Masayuki, Kiwamu,

With the two functions that we implemented yesterday (alog 25316), today we tested the automation to see how they improved the engagement of the second loop.

It locks without a fail out of more than 20 trials within approximately several seconds every time. Very good.

On the other hand, I have noticed that I was missing some signal conditioning filters in order to handle multiple error signals in a user-friendly way. So we need another round of modification on the front end model which should finializes this implementation.

I reverted the settings back to nominal. So the guardian still handles the engagement as usual.

[Advantage over guardian in this case]

The front end code allows for a fast and more complicated servo filter. This is the biggest advantage and therefore the key point of the succeess in the test today.

The reference signal or offset of the second loop has been adjusted by the IMC  guardian which employed a simple PID for servoing the reference signal so as to minimize the "acquisition kick" during the engagement. Looking at the behavior of the suppressed and unsuppressed signals, I found the UGF of this loop to be lower than 0.1 Hz. At one point in the past, this servo became insufficient due to excess intensity fluctuation imposed by the IMC. The IMC seemingly adds extra intensity noise below 1 Hz presumably via misalignment. As a consequence, the guardian servo became unable to keep up with such a large and fast fluctuation. We started assisting the guardian by the manual engagement (alog 22449) where the operator activates the second loop when the second loop error signal momentarily crosses zero.

The new implementation tested today was able to achieve a UGF of 0.8-ish Hz which sufficiently suppressed the fluctuation and therefore a smooth engagement of the 2nd loop with a help from the trigger. In the digital filter, I had to install 3 pairs of poles and zeros ( i.e. zpk([0.03;0.03;0.05], [2;3;3], 1) ) in addition to an integrator in order to compensate for a 0.1 Hz second order analog low-pass. Currently the UGF is limited by the DAC range. The rms fluctuation monitored by SECONDLOOP_SIGNAL reduced by roughly a factor of 5. The dominant component was from around 0.1 Hz and can now be well suppressed with the fast servo.