Focus should have been corrected for sm3, so I might find a slight difference.
Also the report wasn't covering the n-band, which is now available on both spectrographs, I'll add it if time allows.

Just posted early report on throughput stability using quartzes data alf_throughPutStability_20230922.pdf

Continuing my investigation, I added box car extraction to my sanity check
And surprise, r3 suspicious behavior disappeared.

Then I looked at the chi from the regular extraction and those appeared to be very poor in the first visits, including my reference qth.
The problem was in fact that we haven't propagated the July profiles to the beginning of the run, so April profiles were used and we know those are bads.