a BSI sensor riddle

Started 11 months ago | Discussions thread
OP Bernard Delley Senior Member • Posts: 1,061
Re: a BSI sensor riddle

bclaff wrote:

Bernard Delley wrote:

bclaff wrote:

Bernard Delley wrote:

I would have thought that the technical challenge of back illumination was taken for the benefits of improved light gathering. So I would expect a higher quantum efficiency for a BSI sensor.

For very similar pixel sizes, Bill Claffs DXO derived sensor characteristics shows

Nikon D500 (standard sensor chip) QE = 60%

Nikon D850 (back illuminated sensor) QE = 44%

That lower QE is counterintuitive by my expectations see above. So what is wrong ?

Covered in this excellent thread: https://www.dpreview.com/forums/post/60222399

I found a lot of sidetrack to colors. Your answer was to the point and among the few interesting answers.

Thank you.

And, FWIW, my values are a bit closer D500 at 57% and D850 at 49%

today I still find 60 and 44 % respectively on your DXO derived sensor ... page

You misunderstand. I don't consider the DxOMark numbers "mine"; they are simply derived from the DxOMark data.
The 57% and 49% are my own measurements; which are not currently published anywhere at the site.
The reason that my QE values aren't published yet is that some of my values are clearly pretty far off and I haven't put in a good mechanism to handle excluding individual results.

QE is hard to measure without a sophisticated setup (that I don't have).

If you have two cameras with the same mount, and a stable light source, it is not too difficult to do a side by side comparison keeping all else the same in a controlled way.

For white light and the same exposure, I find for example 300 electrons/ mum^2 for D500 and 230 for D850. so the relative QE are like 60% to 46% according to my measurement. I think that is to better accuracy than 1%.

I you care to continue this discussion, please do on the other thread which I started in the right forum. Nikon lens talk forum happened by my mistake, totally wrong forum for this.

I understand this concept. I actually handle very few of the cameras that I test.
This type of procedure is beyond the capability of almost all my collaborators so it's not part of the testing protocol.

well some of your contributors might be able to test several cameras for their relative QE with good accuracy along above lines.  That might give you a database where you can check consistency.

Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow