[JohnR]

Well, the specs say:

Output noise ..................................... Noise floor typical -135dBV, 20-22kHz
SNR ...................... > 93dB (20-22kHz) both ch driven @ 1kHz / 4Vrms (No weighting filter)

I gather the latter is the noise summed over the complete spectrum, plus THD components from the 1kHz input. This seems like a disadvantageous way of specifying it.

For power amps, your argument leads to the conclusion that 1 Watt is just as arbitrary and meaningless a reference as full power, given the widely difference sensitivity in speakers.

Still, again I see your point. My counter point would that, somehow, reducing it to these figures doesn't quite make sense. After all, since lots of speakers will easily generate 1% THD, then it doesn't matter if our amps do either. I dunno, something's not addingup here

J

[Dan Banquer]

O.K. let's start from here. the manufacturer of your pre amp gives an SNR of 93db referred to 4 vrms. So  referred to 2 vrms, ( 6db down) you will have an SNR of 87db. Let's go another 6 db down to 1 vrms and you will have an 81 db SNR down from one volt. Lets do a quick approximation here. 80 db is equal to one part in 10,000. 1 vrms divided by 10,000. I don't have my calculator but that sounds like 100 microvolts of noise, as a rough approximation. The real number is probably closer to 97 microvolts of noise.
THD at 1%? At what power level, and at what part of the frequency band for the loudspeaker.
You're right John. Things don't add up. What's missing in your opinion?

[8thnerve]

Sure, but first, did my tax return example make sense?

I'm sorry John, I did not realize that was not rhetorical.

Digital  adj.  A description of data which is stored or transmitted as a sequence of discrete symbols from a finite set, most commonly this means binary data represented using electronic or electromagnetic signals

Binary  adj.  Any file format for digital data encoded as a sequence of bits but not consisting of a sequence of printable characters (text). The term is often used for executable machine code

Both digital and binary are unable to store information past the resolution of their significant digits.  So how in this case are they different in a manner that makes binary incorrect to a degree that you felt it was necessary to point this out?

At any rate, the possibility of new advanced analog electronics is fascinating.  In my experience, it is the ideas that do not simply accept conventional wisdom that prove to be the most revolutionary.

[JohnR]

Both digital and binary are unable to store information past the resolution of their significant digits. So how in this case are they different in a manner that makes binary incorrect to a degree that you felt it was necessary  point this out?

Right. Earlier you stated that that "more levels" would somehow make it possible for digital computers to represent analog better. My point is that it makes no difference whatseover: binary, ternary, decimal, whatever, there is no fundamental difference.

In general, I find it odd that people "explain" digital by saying that it has zeroes and ones, as though that was, in and of itself, somehow significant. It's much more useful to think of digital processing as NUMBERS, not as bits. Numbers make sense to people, and if you're working with digital signals, that's what you use, not bits. Only hardware engineers and assembler programmers care about bits. (Not quite true, but close.)

Hope this helps...

JohnR

PS Eduardo, I followed a couple of links and found some intro material. Looks pretty interesting! But a lot more than 20 or 30 years away from practical application (if ever). What do you think?

[JohnR]

You're right John. Things don't add up. What's missing in your opinion?

Presumably the dynamic spectral content of the noise and distortion. Let's try some numbers...

As you know , 0.1% second-order distortion is inaudible. That is 60 dB down, so we have:

• Hypothesis 1. Second-order harmonic distortion at -60dB is inaudible.

What about white noise? 60dB is I believe beyond what analog tape is capable of delivering, not to mention a lot of tube preamps. And you can "get used" to that... so how about this one?

• Hypothesis 2. Constant (unmodulated) white noise at -60dB is audible but not objectionable.

I heard a very interesting talk at the AKSAFest by Ben Williams, who is working with Aspen on a DAC design. He played samples showing the audible effect of four different types of jitter. Quite fascinating. He also mentioned that studies have shown that listeners can detect jitter at levels as low as 1 ppm. Noise due to jitter is broadband and not harmonically related. I don't know what 1ppm jitter translates to in the signal, but one-millionth is -120dB so let's try this one:

• Hypothesis 3. Wide-band non-harmonically related distortion is audible and objectionable somewhere between -100 and -120 dB.

What say ye to hypotheses 1 to 3, Dan?

