Edge glow on master bias

[sharkmelley]

I thought I would highlight a interesting phenomenon that seems to be becoming more common: edge glow on bias frames of CMOS sensors.

Here's an example taken with my Sony A7S.  A stretched average of 250 dark exposures of 1/8000sec at ISO 10000 at room temperature (20C) in complete darkness:

 

"Amp glow and/or light leaks", I can hear you all say.  Actually no, because dark current cannot possibly accumulate in 1/8000 sec and the exposures were taken in a dark cupboard.  To confirm this glow is not thermal, here is an average of 50 frames at -20C (these were shot in a dark freezer after 8 hours of cooling down time):

The stretching is exactly the same in both images - in a single exposure bias frame the bright patch is almost one electron "brighter" than the rest of the frame.

The other interesting point is that a 30second dark frame has no more edge glow than the bias - if I subtract the bias from the (much longer exposure) dark then the edge glow is removed.

It's not just the Sony A7S that is affected. Jim Kasson has seen it on the Nikon D800 and D810:

http://blog.kasson.com/?p=7665

http://blog.kasson.com/?p=7789

I have also seen examples from the Nikon D600.

Is it just a DSLR problem or are cooled mono CMOS sensors similarly affected?

I should emphasize that for deep sky astrophotography it causes no problem at all because it calibrates out. But it is an interesting "feature" and what exactly is causing it?

Mark

 

 

 

Edited December 20, 2016 by sharkmelley

[Xplode]

This seems to me like a trying to make a problem where there isn't one, if it calibrates out it's certainly not a problem so why waste time on it?

You should take a look at calibration frames from CCD cameras, they look horrible compared to this and they still calibrate perfectly

[sharkmelley]

15 hours ago, Xplode said:

This seems to me like a trying to make a problem where there isn't one, if it calibrates out it's certainly not a problem so why waste time on it?

 

Yes it calibrates out, so there's no problem but I'm still intrigued to know what causes it.  Most people who have noticed it have (wrongly) assumed it is thermal glow.  But what is it actually?

Mark

[hughgilhespie]

Hi Mark,

Probably teaching Granny, but perhaps posting on Cloudy Nights might be more likely to get a response? They seem to have some very knowledgeable people posting. As you said, it will be interesting to track down what is happening.

Regards, Hugh

 

[Oddsocks]

There are two possibilities that I can think of:

1: Capacitively trapped electrons in the rows and columns closest to the left side and bottom edge of the array where there are, depending on the particular deign of detector, a number of "dark" rows and columns used for on-chip dark calibration, these are covered by a vapour deposition metal plate a few atoms thick and the plate could, in theory, form one pole of a capacitor.

2: Beyond the dark rows and columns are the row and column amplifiers that are a source of IR. The amount of IR emitted is a constant since the read time of the row and column amplifiers (and therefore their "on" time) never varies irrespective of individual exposure time and would be a source of low level read noise added to every exposure. The IR, depending on source intensity, can only penetrate so far into the wafer.

Without knowing the actual structure of the detector, which would be a design secret, it is not possible to say which of the above is the most likely, the irregular shape of the "glow" hints at a capacitive trapped electron theory but depending on the layout of the row and column amplifiers the leaked IR from the row and column amplifiers is just as likely.

One test you could do is to flood the detector with light and read the array with a single exposure and then immediately cover the detector and take a dark bias, if the edge glow changes after the flood exposure then that would hint at the trapped electrons explanation since flooding the detector will dislodge some of them, if the edge glow is exactly the same after the flood exposure then that would hint at IR from the row and column amplifiers as being the source.

You might find the linked document of interest:

https://courses.cs.washington.edu/courses/cse467/08au/pdfs/lectures/07-cmos-ccd-imagers.pdf

Edited December 22, 2016 by Oddsocks
Tidy up

[sharkmelley]

10 hours ago, Oddsocks said:

There are two possibilities that I can think of:

1: Capacitively trapped electrons in the rows and columns closest to the left side and bottom edge of the array where there are, depending on the particular deign of detector, a number of "dark" rows and columns used for on-chip dark calibration, these are covered by a vapour deposition metal plate a few atoms thick and the plate could, in theory, form one pole of a capacitor.

