CN FORUMS ARCHIVE Telescope Reviews: new CMOS sensors in DSLRs

Sorry can be very boring to some......

Mike,

Here is my summary on that why this Sony IMX021 (12-bit built-in A/D) can do/is possible doing 14-bit A/D. It's an interesting detective story:

In layman's term:

Theory 1:
Sony used the IMX021 as the basis and then made a Nikon custom for Nikon D300 DSLR; in another view, the Nikon IMX021-like imager is a full featured one, while IMX021 (as used in A700 and available for commercial sale) is a crippled 12-bit version...

Theory 2:
Sony still provides Nikon with the IMX021 but it internally has Nikon custom-feature, not be disclosed to other DSLR mfgs, includign Sony camera division. I.e., trade-secret approach on selective enable/diosable features.

Theory 3:
{It is known that IMX021's internal A/D is a column-parallel A/D design, using one D/A generated ramp as the flash-comparator reference to do column-wide flash A/D.]
Thus there is a way to use an external 14-bit D/A converter (instead the internal 12-bit D/A), via external loop-thru (a.k.a. side door), the 14-bit one can patch in to replace the internal version.
I.e., external circuitry partially modifying SONY imager behavior.

Theory 4:
Although IMX021 has internal 12-bit A/D, it also has the traditional one-channel (or more) analog output method. Nikon can stop IMX021's internal A/D section, but just
uses N's own/separate external 14-bit A/D IC to process imager's output, just like the traditional CCD's processing method.

=========================================

My take on 4 theories:

Theory 1:
Conceptually makes sense, but the smoking gun:
it is highly unlikely, since bond pad and (IC chip) dice pattern is exactly the same the the IMX021.

Theory 2:
Secret microcode or undocumented features to cripple:
Conceptually possible but realistically, this approach can be reverse-engineered in no time. IMO, high sum of investment money shall not be protected by boys or girls scout's honor code

.

Theory 3:
Use external 14-bit D/A to replace internal 12-bit D/A function.
This is feasible, but is a little bit complicated, since it is highly involved on in depth co-working in between two firms:
- need to make sure internal comparators are workable down to 14-bit precision
- need to find a way to alter the timing sequencer
- need to "enable" (if there is such a whole column of pre-constructed/pre-allocated 14-bit counters) 2 extra bits of the column counters.
In my view, this "side-door" approach is far too complicated and requires detailed involvement. It will also have many service/support issues. Given that "N" is also working on its own JFET/CMOS sensors, "S" wouldn't go that far to be that cozzy.

Theory 4:
Just tell the imager not to use its own internal A/D, but simply passes the analog sensor signal out as-is and let the external circuit to take it from there.

Smoking guns:

4a) (indirectly) in ISSCC 2006 conference, Sony has 2 papers on column-parallel CMOS sensor technology, though more on the smaller size lab version of the IMX017CQE (12-bit internal column-parallel A/D converter) CMOS image sensor. In the table describing its features, it has one line: base feature #4: one-channel (analog) output
{yeah! Not forgetting its CCD root

}

4b) (indirectly) in D300 and D3 (D3 has 14-bit RAW) main circuit board pictures, there are many common components, one smaller IC (same part number) looks like an AFE (analog front end, a fancy name which has A/D inside). Note that D3's fullframe sensor is Nikon in-house developed and fab sensor with 12-channel analog output. This means in main circuit board, there must be an AFE IC. Since the boards are so similar, it makes sense why not just use the AFE IC to perform external A/D function anyway.

I managed to convince myself. That's good enough.

===================================================

Why I started to mention some APS-C CMOS activities:

1. IMX021 is a commercial available IC, sampling nwo and unit price is $400
2. From few recent pictures, its sensitivity seems to be topping Canon's by one stop or two. I.e., Sony seems to have a way to control noise (in room temperature)
3. IMX021, in native 12-bit digital output, is a digital only IC. This makes camera-head type of imager design extremely easy, just like the recent simple LPI/SSI/Q-Guide cameras etc.
4. There may be new sensors being introduced, e.g.1, new influx of funding on Sony sensor R&D and e.g.2, ISSC 2006 Sony has a paper mentioning its own back-illuminated CMOS sensor (not seen commercially yet) ...
It's going to be a new era.

Hmm... I'd better start my part 11 article on the CCD Intro topic and devote it to some of the CMOS technologies ...

Clear Skies!

ccs_hello