JohnR

[JohnR]

THD at 1%? At what power level, and at what part of the frequency band for the loudspeaker.

Dan, I started a thread with some graphs of driver distortion. Hope you find it of interest.

http://www.audiocircle.com/index.php?topic=2886.msg23499#23499

[Dan Banquer]

I think the statement on white noise could well be true. It's audible but not objectionable to most people. However; if you are trying to increase dynamic range with a limited voltage swing then our -60 db of white noise becomes a real limiting factor.
On a more personal note; one of the things that really piss me off about most audio electronics is the noise level, because it masks the harmonics of acoustic instruments. Or let me put it this way; what the hell good is listening to Itzhak Pearlman play a Stradivarius if the harmonics of that instrument are buried in noise?  If you have been lucky enough to stand next to something like that, or be in the middle of something like that, then you instantly realize just how much information get's missed due to electronic noise. I am truly cursed if you think about it for a minute.
The stuff on dither I've know for a while, but thanks for the refresher. I will get back to you on speaker data when I get a chance.
I would also not be surprised if number three is true. The other thing we should also note is that the tuning of a piano is "bent" a little bit at the higher end of the spectrum if I remember correctly, which could well mean that higher order electronic distortions which are exact become even more pronounced at the treble frequencies.

[EDS_]

8thnerve hit a nerve for me. I, moons ago, missed an exam question about the ultimate resolution of analog v. digital.

Analog wins.

[JoshK]

May I humble attempt to further explain JohnR's correct point about digital versus binary?

Forget base 2 for a second and lets look at base 10 (something we are all familiar with).  If I want to represent the number 1/3 using decimals I need an infinite number of decimal points... i.e.

1/3 = 0.3333333.... etc but lets say I only have two decimal places of space on something (like a tax form) to fill in.  So I can only write 0.33 the rest of the 3's get truncated.  This is the key difference, the truncation.

I can represent any number using base 2 that I can with base 10, the base is irrelevant like John says.  Its the truncation that makes digital imprecise because we don't have enough bits to store the number 1/3 with infinitesimal precision.

Don't know, maybe John's tax example was clearer.

[nathanm]

If you have been lucky enough to stand next to something like that, or be in the middle of something like that, then you instantly realize just how much information gets missed due to electronic noise. I am truly cursed if you think about it for a minute.

Well, there's also the issue of how much detail is desirable.  For instance, there's a lot of noises that instruments make that are, dare I say, unmusical.  Spit crackles on horns, clacking of sax keys, string squeaks on guitars, pick noise, flutists breathing, saliva noise on a close-miked singer, buzzes in a drumkit - all of this stuff is part of the performance, but it's kind of a side effect of the tone itself.  It's possible that this extra detail, while it does make the sound seem more real, is intentionally avoided in the recording technique.

But I agree that low noise components and a low noise environment goes a long way towards making the reproduced sound more lifelike. Steady state hiss seems to remind you 'hey, it's just a recording'.

[Dan Banquer]

If you don't like that kind of detail then tell the recording engineer not slam the microphone down the bell of the horn or down the throat of the singer. Recording studios love to do close miking, especially for small groups in the studio. Then they add a little reverb to "give that live room effect" Huuuuuuuuuuuuuuummmmmmmmmmmmmm.

[nathanm]

Whoah, I think Dan just went into deep meditation at the end of that post!

Maybe my reply was a bit cryptic.   I was just suggesting that the recording techniques themselves may have more to do with the obscuring of detail than the noise level of the playback system.

On most pop acoustic guitar tracks I would suggest the opposite, that harmonic detail is about the ONLY thing they let you hear. You get those strummed chords which are boosted in the treble and cut in the bass, then compressed so all you end up with is pure scratchy string noise.  No tone, no nothin'!  No amp in the world can fix that! Heh!

I know I know, "what does that have to do with analo...etc."  I'm sorry in advance.

[Dan Banquer]

Hi John R. I took a look at the graph you posted on the woofer. I confess I am having a hard time reading because it is so damn fuzzy. In any case the question I am asking has to do with low level linearity. I think people misunderstand that linearity is only distortion. This is not the case. For this application it is not only distortion but the relationship of power in to power out over frequency and and at different power levels. If the device is linear the reponse will be uniform over the designed for range. Harmonic distortion at 6.3 watts over frequency tells very little. Do you have any more data on this woofer that might help us?