> Perhaps the PI anti-windup change could be re-visited. When the output is > clamped and anti-windup is active, at xmlauto.cxx:625, the line: > int_sum = clamped_output - prop_comp; > seems to introduce the Kp, proportional gain factor, into the integral > sum. With the proportional gain much bigger than Ki the integral part and > proportional part > effectively cancel after this, with the result: output close to zero. > Replacing that with : > > if( output != clamped_output ) // anti-windup > // int_sum = clamped_output - prop_comp; > int_sum -= error * Ki.get_value() * dt; > > .. seems to restore altitude hold to at least the B777 although I don't > know if that's the optimal correction to the integral term during output > clamping. Thanks. > I used a slightly different approach to fix this. The problem occoured when the proportional amplifier was driven into the nonlinear range and it's output was clamped. This resulted in a feedback into the integrator. Clamping the amplifier before adding the integral sum should solve this issue.
Can you check, if this works for you? Thanks, Torsten ------------------------------------------------------------------------------ Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev _______________________________________________ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
