Clusters with impossible addresses
Each day comes with its batch of problems...
During runs at large scale (i.e. p5), it was discovered that GEM sends clusters with an impossible address to the OTMB and EMTF, that is, an address between 192 and 254 (inclusive). The contamination rate is small.
Still, the behavior is entirely unexpected and it is unclear where the issue is coming from. Both the OTMB and EMTF DQM show the problem. Work is ongoing to check whether those rare events are also seen by GEM or if the link quality could be the source of the issue.
In any case, the way to go would probably be the following:
- In the triggering algorithms, use only the clusters with a valid address (i.e.
< 192
). - In the DAQ path, save all clusters (without ZS) or clusters with an address
!= 255
(with ZS).
That raises two questions on how the trigger data is treated on the GEM side:
- How are the trigger/clusters rates accounted? Only clusters with an address below 192 should be included.
- Which clusters are save in the monitoring registers (S-bit monitor and cluster DAQ)? All clusters but the ones with an address equal to 255 should be recoded.
What do you think? (The values are to be adapted for GE2/1 or/and ME0.)
Now, some plots! The last version of the CSC local DQM clearly reports clusters with an impossible address.
The EMTF DQM shows VFAT out-of-range. This is due to how the VFAT number are computed from the eta and phi coordinates.