Skip to content

Finetuning for LIN

Michael Daas requested to merge finetuning into development

This MR fixes binary search tuning in tune_local_threshold.py, finally yielding the same results as mete_tune_local_threshold.py in a fraction of the time.

Full tuning of the linear flavor now takes about 45s, yielding the following threshold distribution at an absolute threshold of 1000e: image

In comparison, the meta tuning yields the following distribution in about 15 minutes: Screenshot_from_2019-10-18_18-17-32

Note that the meta tuning converges about 15 DVCAL too high (target value was 80).

Additionally, with this MR, most tuning algorithms now support scanning over all FEs in a single run. This finally makes the use of the new meta scripts meta_tune_threshold.py as well as meta_tune_threshold_and_tot.py for full tuning of the full chip possible.

Furthermore, it fixes #269 (closed), #291 (closed), #303 (closed) .

Edited by Michael Daas

Merge request reports