth of the belt entering the pulley encounters a non-straight pulley, wich results in noise and possibly ruining the belt in the long term. We love the coreXY beacause it’s essentially a great and smart solution. Many things like belts crossing could still be improved in the FABtotum and so we started working on it tirelessly to squeeze each and every last Newton we could.

But before continuing let’s have a recap of some of the belt-driven systems we considered during the Alpha and Beta development.
Our interpretation of the Core XY solution was to avoid crossing the belts, like in a H-bot style design, but still having the advantage of balanced forces of the Core XY design.
The solution we adopted was to have 2 closed belt loops, on two different planes.

 

 

 

This way effectively compensates the Hbot torque on the gantry with the help of the belt on the other side. At the same time avoids unortodox pulley mis-usages.
The system works like this:

M1, the left motor, drives the red circuit, while M2 drives the blue circuit.

Those 2 belt circuits run each one on different levels, always straight and do not cross.
Two equal and opposite forces “F” separated by a L distance generate a torque of M=2FL.
In this solution the forces generate a couple on the carriage, wich is effectively mitigated by the short distance between the 2 forces.
If we compare this to the Hbot, we basically moved the problem in a location where it couldn’t affect the system anymore.
In other words, while in the Hbot “L” was really big (the distance between the motors) now L is the distance between the 2 anchor points on the carriage/head wich can be as small as we want. 2F*(small L) = Small torque.

 

Of course this is just a part in the bigger game that is “making the thing faster,better,stronger”. Part of the improved machining capabilities can be traced in improved stepper power management, but also in a lighter carriage and moving parts.
Le’ts see more of that, shall we?

Structural superiority, part II

With the belt design out of the way, the mechanical team lead by FABtotum’ Chief Engineer Alfredo Marinucci started working on the improved structure for the Beta Version.
When we started the Indiegogo campaign we developed the simpler and most effective way of keeping things together, but thanks to the campaign’ success we could afford to be more bold, and to push the structure even further.

The pursuit of a good-looking, solid, simple and reliable structure is the main problem of any design endeavour, and this was also true with the FABtotum’ structural outer shell.
Italian Design is well known to have never turned down this challenge before, from the field of engineering to pure aesthetics choices.
With the design process of the outer shell we went from a flat design that was a mere accessory made in a single material, to a complex and reliable part.
The shell is made with a composite material: an fiber-enriched polymer, that is capable of withstanding more load than the previous one.

 

We didn’t have any prototype of the outer shell until last week, when we finalized the molds (see last IGG update). From then we were able to start building the pre-production prototype with the outer shell, covers, bells and whistles, instead of the “naked” prototype we showed recently. The company we are working with to produce these parts has a long history in automotive industry. The fact is that the new shells are in no way cheaper or weaker than the Alpha side panels, but they add a new level of reliability and quality to the FABtotum.
With Alfredo’s passion for automotive engineering and the materials used, you could as well compare the FABtotum’s structural shell to the 3d-printing industry equivalent supercar racing chassis.

Hybrid manufacturing
One of the most important parts of the hybrid manufacturing design is how you actually mill/cut in the end.
The FABtotum uses a 200W brushless motor with a custom made (and soon to be open source) ESC to pilot it. This setup has been proven solid and a definitive improvement over any other Beta prototype we tested, including AC spindles.

The chuck and the axle of the motor are machined for precision, but we have different suppliers with different specs to look into it as one of the last assignement in the development of the FABtotum Personal Fabricator.
Those are the last things we are deciding this month before ordering parts.

Overall, hybrid fabrication on the FABtotum is possible thanks to several solutions adopted.
From the alpha we where able to change many times our approach and end up with something unconventional but reliable.

As with the mechanics many things contribute to the effectiveness of this sub-assembly, like bearings, power control and cooling.

Closing words

This is it for today’s Dev Blog entry,
Hopefully we showed some light on the mechanical development process and how important it was to us, how much effort we put into it and why we think we did a good job.
We know there are loads of things you would like to know, and things are getting ready pretty fast. We ourself can’t keep track of the advancements.
Proof of that is this very post that has been postponed because it became obsolete in one week!

 

It has been a long way from that prototype (wich admittedly has been savaged for spare parts from time to time). While the time passed very fast, we can’t really make up our mind on the sheer number of things we investigated and tried before committing to a change.
With Indiegogo’s upcoming campaign update we plan to show the final pre-production prototype as we prepare to start manufacturing during April.

hold on until then!

 

related links
https://github.com/ErikZalm/Marlin/issues/463
http://www.edn.com/electronics-blogs/mechatronics-in-design/4368079/So-you-want-to-build-an-H-bot-
http://www.anthonyvh.com/2013/05/21/halubot-part-1/
http://www.cnczone.com/forums/diy_cnc_router_table_machines/51485-make_gantry_rock_solid.html
https://groups.google.com/forum/#!topic/h-bot-and-corexy-3d-printers/1gb6oo291zI