How are counts per frame calculated in CCD for spectrometry?

[Drakkith]

Hmmm. I thought readout noise was centered at zero, but I could be mistaken.

 

[hutchphd]

I think it depends upon the CCD but I have never worried about it much. Certainly the background can be subtracted off as required.

 

[new6ton]

When the signal is comparable or close to the read out noise that more exposure can spell a difference? Remember I wrote this thread about Counts per Frame in CCD in visible as well as IR Spectroscopy. So please share your experience in IR spectroscopy now. With these feature soon available in ordinary smartphone, and we can scan for say date rape drugs in drinks or how sweet is the apple at grocery with our smartphone. It would be handy knowledge.

 

[Drakkith]

When the signal is comparable or close to the read out noise that more exposure can spell a difference?

More exposures always improves SNR. It's just that when your signal is very strong the SNR is already very, very high and you just don't need to improve it.

Remember I wrote this thread about Counts per Frame in CCD in visible as well as IR Spectroscopy. So please share your experience in IR spectroscopy now.

There's no difference between the two in terms of noise and how sensors work. Everything that's been discussed here applies to both IR and visible.

 

[new6ton]

Is readout noise dependent on wavelength?

Anyway. I just have one more question and don't want to post in a new thread to save pages.

Can you give example of an actual photo where the sensitivity response of the CCD is flat (uniform from red to violet) instead of curve? And how an image would look like if your eyes have flat response too able to see visible light with equal sensitivity?

 

[sophiecentaur]

Red and blue wavelengths human sensitivity is lower than greens. So a flat response would boost reds and blues, giving a strong magenta tint to all images.
But that’s an unrealistic answer to an unrealistic question, I think.

 

[Drakkith]

Is readout noise dependent on wavelength?

No, because readout noise has nothing to do with wavelengths. It has to do with the sensor's ability to accurately count electrons shuffled in from each photosite (pixel).

Can you give example of an actual photo where the sensitivity response of the CCD is flat (uniform from red to violet) instead of curve?

No such CCD's exist. All have some sort of curve in their response sensitivity.

 

[Drakkith]

And how an image would look like if your eyes have flat response too able to see visible light with equal sensitivity?

Well, we don't, so we can't say what it would look like if we did because our brains would also interpret this differently. To throw out a random guess, I'd say that everything would look very much the same as it does now.

 

[hutchphd]

The quantum efficiency of all both CCD and CMOS detectors peaks slightly above 900nm. Any other wavelength will be intrinsically noisier. Whether that noise is significant depends upon a host of factors including details of fabrication and signal strength. etc etc.etc