Corrupt clusters in FT decoding

[This issue is being discussed internally in the SciFi team]

Currently in data-taking we have some corrupt clusters, which are all large clusters. This comes from the fact that the (size,frac) bits are set for them in a specific, predetermined way, which encodes the start and the end of a large cluster. Currently, the FPGA clustering does not do it the same, which results in all large clusters being ignored. This drastically reduces our hit efficiency.

The error message displayed when using FT/FTDAQ/FTRawBankDecoder is:

FTRawBankDecoder                      ERROR Possibly corrupt data. Ignoring the cluster.

Fixes will be checked in both Allen and Moore.

Road to being fixed:

• documenting exactly both the expected and actual (size,frac) flags on large clusters;
• transmit to the FPGA team a document with requirements
• implement the new clustering.