Parallel flexure test

This will be my first attempt at a 3D printer, went a little crazy on the design. Just got done throwing together my proof of concept. Hopefully with the lessons learned from this a working machine can be built. Lots of things to fix. The screws holding the fishing line need moved further toward the ends of the boards. I need to play with the current limit on the easydriver boards, the motors only have enough torque to move the platform about 1/4" in any direction, which is about 800 steps. Also, probably need to preload the stages with springs to reduce the torque and use springier metal in the flexures(this stuff is pretty soft). At this scale I will likely only get a 4"X4"X4" build area. Despite all that, I still think it has potential.

Were you planning on having a Z axis also?

It has X, Y , and Z. Each flexure has two degrees of freedom, one is X and Z, the other is Y and Z. They lose off axis stiffness the further they move from the neutral position, so I divide the Z movement between them.

Interesting design. I see the stepper motor drivers, what kind of micro controller are you using to control them?

---

[scara3dprinter.wordpress.com]

I'm using an FTDI Cable 5V in bit bang mode with a CD74HCT4094 8-bit shift register and doing all the processing on the PC. I can daisy chain more shift registers to control more motors/accessories and so far I can generate step rates higher than my motors can reliably step, so I haven't invested in a proper microcontroller. The power source for the motors is a pair of old 5V cell phone wall warts each powering two motors, I will replace them with a 12V PC power supply in the near future and dial up the current limit on the driver boards. This nice thing about such anemic power supplies is if something goes wrong in initial testing phase fire and sparks in the expensive parts are less likely, kind of a very cheap current limiter.

I started off playing with some of the super cheap ULN2003/28BYJ-48 Stepper Motors from ebay. But they had massive backlash in the gear train, so I opted for something a little more upscale.

I attached the hodge podge of python I threw together from various sources to have something to play with. The responder and rhombic code is for a different design I haven't built, so much of the script is dormant until I modify it a little bit to take the drawbot like math this motion platform should use. The squares in the pictures were drawn using the fact that in the middle of it's range the printer is fairly linear so the movement looks good at this scale but will become more distorted as the size increases.

test_flex_squares(9) produced the squares around the pen in the last picture.

Clearly you know exponentially more about programming than I! This leads me to a bit of an off topic question that perhaps you are capable of answering. I've spent a bit of time trying to build full color FFF printers but for obvious reasons its a difficult problem. The other day it occured to me that perhaps I'm putting the cart before the horse. What if there was a post proccessing machine that added color after the object was made. To that end I thought of a ridiculous movie. In Mission Impossible Ghost Protocol the characters use a machine to make a mask.

I can make this machine, but how hard would it be to write the firmware and programs to control it? I would add a mechanism that stayed in contact with the part while it painted to add reference.

I don't mean to hijack your thread, just curious if it can be done.

---

[scara3dprinter.wordpress.com]

Hmm, printing on a curved surface looks challenging. Took me long enough just to figure out how to draw a circle in 3D let alone follow an arbitrary curve. If all goes well with building my printer and eventually scanner, I suppose this would be a fun next step. So I'll keep it in mind but I think I'm a bit further from being ready to write the control software for it than you are to building it.

On that note though, I did manage to fix a lot of bugs in my script and implement circular arc code G2 and G3, so I can use the plugin for inkscape to make engraving paths.

Using flexible metal, how much Z axis do you anticipate having?

---

[scara3dprinter.wordpress.com]

From the literature I read, depending on the material you can laterally flex this kind of parallelogram flexure between 10 to 20% of the spring length and still have long to infinite fatigue life. So if my springs are 20 inches long I should be able to flex 1 to 2 inches without ever breaking it. If all went according to plan I would have a 4x4x4 inch build area.

Someone donated an old laptop supply to the project, it is 20V and says up to 9.5A! ( some kind of dual this, quad that gaming luggable that died ). Anyway, although the beefier supply more than doubled the travel range before the steppers slip, they are still a little low on torque to the 26 gauge steel far enough and I'm still not sure if my new geometry calculations are correct. I'll attach a pic of some text from inkscape and the new code. I'm using a com0com virtual null modem to connect pronterface to the script which still is just a collection of functions I run from the python console.

Also I had to write a script to clean up some the comments and extra zeros in the gcode produced by inkscape.

This is a really cool, original design, Delgir. Nice work! I just saw this today and I'm super excited about it! Long ago I posted some links and notes about flexure hinges and compliant mechanisms here but as far as I know, your printer is the first that does X,Y,Z with only flexures. Very cool.

Do you think you would gain anything if, instead of using the large metal flexures, you were to make the linkages by printing a "living hinge" for each end, and then using a stiff material (wood, metal, printed strut, etc) for the body of the link? The pen holder on this page is my favorite example of something with a printable living hinge: CalcBot pen holder.

I saw the info you posted on that page, lots of good info. In fact that is part of what inspired me to build this test. I like the paper by Trease in the links at the bottom, since it gives a handy comparison of various joints.

I haven't played with any printed materials yet, so I don't know how well they hold up to repeated bending. My first concern was that plastic would creep under the load and loose calibration or worse fail completely. Having stiff members for the body does improve the off axis stiffness but also increases the demands on the living hinge. By distributing the motion I hoped to be able to to use cheap materials and still get good performance.

At some point I should try out plastic as an easy to work with alternative to metal, especially since I have been buying metal that isn't that great anyway. I never did get enough force from the steppers to pull both platforms far enough down to get a reasonable work area. So I'm currently looking at some other compliant hinges that have a greater range of motion and less torsional stiffness, such as split tube and cruciform hinges. My next iteration will probably use something like strut channel to approximate a split tube.

I hope you keep us updated on your progress. I would be interested to see how a printed split-tube flexure would work - I have a feeling it might work quite well.

I thew together a test of the concept you posted before. I need to run by the hardware store to get more materials, didn't have any PCV pipe around. So, I hacked up a couple pieces of slotted angle to try something akin to a cruciform revolute joint. The performance was so-so, it really needs full symmetry and longer angle springs. Will probably be even tougher to fit with an orthogonal copy for the other axis than my first model, still looks doable though. Will keep experimenting as time permits.

Delgir, you inspired me to try something similar using stuff I had sitting around. The wood is 1X2. The tubing is 1/2 inch diameter, 0.07" wall thickness plastic. The tubes are "split" by slitting them once lengthwise with a razor. The plastic is pretty "gummy" - not especially springy - so it is not very well suited for something like this. Even still, the flexibility in the desired mode is fairly good, and it is quite rigid (sorry I don't have any quantitative measurements) in the off-axis modes.

I am working on an OpenSCAD design for a printable split-tube segment, will post when done.

Here is a quick design for a split-tube hinge in OpenSCAD. It's probably not the best design for printing, but I will upload a few others. A cruciform, as suggested by Delgir, would probably print better than a circular split-tube.

Sorry for reviving this thread, but this is so original and so much thinking out of the box that I just can't help expressing my amazement!

Very impressed, even if the practicalities of the implementation seem slightly insurmountable.