Large Detectors
Angelo's slide from May 2023 (see below for full talk):
for epixhr (ignoring sparkpix)
estimate 100 lr4 fibers for 16Mpx 30kHz assuming 25Gbps per pgp lane (per dionisio)
16Mpx*2bytes/px*30kHz/2.5GBpers/4colorsperfiberpair~=100fiberpair
need drp to be in the same bos, so won't fit in one current bos
how many cameras do they want to run simultaneously?
ask calient about their roadmap? big bos?
what is sparkpix data reduction factor? will ask hutch scientists.
For sparkpix:
- 0.5Mpx in XPP
- 2Mpx+0.5Mpx in DXS
- 0.5Mpx in MFX (can instead be epixHR, but not both)
Diling writes about occupancies in XPP:
"I think for high rep rate XPCS 1% as Angelo mentioned previously is a good starting point as a design goal, unless we are hoping for a lower number? It also really depends on how much processing can be done on the detector to further reduce the data. I can think of many on board reduction options, but I don’t know when we will get to a design phase when we start talking about writing down specifications. If all it does is threshold and readout pixels above, then I’d stick to the 1% number for now."
So data volume: 0.5Mpx*2bytes/pixel*1MHz*0.01occupancy=0.01TB/s or 10GB/s which feels doable.
Presentations
(old, from 2020)
(new, from 2023)
Overview
Content Tools