Max F., Keita K., Daniel S.

We had a hard time finding the beat node by adjusting the crystal temperature. Once we came close to the correct frequency the line would simply vanish. This seems to be related to mode hopping. The diode current was about 1.43A. We noticed that increasing the diode current by simultaneously adjusting the temperature to keep the beat frequency close to 40MHz allowed us to widen the acceptable temperature range and improved the beat node strength. Around 1.9A there was another mode hope, so we backed off the diode current to 1.666A. This re-adjustment of the diode current seemed to have increased the gain in the PLL. We had to reduce the common gain to -23dB (down by 3dB) to make it stable again.

We also found that the gain in the digital temperature stabilization servo only has a small range where it would work. Adding 10dB would make it unstable, whereas subtracting 10dB made it go to the rails. Even with the correct gain, we noticed that the PZT voltage would often go beyond +/-1V. Our readback has a gain of 10 and saturates just above 1V. (This readback is used as the error point of the temperature stabilization servo.) The following changes were made to the ASL common mode board, S/N S1102638 (H2 EY): 
R272 removed (from 374), 
R271 replaced by 0 Ohm (from 3.32K). 
The readback now has a gain of 1 and the temperature servo should work over the full PZT range.