Bamboo Printer

From RepRap
Revision as of 20:22, 28 November 2014 by KalleP (talk | contribs) (Added some links)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
Bamboo Documentation
Crystal Clear action run.png

Release status: concept

Bamboo printer.JPG
A compliant link mechanism, no bearings
CAD Models
External Link

This will eventually become a development page. Please comment on the discussion page if you have any thoughts on the ideas here.


At the moment it is just to store a couple of ideas that I have that may allow the use of natural herbal supplements instead of vitamins. Basically I am thinking of using cheaper mechanical components to allow for building in less industrialised regions by making use of more parts from nature or the hardware shop. A couple of the other components may need to be less standard but may become cheap if there is a little development added to them. Parallelism at a low complexity has been my goal. However it is also of no consequence if the rotational error is measurable (then it can be calculated away) or the optical centre of the absolute position detection camera is positioned with the extruder nozzle (it does not matter if the part rotates slightly under a point source dispenser) but this configuration cannot use the simple double wishbone Z-axis with the Z dependent X error.

A specific drawback to this idea is that it is likely to have a larger in footprint than an industrialised RepRap but there is room to make modifications especially if some of the parts are printed (or laser cut) instead of found in the back yard.

This project cannot work yet as some of the ideas have never been tested and I do not have the personal skills to do the development. I hope to develop it to a point where just one or two people/teams might be able to complete the missing bits of this idea at which point it could become a viable development path.

Compliant linkeages

I have decided to call this the Bamboo Printer not because it prints in bamboo but because of the innovative use of bamboo (or other flexible material like glass fibre reinforced tent frames or ski poles) skewers is used to replace much of the linear and rotary bearings of the present RepRap variants. The compliant nature of the mechanics may result in various non linear and other geometry issues but is ignored due to the position feedback system described below. Compliant links

This design will not have a large portion of printed parts though the 'bamboo' bits could be partially printed I expect, though at a higher cost. Possible compact versions It will however have a much reduced vitamin cost (or that is my thinking).

The bamboo sticks will supply a restoring force if only flexed in one direction from rest. Over a certain range they have a reasonably long linear operation region but with a additional spring tension they would have a movement range of double if they were to move both sides of the rest position. If always under simple tension the motor arrangement may be simplified.

Software issues

Optical 2D feedback

The second part of the project that needs new development and is a bit beyond me personally at this time is the optical camera mouse position feedback system (servo control). It is possible that the software could be simplified in part if one were to replace the mouse cameras with a more expensive USB webcam or something more expensive (especially for proof of concept testing with possibly a PC running LabView or similar just for motion sensing) but the idea is to make use of the ubiquitous optical mice to do the tricky part of the project. In short one places one mouse guts upside down looking up at a printed high contrast graphical pattern on the underside of the build platform that is sliding on a sheet of plate glass (or other smooth surface with a hole for the lens). The camera observes the motion and sends incremental signals back to the motor controller to position the platform as per host commands. Absolute position can be established by having a second mouse camera (certain debug modes allow picture taking) looking at the pseudo random pattern. No limit or home switches will be needed, mechanical end stops can be driven against and used to check motion function and build limits.

The mouse brain and optic could be placed under the build bed facing down working just like a regular mouse as long as the build surface was kept reasonably clean and has enough optical features for a good tracking signal. Some of the laser mice apparently have a very low angle of illumination and can work on almost totally smooth and featureless surfaces.

The mouse brains offer from 400 to 3500 dpi resolution, 1500 to 6000 samples per second using the internal DSP and the 16x16 to 30x30 camera and oblique LED or laser illumination. The high resolution mice favour blue illumination and lasers. The maximum tracking speeds are between 400 to 1000 mm per second (16 to 40 in/s).

Here is information for the start of the feedback controller, it is single axis but with more horsepower if should be able to do two and read the camera/s. Microstepping with optical feedback.

There are links for optical feedback servos and some for position sensing using opical mice. Actual servo feedback in two positions will follow soon an will be linked here when found.

Some theory and specifications

This method has since been suggested for other mechanical arrangements but with the simple mouse brain it is limited to rectilinear motion build platforms or heads as the mouse DSP is designed to track motion in two fixed orthogonal coordinates only and unless the angle of the sensor is known it is not possible to know which way the mouse tracking signals are moving. Facing a mouse up above the extruder and having a surface for it to look at would allow use in a XY-head configuration if the filament was fed in from the side (or an extra Morgan arm was used) and the head was kept evenly pressed up to the 'ceiling'.

