on sensor ADC?

I'm posting this as a question, as I would love to better understand how the latest generation of sensors, particularly from Canon, are realizing better dynamic range even on APS-C sensors. I have read only a little about the move to on sensor analog to digital conversion. My assumption is that a shorter pathway for data from light gathering to data value permits less inherent digital error / noise. Is this what has allowed the 5DIV, M5, 80D to realize better DR and overall less noise at higher ISOs? Appreciate any info!

Hi Roger,



I am curious where you found these claims?

First, most of the stuff you read on the internet is speculation and in most cases has no technical merit at all (alternative facts!). Only and only Canon engineers know the design details of the image sensors in Canon camera and those folks are not commenting on any internet forum.

Second, ALL ADC's are on-chip. What the internet forum readers and writers often confuse and do not understand is the so-called column-parallel ADC design which is a design feature in select Sony sensors. It has nothing to do with "data path" but it relates to how the sensor subtracts the dark current or the leakage at each pixel from the actual signal. These sensors perform better than sensors that use a discrete CDC (correlated double-sampling) circuit when it comes to fixed pattern noise that limits the DR in the deep shadows of low ISO images. Explaining the details requires some background in electrical engineering and physics and is beyond the scope of internet forums. If you are interested use google scholar to search for the keywords mentioned above to read the technical literature on this topic, they are mostly in the proceedings of IEEE's ISSCC conference.

Canon's sensors design is heavily guarded by NDA's and trade secrets, they don't publish it and no one really knows except for the engineers in Japan. I haven't seen any papers from Canon in the past 5 years published in the public domain...

hope this helps

Many thanks Arash for the reply. This is really helpful. The topic came up at work (I work in the digital library unit of UC San Diego Library). We toss around topics like this occasionally and often as a result of my interest in photography outside our studio :-) A cursory review of stuff on the web fell into exactly the category you describe; not very thorough and not authoritative. Your answer really does provide a base of understanding for the issue. Thanks again for the response!

Quote:

---

Originally Posted by Roger Smith

I'm posting this as a question, as I would love to better understand how the latest generation of sensors, particularly from Canon, are realizing better dynamic range even on APS-C sensors. I have read only a little about the move to on sensor analog to digital conversion. My assumption is that a shorter pathway for data from light gathering to data value permits less inherent digital error / noise. Is this what has allowed the 5DIV, M5, 80D to realize better DR and overall less noise at higher ISOs? Appreciate any info!

---

Probably you are referring to this interview to Masaya Maeda, which at the time was the senior managing director of Canon inc.

http://www.imaging-resource.com/news...s-new-printers

However, the informations provided publicly by Canon are very limited, and as far as I know there are no "official" details regarding how the analog to digital conversion is performed.

For example, in this interview to one of the developers of the 5d mark IV sensor, it is (partially and informally) explained how they improved the SNR on the new sensor, but no mention to the dynamic range.

https://snapshot.canon-asia.com/arti...4mp-resolution