2: Beyond the dark rows and columns are the row and column amplifiers that are a source of IR. The amount of IR emitted is a constant since the read time of the row and column amplifiers (and therefore their "on" time) never varies irrespective of individual exposure time and would be a source of low level read noise added to every exposure. The IR, depending on source intensity, can only penetrate so far into the wafer.

 

Very interesting - thanks.  Eric Fossum also suggested the possibility of NIR glow from adjacent circuitry:  https://www.dpreview.com/forums/post/58819811 

I very much like your explanation of why the amount of IR emitted would be constant i.e. not increasing with exposure time.

Mark

Edited December 22, 2016 by sharkmelley

[David Magro]

Hey Mark,

Did you solve the mystery behind the glow on your A7s?

I also have the issue with the A7sii when shooting at 1/40" ISO409600.

David Magro

[tooth_dr]

On 22/12/2016 at 00:41, sharkmelley said:

Yes it calibrates out, so there's no problem but I'm still intrigued to know what causes it.  Most people who have noticed it have (wrongly) assumed it is thermal glow.  But what is it actually?

Mark

Mark

I thought I would post looking for advice regarding using the Nikon with the Epsilon 180ED.  I have noticed that glow along the bottom of my bias frames, but it doesnt seem to calibrate out and I'm left with a band along the bottom of stacked image.  I took a series of 300s lights at ISO400, and dithered my captures.  I then took bias frames and flats at ISO400.  I've tried using the bias as darks, and created a bad pixel map, but all combinations of calibration do not seem to be removing the glow.  Would you mind if I asked how you dealt with this properly?

[tooth_dr]

[sharkmelley]

1 hour ago, tooth_dr said:

Mark

I thought I would post looking for advice regarding using the Nikon with the Epsilon 180ED.  I have noticed that glow along the bottom of my bias frames, but it doesnt seem to calibrate out and I'm left with a band along the bottom of stacked image.  I took a series of 300s lights at ISO400, and dithered my captures.  I then took bias frames and flats at ISO400.  I've tried using the bias as darks, and created a bad pixel map, but all combinations of calibration do not seem to be removing the glow.  Would you mind if I asked how you dealt with this properly?

You're having problems with a Z6 ?  I wonder if the band you are seeing is something to do with the PDAF rows.

Mark

[tooth_dr]

2 minutes ago, sharkmelley said:

You're having problems with a Z6 ?  I wonder if the band you are seeing is something to do with the PDAF rows.

Mark

Hi Mark, sorry if I have misled you, I have a Nikon D800E.

[sharkmelley]

I don't know what kind of edge glow (technically electroluminescence) the Nikon D800E has.  But it ought to calibrate out.  However using a DSLR (i.e. not mirrorless) with a fast scope like the Tak Epsilon will always leave a band along the bottom after calibration because of the parked mirror in the mirror box.  I always had problems with my DSLR until I physically removed the mirror.

Mark

[tooth_dr]

12 minutes ago, sharkmelley said:

I don't know what kind of edge glow (technically electroluminescence) the Nikon D800E has.  But it ought to calibrate out.  However using a DSLR (i.e. not mirrorless) with a fast scope like the Tak Epsilon will always leave a band along the bottom after calibration because of the parked mirror in the mirror box.  I always had problems with my DSLR until I physically removed the mirror.

Mark

Thanks Mark. Perhaps I’m seeing that at the bottom?  The mirror actually impinges permanently on the image?

[sharkmelley]

The mirror impinges on the image creating a bigger area of shadow and an extra vertical diffraction spike on stars.  But there is also diffuse scattered light off the back of the parked mirror that prevents the flats ever properly calibrating the lights.  You'll soon become an expert at gradient removal !

Mark

[tooth_dr]

22 minutes ago, sharkmelley said:

extra vertical diffraction spike on stars.

Ahhhh! That explains that, I was doing some head scratching earlier !

Edited September 21, 2020 by tooth_dr