Day 2 started with a session on Time-of-Flight sensors. ToF seems to be the logical choice for depth sensing, but also for fluorescence detection. The session kicked-off with a kind of overview paper presented by Robert Henderson (although he did not contribute to the paper, he gave an excellent talk!). In this overview three different techniques to perform ToF were compared :
– Buried channel demodulator, which suffers from the fact that a special technology is needed,
– Current assisted photonic demodulator, which suffers from power consumption, but was mentioned to be the best choice for low cost applications,
– The pixel with in-pixel switched cap circuitry, suffering from its complex pixel structure but seems to be the preferred choice in the case of industrial and/or security applications.
Another interesting observation with ToF is their move towards pinned photodiode pixels as well. Apparently most foundries that supply the silicon for these ToF, all have pinned photodiodes available these days. In other cases where still photogates are used, the photogates can be biased negative to get the same accumulation at the interface and to lower the dark issues.
A very nice invited talk came from Masatoshi Ishikawa : “New application areas made possible by high speed vision”. The talk had a lot in common with the one he gave last year at the ISSCC forum. But it was still great to hear it again, also because of the great movies to highlight the capabilities of custom designed high-speed sensors. In many high-speed applications regular devices are used in combination with very highly sophisticated algorithms. But the lesson of M. Ishikawa is the following : exchange spatial resolution for temporal resolution and the algorithms will become much more simple. For normal sized robots a frame rate of at least 1000 fr/s are needed, for micro-machined type of robots 10,000 fr/s is a must.
In one of the papers the use of ToF with a standard CMOS 5T pixel was illustrated. The sensor originally was not intended for this application, but nevertheless, it is possible to detect depth. The results are not up to the same quality level as the specialized ToF, but a bit more optimization of the sensor and/or timing could lead to better results. A very interesting side-effect was mentioned during the Q&A : the ToF mode of this device can be used to measure the delays in the sensor’s metal wiring.
Dark current, dark current non-uniformities, hot pixels, they all seem to be a joy for ever ! Of course we all knew the positive influence of having an accumulation layer at the interface, but a couple of papers illustrated how you can do this dynamically in CCDs as well as with the transfer gate of the CMOS devices. This is absolutely the way to go if the sensors need to show dark currents close to their theoretical limit. Although, it was mentioned in one of the papers that over time, a negative bias of the transfer gate can introduce some nasty aging effects.
Day 2 ended with a 25th anniversary talk of the workshop given by the father of the workshop himself : Eric Fossum. The talk started with the statement of Bernard of Chartres (1115 AD) “We are like dwarfs on the shoulders of giants, so that we can see more than they, and things at a greater distance, not by virtue of any sharpness of sight on our part, or any physical distinction, but because we are carried high and raised up by their giant size”. This was a quite nice start to acknowledge the contribution of several great solid-state imaging pioneers that prepared the path on which all of us are walking these days. The older images and papers that Eric showed brought back good memories of several previous workshops.
Tomorrow more news.
Albert, June 11th, 2011