Occasionally setup_tracking_params will fail to converge for some bands (see plot below). The cause is often that the value of frac_pp it found is off, and the solution is to inspect the frac_pp for good bands and manually set it to their mean for the bad ones. Although the underlying pysmurf algorithm could be made more reliable, this is a parameter that changes very little and having it be set dynamically is unnecessary.
Max in Slack:
Ya this was brought up by Suzanne, basically it doesn't change unless something goes wrong (like you develop a short). She was requesting we split the dev_cfgs and don't allow such long lived parameters to be overwritten unless done so by hand in a git controlled way or something
I think this configuration is already possible by setting fraction_full_scale in the pysmurf config, which is tracked in the simonsobs/ocs-deployment-configs. Then we would need to make the sodetlib optionally not update that parameter (and not track it in the device config). We'd need to think a bit about operationally how this would get set.
Occasionally
setup_tracking_paramswill fail to converge for some bands (see plot below). The cause is often that the value offrac_ppit found is off, and the solution is to inspect thefrac_ppfor good bands and manually set it to their mean for the bad ones. Although the underlyingpysmurfalgorithm could be made more reliable, this is a parameter that changes very little and having it be set dynamically is unnecessary.Max in Slack:
I think this configuration is already possible by setting
fraction_full_scalein the pysmurf config, which is tracked in the simonsobs/ocs-deployment-configs. Then we would need to make thesodetliboptionally not update that parameter (and not track it in the device config). We'd need to think a bit about operationally how this would get set.