[beholder3]

I just read an interesting comment on back-side-illuminated sensors.

The author explains that with non-BSI sensors the heating effect of longer exposure times on image noise mostly is not relevant for shots < 1/10 second.


He explains this with the dark current being 0,5 to 1 e-/Pixel/s at 20°C.

His other comment is that BSI sensors can have as much as 4 times that dark current (2-4 e-/Pixel/s).

I gather that for both astro and other long-term-exposure-in-the-dark shooters any BSI sensors are not a good idea.

[joergens.mi]

can you give a link to the source?

Why is the dark current 4 times bigger and is it normalized to the area of one pixel.
Because all the connection lines are vanishing from the active side, the pixel size is increasing significantly.

In the german language Wikipedia ( https://de.wikipedia.org/wiki/R%C3%BCckw%C3%A4rtige_Belichtung#Einsatzgebiete ) it is stated that it was first used in the astronomy especially for uv light, in Hubble is a BSI-CCD used.

The Nikon D850 is equipped with an BSI CMOS Sensor ( https://www.nikon.de/de_DE/news-press/press.tag/news/bv-pr-wwa1711-nikon-d85...-100-score.dcr ) this camera gets one of the highest DXO MARK results because of the picture quality and the capabilities of the ISO range ( https://www.dxomark.com/nikon-d850-sensor-review-first-dslr-hit-100-points/ )

[Digitalis]

I gather that for both astro and other long-term-exposure-in-the-dark shooters any BSI sensors are not a good idea.

There may be a nugget of truth there. Most Astro shooters use dedicated, cooled* CCD sensors with numerous filters. BSI CMOS and BSI CCD sensors exist, it would be interesting to compare the two for astrophotography.


* The application of active cooling reduces the effects of dark current

[biz-engineer]

BSI is more sensitive (native sensor ISO is a about a stop better than non-BSI), allowing shorter exposure times, would lower the impact of dark current. Real world comparisons need to be made.

[beholder3]

can you give a link to the source?

DSLR-Forum - Einzelnen Beitrag anzeigen - Pixeldichte und Bildsch

The author usually is very knowledgeable and trustworthy. Ask him yourself if you need more detail.

The statement also is generic and one would need specific sensors to compare in real life.


Here is some interesting article: https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=5&ved=2ahUKEwi...mc6liO9YVE1KPK

dxomark type noise performance is not the same as thermal noise performance caused by longer exposure times.

[Wasp]

I tried a bit of astro photography with my BSI equipped Samsung NX1 a few weeks ago when I was in a rural location. I also had a go with my trusty Pentax K-x. The results were not stellar(!) but I put that down to the light of the full moon messing up my exposure. I kept exposure times to 30 seconds and below, so I doubt that noise played a role.

[joergens.mi]

But he is talking about a little bit more sensitive to dark current. I think you put to much emphasis on the factor of 4, which is one the factors relevant for the quality of pictures.


He also says, that there isn't any problem below 40°C on the sensor, and tells us that an active cooling will improve the results for dark scenes and times above 1/10s.

I would interpret it, that in normal temperature situations there isn't any problem, above 40°C we will get normally enough light to be below the problem zone. I may be problematic when you are in pit or cave 1200 m below surface, because the environmental temperature will raise above 35°C an there will be a lot of light missing there.

[beholder3]

But he is talking about a little bit more sensitive to dark current. I think you put to much emphasis on the factor of 4, which is one the factors relevant for the quality of pictures.

The numbers he gives is factor 4.

He also says, that there isn't any problem below 40°C on the sensor, and tells us that an active cooling will improve the results for dark scenes and times above 1/10s.

Sure. The question is, what is "problematic" given the detail loving discussion we have on forums?
If it is worth mentioning that cooling will improve results from 1/10 s on then obviously there is impact enough.


There is a good reason for LENR / darkframe subtraction even in normal environments for long exposures, because even then you get visibly more noise in the images with non-BSI sensors. And all this is being caused by the sensor heating up a few degrees.

It is not dramatic, but it is there.

[biz-engineer]

The numbers he gives is factor 4.

Pentax will improve that when they will customize their BSI sensor for Pentaxians. Isn't Pentax KP already using a BSI sensor?

[clackers]

BSI is more sensitive (native sensor ISO is a about a stop better than non-BSI), allowing shorter exposure times, would lower the impact of dark current. Real world comparisons need to be made.

It's a trade-off isn't it? A bit more real estate for light gathering versus the problem of more thermal noise and cross talk from the wires being close to each other.

It was only attempted originally with small phone sensors and their low currents.

[biz-engineer]

It's a trade-off isn't it? A bit more real estate for light gathering versus the problem of more thermal noise and cross talk from the wires being close to each other.

According to this paper [ https://www.researchgate.net/profile/J-P_Carrere/publication/269292896_CMOS_...rk-current.pdf ] (go direct to the conclusion) , dark current greatly depends on the process steps of manufacturing (etching, etc..), not really whether the pixel is front illuminated or back illuminated. There is also some info about blooming pixels (the infamous white dots of the D810 and K1 sensor, due the presence of metallic contaminant in the silicon substrate).

[clackers]

According to this paper [ https://www.researchgate.net/profile/J-P_Carrere/publication/269292896_CMOS_...rk-current.pdf ] (go direct to the conclusion) , dark current greatly depends on the process steps of manufacturing (etching, etc..), not really whether the pixel is front illuminated or back illuminated.

Sure, BE, but thermal noise and cross talk aren't dark current phenomena.

[reh321]

It was only attempted originally with small phone sensors and their low currents.

I fixed it for you.
I believe the sensor used in the Q-7 and Q-S1 is BSI.

[clackers]

I fixed it for you.
I believe the sensor used in the Q-7 and Q-S1 is BSI.

Yeah, next up were point and shoots, compacts.

My Canon G7x has the 1" Sony BSI Exmoor sensor.

[biz-engineer]

Sure, BE, but thermal noise and cross talk aren't dark current phenomena.

they also mention dark current, if you read carefully.