[Ian Bell[_2_]]

Arny Krueger wrote:
"Ian Bell" wrote in message

Arny Krueger wrote:
"Ian Bell" wrote in message


0dB has no meaning.
Actually, it does. In the digital domain 0 dB is
commonly equated with FS.

No, that is 0dBFS.

Yes, so your use of the word "no" must be some kind of terrible mistake on
your part.


No, 0dBFS refers to full scale signal in a digital system. It is common,
especially in live recording to set '0dB' to -15dBFS so as to ensure
sufficient headroom. In other words, 0dB can be anything you like and
therefore on its own without context is meaningless.

While the units are arbitrary and may be confusing to
people who still think 100% analog, it is both
meaningful and commonly used.

No it is not.

Assertion without support, which should be immediately dismissed.


OK, then we will dismiss your original unsupported assertion that 'it is
both meaningful and commonly used'.

Input termination in both cases is unspecified.
If the input is a mic input, then the usual standard of
300 ohms applies.

It is a power amp, it does not have a mic input. 300 ohms
is NOT the 'usual standard for a mic input'

Good modern power amps are common, and do not have that much variation in
their noise performance with normal variations in source impedance.

If the input is a line level input, then the source
impedance is usually such that system performance is not
affected that much by probable variations.

You miss the point. How is the input terminated when
noise is measured - obviously it is not left open circuit
is it?

I guess you've never measured the actual noise coming out of a good modern
power amp with the normal range of source impedances. They often don't vary
all that much. Remember, this is not legacy vacuum tube equipment which was
generally far noisier.


A 'good modern one' should not vary much. We have no idea if the OPs's
device falls into this category. However, if the noise is being
minimised then a point will be reached when the source impedance is
relevant because if the amp was extremely noise free that would be the
ONLY source of noise.

OTOH, it will not vary much with source impedance if the amp is very
noisy to start with, like an old tube amp. SO precisdely the coinverse
of what you state is in fact true.

Output termination is unspecified.
The noise performance of line level outputs is minimally
affected by probable and reasonable variations in load
impedance.

No bandwidth is specified.
Relevant and already mentioned

No weighting or not is specified.
Relevant and already mentioned

No mention of whether the measurements are rms, peak,
quasi-peak
Both SNR and DR are ratios of two measurements. It is
most important that both measurements be done under the
same circumstances.

In fact they generally are not since a signal and noise
have quite different characteristics. The signal will be
measured rms. The noise can be measured in several ways
each giving a different figure.

While people could be stupid and compare a noise level measured in
peak-to-peak volts to a signal measured in average volts, I know of no
actual cases where this happens, except perhaps in your mind, Ian.

No, but as I keep saying, manufacturers will show their product in the
best light and judicious use of weighting and bandwidths will alter the
'measured' value considerably. 'A' weighting, which is pretty commonly
used by manufacturers will often improve and amplifiers noise spec. by 10dB.

There
could be equipment noise that has a high crest factor, but nature does not
usually go down that path. What is then left is a few dB of ambiguity, and
most good modern equipment (which is common) is not so noisy that a few dB
is a deal breaker in actual use.


It is more than 'a few dB'


Variations in the way the above are specified can make a
large difference to the 'measured' value

Some yes, some no.

and I bet you
can guess which set the marketing department will want
to choose.
One other benefit of the solid state/digital revolution
is that SNR and DR performance is often so good ( 100
dB) that most variations in how they are mentioned don't
matter that much.

Rubbish.

Dismissive, unsupported claim, again itself worthy only of dismissal.


As I have said before DR and SNR are not the same and they have little
relation to the actual level of noise heard in the speaker. The OP's amp
for instance has an output power of 800W. Suppose this is into a load of
4 ohms then this requires over 56V rms output signal. The DR is 100dB so
the noise at the output is 100dB below 56V rms which which works out at
a mere -65dBu.

The original all tube Leak Point One in 1949 had a measured output noise
and hum of -80dB below 10W into 15 ohms which works out at just over 12V
rms. So its output noise is 80dB below 12V rms which works out to be a
mere -58dBu.

So in 60 years of development there's been only 7dB (oh sorry I should
have said 'just a few dB') of improvement due to the digital revolution.

Once again, SNR and DR are different animals.

A truism - therefore something that actually sheds no significant light. Why
do you obsess over these things, Ian?

Achieving a 100dB DR is not hard but unless you run your
amp close to clipping all the time you will not achieve
that as a SNR.

Another truism.


Excellent, so then you must agree that SNR is always less than DR.


So to get back to the OP's original problem, having a DR of 100dB is
irrelevant. What matters is what his SNR is. The fact he can hear hiss
from an 800W amp with his ear right next to it further demonstrates this
because that is not where listeners will be when the spekaer pumps outn
800W.

The marketing guys are as active as ever in trying to
make their product appear superior to its competitors and
will choose the measurement methods that best do that.

The myth is that some equipment is head-and-shoulders better than its
competition. Everybody pretty much feeds from the same trough.



In reality, marketing departments will aim to make crappy equipment
appear much better than it is using specmanship which is what they have
always done.

Cheers

ian