X correction

Finally similar to the ideas of the parallelogram idea for the Z-axis platform that is also proposed for the 5 link SCARA Wally design that is being developed, the Z-axis would also be a compliant bamboo double wishbone and the sine/cosine error affecting the X offset would have to be corrected for.

Consumer grade motors

The position feedback arrangement will allow for use of salvaged or consumer motors; DC, brush-less, stepper, pneumatic, linear motor or what have you as long as the controllers can be found or developed.


Any standard extruder and hot end design can be used. This will possibly add a bearing or two to the design but some bearing free options are in production.

The cold end can now be done with a geared stepper and a suitable Pilot pen pocket clip as the pressure pad or a Scoop Extruder these require no bearings. The hot end will be best done using a Bowden set-up to keep the weight on the Z-axis down.

Another idea for making use of optical feedback servo motor drives is with a Bowden extruder. If you had a small rubber wheel that rode on the filament (in a scoop in the PTFE tube just before it enters the Hot End you could monitor the actual filament movement. This has been proposed before as a way to monitor for slip and slack in the tube but not so often in connection with a optical feedback servo controller. Secondly if one already has a servo controller design using an optical mouse brain it might be possible to have the wee mouse eye looking at the filament and controlling the remote drive motor (stepper, DC, brush-less whatever). This would be an ultimate test bed for the mouse sensor as it is a single axis and it is also insensitive to an ocasional missed step as the error cannot acumulate like it might on the X and Y axis.

Drive techniques

String drive

A number of delta-bots have tested the use of 'Spectra line' (a braided fishing lure product) instead of timing belts and this would be the preferred mechanical transfer medium. Again the feedback will allow for simple coupling using variable diameter capstans (like on a sailing boat or an old radio frequency dial cord arrangement) pulley that may or may not slip a bit as this will be corrected automatically. The drive could be a single line wrapped around a pulley/capstan with an elastic to tension the free end that pulls against the spring force of the bamboo skewers, if the movement is more than an elastic can handle then the free end can be looped around an idler pulley and attached to the moving part with a spring (like in the dial cord) to take up the change in string path length due to non-linear movement.

Dial cord is also a woven line that is reasonably durable to flexing but not to chafing though it is getting harder to source under that description. I believe there are types of suture materials (medical stitching gut) that are also woven and could be a possible source of high quality line if sourced in bulk.

Thin wire 'rope' is another option, it has negligible stretch. This was used in a lot of XY plotters and older impact printers. It is also known as aviation wire and used as fishing lure leader as it is bite proof. The drawback of a steel wire is that they are more expensive and they are susceptible to kinking, it would also not be ideal for the capstan type of pulley as the friction will not be very high. Coated versions that would run smooth and not cause pulley wear are also available.

With the direct position feedback, even with the possible stretching, monofilament fishing line (gut) would also be a low cost option, A spring or elastic could be used to take up the stretching.

The innovative DIY flat belt drive is also an option.

The differential capstan method could be diret connected to a generous motor as a garing system to pull the stage against the compliant restoring force. It would allow for a very low cost system that could have any reduction ratio, the inherent lack of absolute repeatability is not an issue due to the position feedbck from the stage, this system is ill suited to a push pull arrangement but ideal for working against stiff restoring force. It could be used in the same way for the Z-axis.

Threadless Ball screw

The Z-axis can be driven with a threaded rod or a timing belt as traditionally but a suitable possibility is the threadless ball (bearing) screw. It uses angled ball-bearings to force a rotating rod forward or backward much like a real linear ball screw. However they can be made cheaper and only need smooth rod and do not have backlash like a lightly loaded ACME screw. Problems are that they do not have absolute accuracy and the creep is not totally predictable though slight and the ball bearing nut is more bulky.

With feedback the creep is not a problem and they can be designed with any transmission ratio (lead) required (within limits).

The feedback could be arranged with yet another optical mouse brain on a strip of tape that moves with the vertical motion of the bed. This would allow for use of any type of motor again and have another added advantage. You could crash the head into each corner of the bed at the start to determine the bed level. If the head came down with gravity it would stop and the servo motor would fault when you try to move it past the hard stop.

A working design is available for testing on Thingiverse.

Other links

  • The start of the Bamboo Printer discussion forum thread
  • If this ever gets made it could get renamed to the Kelchner printer after Scot Kelchner and his team that studied bamboo genetics
  • Forum post with picture A picture that has the same sort of feel to the bamboo printer, it closes the square sort of like a pantograph and uses joints instead of flexural members but looks very clean.
  • Brushless motor control It seemsthey use shaft encoders on one of the pair of motors for each axis. The other motor seems to be working as a parallel connected assist motor that will reduce the drag on the other end of the rails.