Re: Read noise: get it 30x down?
- From: davem@xxxxxxxxx (Dave Martindale)
- Date: Thu, 13 Dec 2007 09:04:07 +0000 (UTC)
floyd@xxxxxxxxxx (Floyd L. Davidson) writes:
This does not make any sense at all. CCD does not produce (any
measurable) power; the process of "reading the capacitor of CCD" is,
in fact, lossless. CCD produces voltage. ADC does not measure power.
ADC measures voltage.
So the input impedance of the ADC is infinite...
Not quite. The input impedance of the sense amplifier, which is
probably on the CCD chip, is infinite. The input impedance of the ADC
is probably finite and resistive.
The problem is that 2X the amount of light delivers 2X the electrons
into the sense capacitor, which results in 2X the voltage out of the
sense amp, which produces 2X the voltage *and current* into the ADC.
So a 3dB change in charge (and power) into the sense amp results in a 6
dB change in power into the ADC. Which of these is the right place for
calculating signal-to-noise? Because you'll get different answers in
the two places.
The 6 dB figure is only applicable if the current
increases proportionally to the voltage. If not, it's
3 dB...
People are using the "20*log(v2/v1) to calculate signal to noise,
and some of us are suggesting that's the wrong formula. Perhaps it
should be 10*log(v2/v1) since that's the power ratio *in the light from
the scene*.
Dave
.
- References:
- Re: Read noise: get it 30x down?
- From: Floyd L. Davidson
- Re: Read noise: get it 30x down?
- From: Ilya Zakharevich
- Re: Read noise: get it 30x down?
- From: Floyd L. Davidson
- Re: Read noise: get it 30x down?
- Prev by Date: Re: Question: Comparison of Compacts
- Next by Date: Re: Read noise: get it 30x down?
- Previous by thread: Re: Read noise: get it 30x down?
- Next by thread: Re: Read noise: get it 30x down?
- Index(es):
Relevant Pages
|