Re: Read noise: get it 30x down?

davem@xxxxxxxxx (Dave Martindale) wrote:
- a doubling of voltage or current is a 6dB change (when measured at a
constant resistance or impedance).

Yes. The problem is that people observe the doubling of the voltage and
use the usual formula to calculate the change as 6 dB (4X the power).
That's true of the electrical signal, but *it is not true of the
original light signal* where power has only doubled, and it isn't true
of the energy transferred from CCD to sense amp.

But of course none of that has any significance at all.
For the exact reason that you stated as your last
sentence:
a
"As long as you remember what it is measuring."

What is being measured is *not* the radiant power
compared to the electrical power.

The dB values for ratios of radiant power are valid.
That is the common measure. For example, fiber optic
systems typically have a design parameter for "optical
power margin". That is very similar to the "fade
margin" used in the design of microwave radio systems.
In both cases it is part of the "link budget" drawn up
by the systems design engineer. In both instances that
margin is a ratio, and it is specified in dB. Nothing
else would make particular sense. It allows comparisons
to be made for the source, the receiver, and the
transmission medium without doing some form of
normalizing to get comparable results.

It is *not* used to compare, either with an optical
system or with a radio system, the signal input to the
detector output. Indeed, the SNR at each location is
granted a very interesting thing to compare, but it is
not a linear relationship and is not something that
produces a ratio itself that can be expressed in dB.

Keep a reality check going on his one. The *fact* is
that dB is a very very common way to describe ratios in
electronic cameras. If _you_ find something that
is inconsistent, _you_ can rest assured that what is
inconsistent is your understanding of the topic, not the
way he rest of the world is describing it.

No,

Yes, that is a *fact*, and what you have said here supports
that.

you just have to realize that in electronic imaging, electrical

See, it is "yes". You have to realize... what the Hell
is actually being said! Applying poppy*** personal
notions of what things "should be", is not a way to
determine if something is valid or not.

signal dB does not correspond to optical signal dB, because the sensor
is a power sensor, not an amplitude sensor. A 3 dB change in light power
produces a 6 dB change in sense amplifier output power into a resistive
load.

True or not, that doesn't have any valid significance!
We don't compare across the sensor. We compare radiant
power to radiant power; we compare electrical power to
electrical power.

Describing ratios in dB is exceedingly useful because it
is exceedingly difficult to look at raw numbers and
compare them as a ratio. But 6 dB is the same ratio no
matter where you find it on a scale.

As long as you remember what it is measuring.

Exactly. It is *absolutely* a ratio between two things
that *must* be the same.

--
Floyd L. Davidson <http://www.apaflo.com/floyd_davidson>
Ukpeagvik (Barrow, Alaska) floyd@xxxxxxxxxx
.