Friday, June 24, 2005
Last weekend I gathered some bits and pieces from some old dead printers so I now have some stepper motors and gearing. None of the parts are ideal, but it should be enough to get a prototype working from. Unfortunately I discovered I was out of brazing rod, so I didn't make any progress with putting anything real together. I might be able to get out to the hardware shop this weekend if all goes well.

In the meantime, I've been discussing communications ideas with Adrian and Vik and after a few days of tinkering I now have a prototype comms library that does the right thing in the emulator. If I don't build mechanics this weekend, I'll probably breadboard this and run it on some real PICs.

The implementation is currently a token-ring-like system based around asynchronous serial communications. The packet structure is based on a small open system called SNAP. I have a few reservations about the design and suitability of SNAP now, but it's partially implemented so I'll put up with it for now. It's all fully interrupt driven and deals with most stuff automatically in the background, so it should make building the various controllers a straightforward task.

There are future plans to probably replace the token ring system with an i2c based system which will improve latency and speed. This should be sufficient for early experimentation however.

It's all built in sdcc, which has a few small glitches, but so far I've managed to work around all the problems and things are progressing well. It's on a CVS server, but not available publicly at the moment. Perhaps it will move to sourceforge or some place like that in the near future.

Friday, June 17, 2005
Earlier this week after contacting a bunch of suppliers, I gave up trying to find a good source of gearing, but all is not lost. A lot of gearing can be found in old equipment. The older the equipment, generally the better -- they used to build things really well. Daisy-wheel and similar printers are about the best but an old dead laser printer or even a VCR will likely be a source of something usable.

At this stage without a reliable and repeatable source of affordable gears, I'm going to get the parts from scrap since I already have quite a bit. You can get old broken printers and VCRs for a few dollars. The downside is that the entire machine won't be exactly repeatable, but will still work with other gearing with very minor modifications.

Monday, June 13, 2005
It seems very hard to find the worm gears (and matching spur gears).

I can't currently find anyone that sells these at anything but rip-off prices (such as RS components). Making them will probably be problematic given only simple tools.

Sunday, June 12, 2005
Preliminary design
I experimented with a few different layouts to see what could fit where. This is a little time consuming in CAD but it sure beats building and throwing away real parts.

A few designs were considered including:
This general design was initially selected because it can relatively easily produce circular features and has a stable point for an extrusion head.

The threaded rod was selected because it is readily available at a hardware store (whereas racks are not). This also provides some degree of gearing.

Tubing is being considered for the general chassis construction because it is an innately strong shape, readily available and pretty cheap. On the downside, it's not very easy to work with. I was hoping to find something other than copper which may be a bit too ductile and might deform, but steel or brass tubing was harder to find. Copper can be brazed together with just a cheap gas torch. This will unfortunately be fairly labour intensive.

These diagrams are just roughing out the design at this stage, and some bearing and supports are not shown yet.

This is a general overview (rough layout only):

A slightly closer look underneath the turntable:

And a closer look at the extrusion platform:

incidentally, the round brass things are supposed to represent spur gears, I just didn't want to spend time modelling them.


  1. If there is any inaccuracy in the machining of the threaded rods or the worm gears, there may be some jamming. If this occurs, it would be preferable to use only two nuts on the platform rather than 4. This would make the platform less steady, so an additional polished rail would probably need to be added with a bearing sleeve attaching the platform to the rail. This would be at the opposite end of the platform from the nuts.
  2. The platform raising mechanism depends on a vertical slot cut in the copper pipe. A pin through the threaded rod sits through the slot to lock the rotation of the rod and pipe while allowing the rod to rise and fall. I'm not sure at this stage how easy it will be to accurately cut this slot. It needs to closely fit the pin to prevent a rotational hysteresis type effect. Perhaps there is a better mechanism to allow the platform rotation and raising.
  3. Having one motor that controls both height and rotation of the turntable is a nice simplifying idea from Reprap, but it does rely on the quantity of material deposited from the extruder matching the change in height of the platform after one revolution. I'm not sure how realistic this is. It sounds like too much material for accurate production, but we'll see as it progresses.
Perhaps a third motor is in order (to prevent concern #3). For example, if the platform freely rotated and a third motor was reponsible for raising the platform, then the amount of material deposited can be changed without problems. One interesting way to do this (suggested by Dave) is to motorise the nut that currently causes the platform to move. If the nut is held still the platform will rise and fall as it rotates. If the nut is rotated in unison with the platform, it would remain at a fixed height. More subtle adjustments could be applied by rotating the nut at intermediate rates.

Saturday, June 11, 2005
Component placement
Here's an idea I had for a slightly different way of attaching components to a reprap that may improve bonding and electrical properties.

click for a larger image
The difference with this approach is that for parts containing electronics, the automated production would be paused at some point and a human would assist by placing components into the appropriate places. The placement would be very straightforward, just resting the components in appropriately shaped cavities that would hold them securely enough for a conductive layer to be applied, holding them firmly in place.

Another advantage of this approach is that you can potentially have multiple layers of components, making use of three dimensions rather than just two. This could be quite beneficial given the small scale integration density of the RP approach. Unfortunately this would require human intervention for each layer. Of course once the basic system is working, no doubt others will build add-ons to automate more of the process, so a little human intervention isn't much of a disadvantage in the early days.

Parts costs
I went and looked at some bits that might be useful.

300mm non-stick pizza tray: $3.99 (Warehouse)
M10x1m threaded steel rod: $6.50 (Bunnings)
M10 steel nuts: $0.16 (Bunnings)
M12x50x50x3 square steel washer: $0.80 (Bunnings)
Copper tube 15mm x 1.02mm x 1m: $9.06 (Bunnings)

Friday, June 10, 2005
This blog relates to the Repstrap project. The Repstrap project is related to the RepRap project and aims to build a boot-strapping system with sufficient capability to produce the parts needed to build a seed RepRap machine.

For those such as myself that are not lucky enough to have access to a rapid prototyping machine this will hopefully provide a means of getting involved in practical development with RepRap at an earlier stage.


Powered by Blogger