We then tested the servo, see attached figure. At the beginning of the plot, alpha is at its nominal value of 0.166. We then set it to 0.2 by hand and activate the servo. The servo behaves as it should and drives the error signal (Alp_Lock_SIG_ALPHA) to zero. The gain of the servo is set to achieve a time constant of 200 seconds.
The servo, however, converges to a value which is about 2 percent higher than the nominal value. As suggested by Paolo, this might be solved by setting the right demodulation phase in the medmod procedure. This trick was tested and seems to work. To calculate the exact value of this demodulation phase, we stepped alpha in small increments, see the second figure. I didn‘t have to analyse the data yet.
