New Feather non-Cartesian printer design

Here's a new non-Cartesian 3D printer I designed. I call it the Feather printer because the total X-Y moving mass (the build table plus slider blocks) only weighs about 150g. The whole printer only weighs 2.5kg.

I'm still working out the kinks, but as you can see from this video (youtube Feather printer video) it does, in fact, print. I built it using my Lobo CNC milling machine (see: [forums.reprap.org]), and at some point, I'll post complete plans if people are interested. I've already posted some info on the extruder (see: [forums.reprap.org]), and I'm happy to post more of the gory details if people are curious.
-Jeff

Very cool.

---

___________________________________________________________________________

My blog [3dprinterhell.blogspot.com]

Yes i like the concept. Very nice project.

But i have some doubt with 'precision' because a small gap in 'pivot' mouvement have many influence on the theorical position.

Indeed, it's always fun to see new idea/concept :-)

... interesting design!

What's the overall accuracy? - looks a bit 'wobbly' with this screw-hinges, but seems to be precise enough

---

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

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

The screw hinges (I'm assuming you are referring to the funky Z motor mount) actually work fine because throughout the entire Z axis range of motion (about 5" at the extruder) the motor tilts only about 5 degrees - hardly any motion at all. The flathead screws stay firmly seated in the countersunk holes. I was mostly worried about backlash in the pivot points driving the table, but in the test prints I've done, there's no visible discontinuity where the motors are switching direction.

One big thing that helps with accuracy is that the moving mass is very light. Aside from out-and-out backlash, any flexibility will give errors proportional to the load, which with the very low mass, is kept to a minimum.

One other thing many people don't realize is that stepper motor accuracy varies with the load. Because each stepper torque detent acts as a spring, even with 16x microstepping, the accuracy of the motor at full load is +/- 1 full step. (Also, even with low loads, at high speeds the motor error can approach +/- 1 full step.) Another reason why keeping the mass low is important.

The thing I'm wrestling with now is the Z axis drive screw. Any errors in it get magnified by about a factor of 3 in extruder height. My layer thickness is currently varying by .06mm because of some small irregularities in the screw. I'm swapping out the screw to see if I can improve that. I may have to punt on hardware store threaded rod stock and go with a more precise acme threaded rod.
-Jeff

interesting design almost an inverse scara design.

keeping it light is indeed very good but still accuracy must be problematic. Any play at the pivot points will translate in wobbbbblying !

Regarding play, at all of the pivot joints, I am using 0.250" dia nylon spacers with 6-32 screws. It turns out that by tightening down the screw, you can force the spacer to bulge a little bit to get a zero clearance fit. This would never work if you had any real load on the pivots, but they only ever see a few oz. of force.

Love it. I've made many MCAD experiments from this concept [www.youtube.com] and your variation with the pivot is very interesting. What's the build area?

Edited 1 time(s). Last edit at 06/03/2013 08:35PM by billyzelsnack.

billyzelsnack Wrote:
-------------------------------------------------------
> Love it. I've made many MCAD experiments from this
> concept [www.youtube.com]
> and your variation with the pivot is very
> interesting. What's the build area?

I like your plotter in the video. I had made a similar one a while ago (not nearly so nice), which, in fact was the inspiration for the feather printer design.

The overall build area is 35in^2 (225 cm^2), with a Z axis range of 5" or so. Of course, the build area is not a rectangle - the build table is, for the most part just 1/4" bigger around the edge than the build area.

One thing I forgot to point out is that the build table essentially has 3 point support - at each of the 2 slider blocks and at the end support for the rails. Because the extruder sits right over the end support, there is never an issue with having to level the table (or even that the table be particularly flat) just so long as the table has uniform thickness.

oh that's not my plotter. I've just done MCAD experiments based on the concept from that video. It didn't occur to me to build a moving table though and I think your design is very interesting.

Is the pivot slot straight or curved? Is the point of the pivot to gain extra "x"

As for Z.. This might be crazy, but what if you rotated the entire thing 90 degrees so XY becomes XZ and then have the extruder move on a belt driven Y?

billyzelsnack Wrote:
-------------------------------------------------------
> oh that's not my plotter. I've just done MCAD
> experiments based on the concept from that video.
> It didn't occur to me to build a moving table
> though and I think your design is very
> interesting.
>
> Is the pivot slot straight or curved? Is the point
> of the pivot to gain extra "x"

The pivot slot is straight. There's probably some optimal curve for the slot, but for +/-40 deg. of table tilt, straight works fine.

In general, your plotter and the feather printer are known as 5-bar linkages. In your printer in particular, each linear rail is the equivalent of a link, plus the two links connecting to the pen, and the base that connects the two linear rails also counts as a link. Each link has a 1 degree of freedom joint associated with it (linear or rotary) for a total of 5 degrees of freedom. For planar mechanisms, when you form a closed loop with the links, you end up reducing the overall DOF's by 3, giving you net 2 DOF which gives you your X-Y plotter motion.

In the feather printer, the slotted pivot joint has 2 DOF (rotating and sliding), or alternately, it can be mathematically modeled as two 1 DOF links superimposed on each other. The net result is the same - 2 degrees of freedom for X-Y motion.

billyzelsnack Wrote:
-------------------------------------------------------
> As for Z.. This might be crazy, but what if you
> rotated the entire thing 90 degrees so XY becomes
> XZ and then have the extruder move on a belt
> driven Y?

Seems like that might work. My idea in having the build platform do all the moving is that you can make a build platform much lighter than the extruder.

Very nice work!

is this thread still active? if so I was wondering how you compensated for the slight circular motion you get with the z-axis movement. I thought about a design very similar to this one earlier and don't know enough about firmware to decipher this.