OpenServo.com

[ginge]

Hi all,

A couple of people, myself included, have been wrestling with i2c bus data corruption, and generally odd read errors. At the extreme of these corruptions, the I2C host loses arbitration, and the AVR gets stuck in an odd error state. (I don't know enough to say what is really going on)

There is a very simple solution to all of these data corruption errors.

Very effective noise suppression can be achieved by soldering a 10nF ceramic capacitor across the terminals of the motor.

This completely cleared up all I2C issues, enabling me to do over 10000 successive block reads while the servo was in motion.

I have spoken to Mike over this issue, and he is going to add provision for a small SMT ceramic capacitor between the motor terminals in the next PCB revision.

It's been a hard problem to track down, with a very simple solution.

Barry

[DKNguyen]

Isn't that already done in the OpenServo schematic? (Or does the schematic use 0.1uF=100nF?)

[mpthompson]

The capacitor Barry describes is directly across the terminals of the motor within the H-Bridge. My guess is this catches higher frequency noise that the other capacitors miss.

-Mike

[ervo]

Well, I must say that I am not alone on this one. I have also been a victim or I should rather call myself a fighter here because I still have no clue whatsoever has to be done here to get rid of the corruption based on the i2c bus data. The solutions being shown here, however, really shows that the health of the ceramic capacitor is at stake. I do wonder as to what could be the possible outcome from this!
_________________
CAAS