OpenServo.com

[mpthompson]

Jim Frye of Lynxmotion is considering a commercial servo product controlled by the OpenServo software. One considered feature is that his board might contain a USB-to-Serial chip his customers would use to control and update the OpenServo software from a PC. With this in mind, I wanted to open up a forum thread discussing what would be needed to implement async serial control of an OpenServo.

Some random issues off the top of my head:

1. The functionality of the twi.c module (the OpenServo I2C communication module) would need to be replaced with an equivalent serial module. I basically would want to see a serial module as much as possible parallel register based programming and control model of the OpenServo. This will keep the two methods of communication in sync for common development and keep the parser that interprets commands form the PC relatively simple.

2. We would need to think about a revised serial based bootloader. Once again, it should mirror the functionality as much as possible of the current I2C based bootloader.

3. By default, the OpenServo using the internal 8MHz clock of the AVR ATmega168. I've done serial communications with these chips and the 8MHz clock doesn't divide well into common async baud rates. This makes communication a bit dodgey at higher speeds (i.e. faster than 19.2kb). A serial based OpenServo would probably require an external crystal oscillator with a frequency that divides evenly into aysnc baud rates for reliable communication. Changing the clock speed of the OpenServo would have some implications across the various peripherals that are timing sensitive.

4. Commercial robotics servos tend to have a modified half-duplex TTL level serial protocol whereby serial servos can be daisy-chained, similar to what we do with I2C already with the OpenServo. Would we want to explore similar capabilities? It would take more effort to implement.

Any other thoughts on serial control of an OpenServo?

-Mike

[wurpfel]

HI Mike


the serial communication has advantages, but not many...
the greatest drawback is the point-to-point topology for the most protocolls.


best canditate for implemetion is the EIA-485 protocoll.
works with 5V, up to 32 clients, fast enough and up to 1,2km distance (in theory )


the PROFIBUS is a derivate of this for industrial use..
but when serial then better opt for CANbus
_________________
not to old to play with scrap