Monday, June 23, 2008

New Sony sensor

Sony has developed a new type of digital camera sensor, claimed to be more sensitive.


Damien said...

It's funny how the current design of sensors just came so close to the biological way of doing things.

The human eye also has its "wiring" of blood vessels and neurons in front of the light-sensitive cells.

You can't see it because the neural system of vision is only sensitive to change and adapts in a matter of seconds to remove fixed features as noise (a bit like dark frame substraction in digital capture ;-) But...

If you stare at a blue sky for a while (no, not at the sun), you'll soon notice a hundred of tiny bright spots zipping around, always following the same trajectories. These are your white blood cells !

Bert said...

They fail to mention the substantial cost reduction that this structure can also bring. The traditional mounting method of a chip on a PCB requires the device to be mounted in an (relatively) expensive package that provides connection leads from the top of the chip to its back, linking the on-chip metallizations to the underlying PCB.

In recent years, we have seen a slew of efforts resulting in many cost-effective "flip-chip" packaging technologies, in which the chips are mounted face down on the PCB (or a glass substrate, in the case of an LCD). This more-or-less direct bonding of the die's metal pads to the PCB eliminates the need for a lead frame, resulting in substantial cost reductions and performance improvements.

The new sensor structure announced here lends itself directly to such packaging technologies. In fact, I wouldn't be surprised to learn that this was the initial motivation behind the research that led to this new sensor... ;-P

eolake said...

Sounds sensible.

It also sounds like this is the obvious way to mount things, I wonder why it was so hard.

Bert said...

"It also sounds like this is the obvious way to mount things, I wonder why it was so hard."

The active regions on a semiconductor die are normally just a few atoms thick, and the die itself is nearly opaque. It is only in recent years that new processes were developed to "sink" active elements deeper into the die, in order to minimize the surface required.

Now, either they have reached the point where diffusions are deep enough to traverse a thin wafer, or they have devised some scheme allowing the interconnection of active areas located on opposite faces of the die. I would tend to believe it's the latter, from a visual hint in one of the images in the article, but I really don't know.

But no matter what solution they are using, it does represent a leap in thinking from traditional methodology. Requires imagination because it's not obvious, hard to implement because it goes against the established way of doing things, and thus even harder to sell to managers. That's why such leaps are pretty rare.jsnzxi