It is surprising how little you need to start playing with mobile robots. I borrowed a couple of OOPic cards from a friend, picked up a a couple of standard R/C plane servos and I became a robotist. Then I visited the local Value Village second hand store for R/C toy parts. I needed wheels and dreamed I might find more. All their battery cars were $3 to $4, and I did find a couple to at least get off-road tires from.
I started with standard R/C servos for actuators for Beverly. Then modified them for continuos rotation for the Tractor, actually a independent two wheel drive tail dragger, but in the photos it looks like a tractor.
Lots of web surfing, I always get real excited about anything new and read everything I can find about it. In the process I run across David P. Anderson's nbot. A two wheel balancing mobile robot. I now have a target design!
Next I dissected an old Epson printer (720). I even had the title for the web page planed, "Make your printer stand up and walk!", but little did I know. Even thought the printer did have "all the parts", I was not up to the task. It has two bipolar stepper motors that looked powerful, but they are 1.8 degree, 42volt and not that much torque or rpm. It did have two stepper motor drivers built in, but very small surface mount technology and different from each other. The paper drive uses a single chip, LB1847, that might be useful, but the print head driver is a 128 fractional step pair of drivers, A3956SLB. It also has 3 LED/ detector pairs with cables.
So the only part I actually used is the plastic tray under the ink cartridge cleaning. It is the body for Standup.
I also have a Lexmark 22 that ran out of ink. $38 for the printer, but $50 for replacement ink cartridges, so I bought a Lexmark 35 for $59 with two new cartridges. So an extra printer! It is also lacking in the parts department. These new printers have fewer and fewer parts! a larger and a smaller bipolar steppers but I didn't figure out the drivers. It wasn't obvious.
Disgusted with the bipolar steppers and the amount of electronics to drive them, I read about the unipolar steppers in old 5 1/4 floppies at Ian Harries on Stepping Motors . Off to the local PC recycle/junk shop, RE-PC and I now am the proud owner of 5 floppies for $10. Two match the article exactly, the Tandem steppers, so I'm in business right? NO!
I get some drive transistors to make an H-bridge, TIP29, and mount the steppers on a piece of corrugated plastic. These little buggers are heavy! The floppy steppers can just barely carry enough batteries to make them move, but not near the torque or velocity needed to balance anything.
Back to the servos. I mounted the modified R/C servos on the end of the printer ink tray and start writing code for the OOPic. The first thing I notice is that I can't control the speed of the modified servos. More research and I find that servos run at full speed, except for the last few steps near the set point to slow down. And that is not dependable.
Here we go looking again. More web research. I find a hint that I can remove all the electronics from the servo and have a small DC gear head motor. It is even easy to do. I had one to practice on. It seems it is possible to let the magic smoke out of a servo if you plug the voltage in backwards. It is easy to modify it with a little solder sucker.
Now I have the modified servos (DC motors) hot glued to the end of the ink tray. I need an H-bridge again, so this time I find a couple of LM18200s and make a compact motor board. I now have variable speed, reasonable torque and it looks like I'm in business.
I need a sensor to manage the balance or tilt input. An easy turn pot might be just the thing, and I have the ones from the servos! First I glue it to the top of the tray and make a pendulum of two tongue depressors glued together. It just bangs into the side of the body and really gets to swinging from the inertia. It might be workable if I had the software tuned right. So, I move the pot to the bottom an drag one depressor on the floor behind. This will have to be replaced with a "real" tilt input at some point.
Much better, but the pot doesn't change enough in value for the tilt angle. I'm guessing less than one count of the a2d for 2-3 degrees, when maybe 5 counts per degree might be more useful. A scaling multiplier is too coarse.
I've ordered two new surplus DC gear head motors that should arrive today. I hope to have greater torque and rpm to overcome the lack of correct constants in the PID tilt loop.
And that's the state as of July 1, 2003.
© 2003 TCL Services