Design Ideas

[mack33]

Okay, a little background:
I have a bachelor's in Electronics and Computer Technology and I minored in computer science. So I can understand allot of what will be said. Obviously i didn't learn much about building robots.

What I want to do:
Make precision robots with as little "out of the box kits" as possible. I would like to start by building small scale precision robots and eventually work on large scale precision robots (robotic arms).

Questions:
What would be the best way to determine motor rotation or axis position. (stepper motor, encoders)
Any suggestion for micro controller starter kits. (Unfortunately I don't have a pic programmer yet.)

[szhang]

"precision" robots are hard to make, small ones even more so because of the smaller margin of error.  What exactly do you mean by "precision"?

Encoders are usually the best way to get motor positions, you can get optical or magnetic ones that have >8000 counts per revolution, though for small robots that don't require that much precision servos are alot easier.

you have two choice with microcontrollers, you can buy a control board like Axon or Arduino, or you can get a PIC/AVR/dsPIC and program them and make your own board.  The latter is obviously more difficult but slightly cheaper, though you don't need a programmer for microcontroller boards like Axon or Arduino.

[mack33]

By precision, I mean that the robot actually knows approximately where it is. Not a robot that allows it's motors run free.

For instance, let's take a simple robotic arm. It would be cool to have a robot that I could program to move it's parts to certain locations.

For instance:
rotate on it's base x degrees (or radians)
extend it's arm to point y
close it's gripper

I have had a little experience with encoders and have some concerns.
First, the only experience that I've had with encoders is an optical encoder that sent pulses along wires when it hit clear spots on the disk.
If something like this is available and I choose to use one, I would then have to connect it to either a sub-circuit to count the number of clear spots transmitted or I would have to find a micro-controller that has a dedicated counter input. Unfortunately, the latter only allows for a few encoders.

So the next question:
Where can i find more information about encoders? (theory, projects)
How could I configure them to relay information to the micro-controller?
How reliable are mechanical encoders?

[Razor Concepts]

Starting out by building desktop sized arms using servos might be a good way to fine tune your construction skills. You don't know what position the servo is at any given time, but as long as you tell the servo "go here and stay there" you can be assured that it is at that position (or will get there eventually). Since you have a lot of experience in electronics, I bet you could wire up the potentiometer in the servo to the microcontroller, so you know what position the servo is at.

[szhang]

I have had a little experience with encoders and have some concerns.
First, the only experience that I've had with encoders is an optical encoder that sent pulses along wires when it hit clear spots on the disk.
If something like this is available and I choose to use one, I would then have to connect it to either a sub-circuit to count the number of clear spots transmitted or I would have to find a micro-controller that has a dedicated counter input. Unfortunately, the latter only allows for a few encoders.

Uhm... no, any digital pin can be used to count pulses.  You only need to use the capture mode if you want to do the counting purely in background.

Also, the encoders you are talking about are incremental encoders (they only give you the change in position).  There are optical and magnetic absolute encoders.  Or you can just built a very small circuit to convert the incremental data to an absolute position.

[Admin]

read up on 'quadrature encoders'