FastVision published datasheet of its recently announced FC300 camera featuring Panavision 3.2MP/180fps sensor with optional JPL BSI processing. White its peak QE is a respectable 60%, the whole spectral graph looks quite interesting:
Cannot understand why there is a dip in the QE curve in the 500 range. Very interesting curve indeed. Maybe a problem with the filter change during measurement?
The data sheet claims that the chip has 5um pixels and is capable of either rolling shuttering or global shuttering.
But if it has the global shutter... well, then it must have a shadowed area where the sampled signal is stored...
so if the pixel has an intentional shadow (metal shield), then why bother with the back thinning at all? it would be better off with a nice microlens on the front side and save the $$$!!!
Even with a micro lens , your fill factor is not 100%. Further it still imposes metal routing constraints, all of which are solved by backside illumination.
Something is odd. The specs on the camera match the specs on the Panavision website for the DYM0032 sensor (not BSI) except that the sensor datasheet quotes 51% QE at 600 nm. See - http://www.panavisionimaging.com/PDF/PB0030.pdf. If the whole depressed section were moved up 20 points so the 600 nm matched, the curve would look a little less strange.
No AR coating and/or a reflector?
ReplyDeleteDecent mid-band QE for a 5um x 5um pixel without either (if indeed that is the case)
DeleteHi,
ReplyDeleteI wonder if the 180fps are correct as the sensor datasheet (DYN0021) claims only 120fps for full resolution.
Any idea?
Cannot understand why there is a dip in the QE curve in the 500 range. Very interesting curve indeed. Maybe a problem with the filter change during measurement?
ReplyDeleteAR coating problem, maybe?
DeleteI am a bit stumped...
ReplyDeleteThe data sheet claims that the chip has 5um pixels and is capable of either rolling shuttering or global shuttering.
But if it has the global shutter... well, then it must have a shadowed area where the sampled signal is stored...
so if the pixel has an intentional shadow (metal shield), then why bother with the back thinning at all? it would be better off with a nice microlens on the front side and save the $$$!!!
perhaps that is why the QE curve is so funky!
Even with a micro lens , your fill factor is not 100%. Further it still imposes metal routing constraints, all of which are solved by backside illumination.
DeleteSomething is odd. The specs on the camera match the specs on the Panavision website for the DYM0032 sensor (not BSI) except that the sensor datasheet quotes 51% QE at 600 nm. See - http://www.panavisionimaging.com/PDF/PB0030.pdf. If the whole depressed section were moved up 20 points so the 600 nm matched, the curve would look a little less strange.
ReplyDeleteI think that this curve is not serious... don't wast too much time on it!
ReplyDelete