ok, thanks for your answers.
I imagine that the controller limits the throttle rising slope (but not the falling slope) in software, implementing a ramp-up.
whenever the low voltage comparator triggers the throttle is zeroed, and from there it must slowly ramp up till it cuts again, which takes about 1/3 of a second. if a filter is applied at the output, only a small amount of throttle will be reduced per throttle sample period, and once the low voltage condition subsides the throttle will ramp up from there, not from zero.
so this means two things:
1) the filter must block from frequencies in the neighborhood of the sampling period (thousands of Hz?), not frequencies near the current oscillation period (3Hz), so you shouldn't be concerned with this:
> I think if capacitor is too big the power will continue a bit when I will release the throttle.
the delay should be in the order of hundreds of a second, completely negligible. (otherwise, a transistor could be added to discharge the cap when the throttle is lowered.)
2) the oscillations will get smaller in amplitude (and thus higher in frequency, since it's a ramp) the more you filter, and can be as small as you want.
so, do this....
-replace R7with a 1K resistor (or add it after R7 (in series), whichever is easier, since it makes no difference whether the resistor is 1K or 1.1K).
-add a capacitor from the motor controller side of R7 (the "Tout" signal) to ground.
a 1.5uF (microfarad) capacitor with the 2K resistance (R6 + R7) will result in a 53Hz cutoff (but I don't know the source impedance of the throttle sensor, which should be added to the 2K!). this should keep the response time of the throttle below 100ms, which seems fine. a 1uF cap would result in a bit less filtering and faster response. I would go with a 1uF cap and raise it to 1.5uF if I still find any oscillations, which is unlikely.
you can get standard aluminum electrolytic caps (and resistors too) from any old radio, cordless phone, music playing equipment, etc. they look like little cans with two legs. just be sure the cap is visually in good shape and that you don't overheat or mechanically stress it while desoldering it.
the electrolytic caps have polarity. the negative lead must be connected to ground in this case.
WARNING: be sure you have the break cutoff connected before testing this mod. should there be a problem, you can pull the break and avoid disaster.
this mod will probably get you going.
good luck!
