Working with MPCCD images
We use the MPCCD Octal detector for SFX experiments. It consists of eight sensor panels (or modules).
TODO
Since April 2017, the compact MPCCD is in operation. This fixes a bug in older MPCCDs that led to the loss of
intensities when saturation occurred. In the compact MPCCD, the total intensities are preserved when spilt
pixels were summed together. However, it has X-ray shields (20 - 25 px wide) over amplifiers located at the edges,
so the effective pixel area is slightly smaller than older MPCCDs.
If you use higher energy (>= 10 keV), you might want to request the phase 3 MPCCD.
Its sensor thickness is 300 nm, five times thicker than other MPCCDs.
This provides higher DQE at higher photon energies.
((Actually, its back-thinned layout provides higher DQE even at 7 keV, but the difference is
smaller.))
The edge pixels (5 px) around phase 3 MPCCD are unreliable and should be masked.
Moreover, the sensor thickness may cause a parallax effect that leads to radial spot elongation at
high angles. This can in principle be modeled in DIALS, but not in the current version of CrystFEL.
The pipeline automatically defines appropriate masks in the CrystFEL geometry file it generates.
When beam line engineers deploy the detector to the DAPHNIS platform, they do their best to place the detector perpendicular to the beam. However, some errors are unavoidable.
Unfortunately, geoptimiser cannot refine tilt and twist, although the latest version (0.6.3) of CrystFEL
allows the z component in the detector fast and slow axis. In principle, we can
refine tilt and twist in DIALS and convert the geometry JSON file to the CrystFEL format.
We have not tested this yet.
In practice, tilt and twist are absorbed into other refined parameters during geoptimiser.