Here's an explanation pitched at a BSEE level for why input-referred noise could drop as gain is added that doesn't involve shot noise.
Consider a noise source, N1, followed by a noiseless amplifier of gain A, followed by another noise source N2. If N1 = 2 and N2 = 1, with A=1 the noise at the end of the pipeline is sqrt(1^2+2^2) = 2.236. Since the gain = 1, that's also the noise referred to the input of the pipeline. Now let's double the gain. The noise at the end of the pipeline is sqrt(1^2+4^2) = 4.123. Referred to the input, that's 2.061. Adding gain has reduced the input-referred noise.
I am sure it was, in connection with valves, which were never part of my professional needs -- it was1964-8 and transistors were the thing already -- thus I used that part of my brain for things that I dealt with regularly! Remember, I am an Engineer, not a Physicist and the thinking is different.
I will follow up the reference, which doesnt get quoted here for some reason...
Yes. I understand the problem. But no intelligent person who genuinely wants to learn minds reviewing basics from time to time. And it should be possible for him/her to skip things that are well-known.
So as time goes by, assuming a static subject, is this effectively increasing the number of bits used to represent the image?
This is all well and good, but for those of us without the desire to relive their days of A-Level physics, are we actually arguing about historically wrong core principles here, or rather terminology? By that, I mean does the fact not remain that if you want a "lighter" photo without opening the aperture wider, or shutter longer, you use a higher ISO? A million books have told me so, and I've seen it in action countless times, so frankly that's all I really care about re. the embattled triangle, irrespective of whether or not one of its components is technically not part of a given definition of exposure.
WHY when I highlighted the whole paragraph do I only get the first word?
Thanks! I need to think about the example, the result of which looks as though it may be a special case.
As far as Terman is concerned, I dont think I still have the book (it was very dry and out of date in many ways in 1960). But he himself says that the aim is to run the vacuum tube in such a way that shot noise is insignificant. The mechanism of transistors, which run cooler, is not the same, of course.
Shot noise is as important with transistors as it is with valves. I would have thought that engineers would need to understand the origines of the phenomena in their equations. But there we are
I find that in these photography discussions a remarkable number of people are very resistant to reviewing basics. I'll refrain from commenting on intelligence.
Depends on the way you're using the term 'bits'. If you mean bits of data, that gets built in by the electronic engineers. If you mean bits of information, exactly so.
If that level of understanding of what ISO and exposure do, then that's fine. The real problem is that conceptualising it in the triangular way is that it tends to stop people going any further than that. Or even from realising that there is a further than that.
OK. So here is what I do: I use the Canon Fv mode, set aperture to give the depth of field I want and what I consider an appropriate shutter speed, then ISO sets itself according to Canon's algorithm. Often I just set aperture and let Canon set ISO and shutter speed.
Is this not a good way to take photos, the raw files of which I then process in PhotoLab?
LOL I spent most of my engineering career avoiding noise. But once I was asked to provide a noise source for a Synket synthesizer and used a zener diode like you. The composer complained that the noise was too white (only that was not how he described it), so I added a capacitor across it!
Using a minimum SS setting and exposure compensation to fine-tune the exposure (or) brightness when necessary, this is a perfectly valid way to shoot, and how I also shoot most of the time.