During the early steps of the CARM offset reduction last night it was observed that the ALS DIFF bandwidth was a little low, causing lock losses around the arm power where the transition to the tranmsission QPD sums takes place. To continue locking work the gain was doubled, but it was noted that this should not really change. Today, the ALS diff loop was re-measured, and found to be close to the expected value (under nominal gain settings), so its not clear what happened. At the same time, it was noticed that the length control was ringing at 1.56 Hz, which is sometimes associated with mismatched L1/L3 crossover. A quick measurement of this did not reveal any problem, and somewhat incidentially it was discovered that the alignment of the reaction chain has a significant effect on the length control at this frequency (the reaction chain had previously been adjusted for a different test), potentially due to angular cross-coupling. This should be investigated further.