After two years of fiddling, I have to admit that my progress has lagged far, far behind that of the official reprap project. My primary problem is the lack of good, simple homemade pistons. Syringes are very good at pushing, but pulling is another matter. The water just boils(from vaccuum, not from heat) and nothing moves. I tried to make better pistons out of copper pipe, but I am not a master machinist, anything with a good enough seal was nearly impossible to move, too.
And the electronics I'd developed, while (I like to think) fairly clever, was totally obsolete by the time I'd mostly perfected it. Reprap has moved on to a simpler all-in-one board. While I still think my pics will find uses here and there, this particular use isn't it.
So the biollante project is off, for now. I've gotten a Rapman 3.0.0 kit from Bits From Bytes, which seems to be about as complete and faithful-to-the-original machine as can be found in kit form. Once I'm able to crank out parts on design, then perhaps I'll be better able to build my own designs.
Biollante is my own personal attempt to build a Replicating Rapid Prototyper, of a design radically different from their present "Darwin" design. At present, I have a frame, and some motor firmware for the PIC16f628a.
Why name it Biollante? Well, at the time, the Reprap Builder's Blog was awash with Godzilla-inspired names. I decided to be a rebel and a lemming simultaneously, picking the lamest Godzilla monster I knew -- an incorporeal mutant rosebush that can cough radioactive spores and fly through spaaaace! It's also a reminder to me to not take this too seriously; after all, just because I'm a reprap builder doesn't make me a good reprap builder. ;)
Instead of designing the structure around a rigid platform like a milling machine, or an "arm" like actuator, I decided to build a Stewart platform. My hope is that such a symmetrical and intrinsically stable platform will be easier to build than a top-heavy structure with few load-bearing points. The electronics and software will become more complicated, of course -- twice as many motors to control -- but this mechanical numbskull considers it more than a good tradeoff for a simpler and stabler mechanism.
Since it is only intended to have plastic laid down on it, not to push metal into a milling bit, it can also be much lighter than I think I could build for a rigid linear axis system. So light, in fact, that I have used needle joints for the universal joints, held together by worryingly strong magnets scavenged from dead hard drives.
The linear actuators are simply 12cc Monoject® medical syringes. They are convenient since they are cheap, seal well, and calibrated to millimeters. Yes, millimeters -- 0.2ml of fluid is exactly 1mm of motion. Their motion will be driven by a fluid pump, probably another syringe.
I'm borrowing a lot of design on the electronics from the RepRap "universal" controller; each module, designed around a PIC16f628a microcontroller, performs one function, communicates serially, and is connected in a ring. The computer sends to the first module, the first to the second module, etc, and the last module sends back to the computer. But there are some important differences:
- Seperate power and data connectors to avoid disaster
- Individual voltage regulators for each module to avoid motor noise
- Modified pin allocation to take advantage of more PIC hardware features
- Discrete transistors instead of expensive, hard-to-find stepper chips
- Human-readable and writable ASCII packet format -- talk to the chip from any serial terminal
- Chip does not busyloop while waiting for a packet to become complete
- Firmware written in nearly 100% pure ANSI C, inline assembler used only for device-specific functions
- Constant values placed in EEPROM memory to conserve code space -- can be updated at runtime
Now, onto the fun stuff.