The Lynx read noise is said to be below 4e- at rates up to its full 100 fps with a power consumption under 200mW. The full well is >25Ke- with image lage <0.1%. The output resolution is said to be 10b at 60fps and "digital output" at 100fps.
Goossen Boers, CEO of PHOTONIS, commented, "The ground-breaking Lynx CMOS digital sensor is a pure product of our Night Vision expertise. It differentiates itself from other sensors in the market by the perfect combination of large pixels collecting the maximum amount of photons, low noise, low power consumption, very high sensitivity and extended wavelength range. The new sensor reinforces our continued commitment to the low light imaging market as well as to our innovation and new product development."
Here is some data from Lynx datasheet to support the claims (judge by yourselves):
Vladimir, 15mLux is not Twilight! It's half Moon.
ReplyDeleteHow compare this sensor to SCMOS from Fairchild?
@ "Vladimir, 15mLux is not Twilight! It's half Moon."
ReplyDeleteThe text in picture is not mine. You can report it to Photonis.
sCMOS pixel size is 6.5um, about half an area of Photonis pixel. Large area is a killer advantage in low light.
large area is clearly an advantage. However, sCMOS with the micro-lens should make up for the reduced area. This sensor has some decent performance at 100 FPS.
DeletePhotonis QE of over 50% suggests that its fill factor is quite high. The QE is just a tad lower than sCMOS' one:
Deletehttp://www.scmos.com/downloads/
So, the area advantage still mostly works in spite of sCMOS microlens.
What happens if you enlarge the 1.4um pixel used in iPhone to 10um? If SONY jumps in this business, can they blast all the players??
ReplyDeleteGood question. First, I doubt that the 10um pixel market size is big enough for Sony, Toshiba, Aptina or others to develop a process derivative for that.
DeleteSecond, what type of improvement do you expect? Higher QE in IR? Better MTF? Lower price?
simply the high sensitivity and on-chip processing. All these CMOS have a lot FPN, hot pixel, etc. An additional FPGA is needed.
Delete10x bigger pixel gives you 1/10th total pixels for the same area. Hard to sell 0.8MP vs 8MP. What application do you have in mind where you need to directly use unprocessed images? (Try CMOSIS 5.5um px with CDS)
Deletethe area ratio between a 9.7u pixel and a 1.4u pixel is 48X... an 8MPix sensor is actually reduced to 0.17MPix for the same area chip with 9.7u pixels. How much of a market would there be for a 512x320 device?
DeleteI can't see the price of the device dropping since the volumes would be miniscule! But pushing the resolution to 8MPix would make the sensor ~35mm x 22mm which is nearly the size of a full frame 35mm DSLR device. (36x24) How cheap are those?
Finally, don't expect the same 2-3e- noise performance from a 9.7um pixel, nor a 48X higher full well!
The last reported works on sub-electron noise (Kawahito ISSCC11, Lotto ISSCC11, Theuwissen ISSCC12) have quite large pixels..
DeleteDo you know who designed the CMOS? How is manufacturing the CMOS? What process?
ReplyDeleteCMOSIS design the CMOS.
DeleteLynx sensor performances looks worse (QE, RON...) than the ones featured by Photonis InXite device (15 µm pitch): see http://image-sensors-world.blogspot.fr/2011/11/photonis-announced-inxite-sensor.html).
ReplyDeleteOne for European/export market and one for US market?
the devices were probably developed by two different design groups, most likely use different fabrication processes and appear to be targeted for somewhat different markets! (Scientific high speed for the 15um InXite and night vision for the Lynx)
DeleteI am sure if the customer has enough money, both sensors could be made available on both sides of the Atlantic!
While it is true that both sensors address low light image markets, the InXite chip is made for the scientfic market and not the surveillance market. It can run up to 1000 fps and the processing power necessary for those high frame rates requires a much larger camera chassis than is practical for surveillance. Lynx was designed to operate at ranges from daylight to quarter moon (typo on datasheet noted!) with low power consumption making it ideal to be integrated into a small chassis or unmanned application. Both chips are designed for commercial applications and are available on both sides of the Atlantic (and anywhere else).
DeleteWhen you can design an exportation restricted CMOS sensor, then your design skill is high enough! :)
DeleteIt can be very useful in amateur astronomy
ReplyDelete