"However, you missed something in my post. The lower limit of audibility of a 25hz tone is just over 80 db, so a 100db 25hz tone will not be that much louder perceptibly than the distortion product I referred to."
Does that limit the isuue to "Dark Side of the Moon" and pipe organ concerts? Judging by the bars on my RTA and what I hear at the same time, 100db spl @ 100Hz is plenty audible and the subway vs. speech studio comparrison should be reasonably valid.
Does that limit the isuue to "Dark Side of the Moon" and pipe organ concerts? Judging by the bars on my RTA and what I hear at the same time, 100db spl @ 100Hz is plenty audible and the subway vs. speech studio comparrison should be reasonably valid.
"You are misunderstanding the implications of having a 15th
harmonic 80db down. By definition this cannot be the only
harmonic distortion, it implies a series at least to the 15th,
and probably much further - 81st harmonics have been
identified in Crowhursts work."
Not at all. The original comment spoke of THD % with a couple of leading zeroes. If you have a 15th harmonic of -80db, the probalility that sum of all harmonics (THD) will have fewer than 2 leading zeroes approaches unity. Conversly, if you start from the situation of a two-leading-zero-THD, the proability that the 15th harmonic will be as great as -80db is rather small.
I admit that my experience does not cover as wide a range of amplifier topologies aws does the experience of other posters, but to-date nearly all the Lin topology amps I've taken a look at are dominated by 2nd and 3rd harmonics with the leveld of subseqent higher orders diiminishing monotonicly.
harmonic 80db down. By definition this cannot be the only
harmonic distortion, it implies a series at least to the 15th,
and probably much further - 81st harmonics have been
identified in Crowhursts work."
Not at all. The original comment spoke of THD % with a couple of leading zeroes. If you have a 15th harmonic of -80db, the probalility that sum of all harmonics (THD) will have fewer than 2 leading zeroes approaches unity. Conversly, if you start from the situation of a two-leading-zero-THD, the proability that the 15th harmonic will be as great as -80db is rather small.
I admit that my experience does not cover as wide a range of amplifier topologies aws does the experience of other posters, but to-date nearly all the Lin topology amps I've taken a look at are dominated by 2nd and 3rd harmonics with the leveld of subseqent higher orders diiminishing monotonicly.
Only if one's hearing is extremely poor. Just try not to forget that such a harmonic would normally only be one of dozens of others along with intermodulation products produced by the same nonlinearities so the total situation is potentially far more dire than I suggested.
And since they add as the sum of powers, the summation of a hypothetical fifty such distortion products of equal amplitude would amount to only -63db total relative to the fundamental.
I don't believe in MHz bandwidths either but I also think an amp going to 20 kHz only is rather sluggish by today's standards.
DVD-A and SACD will definitely go higher.
And don't forget that you shouldn't introduce more errors within you chain than those that are already included in your source.
Regards
Charles
Please forgive me if I am totally wrong...
Let's simplify the situation by assuming that we are feeding the amplifier with a periodical pulse waveform. Its Fourier transform has an infinite number of odd coefficients. Therefore, an amplifier capable of 20 kHz bandwidth would have very poor pulse waveform response at 10 kHz.
I know the pulse response is primarily for market droids but somehow I tend to think waves as superpositions of a number of sine waves, each of them has to be reproduced perfectly in order to the original waveform be reproduced accurately. Thus, my way of thinking would apply to any other signal. In other words, an amplifier with high bandwidth (much higher than the range of human hearing) is desirable. High bandwidth may introduce other issues, for example, instability problems.
I am not a scientist and if you think I am totally wrong, I would be pleased to hear an explanation. Actually, I would really love the fact that amplifier would need only to cover the range of human hearing
Let's simplify the situation by assuming that we are feeding the amplifier with a periodical pulse waveform. Its Fourier transform has an infinite number of odd coefficients. Therefore, an amplifier capable of 20 kHz bandwidth would have very poor pulse waveform response at 10 kHz.
I know the pulse response is primarily for market droids but somehow I tend to think waves as superpositions of a number of sine waves, each of them has to be reproduced perfectly in order to the original waveform be reproduced accurately. Thus, my way of thinking would apply to any other signal. In other words, an amplifier with high bandwidth (much higher than the range of human hearing) is desirable. High bandwidth may introduce other issues, for example, instability problems.
I am not a scientist and if you think I am totally wrong, I would be pleased to hear an explanation. Actually, I would really love the fact that amplifier would need only to cover the range of human hearing
Hi Charles,
Depends on what gain you are talking about. When talking about SS amps there is usually a substantial amount of feedback to keep distortion within reasonable limits. For this you need loop gain. Now if you want 40 dB loop gain at 20 kHz and the amp should have a gain of 20 dB, you will need 60 dB open-loop gain at 20 kHz. This translates in a GBP of 20 MHz and our amp will have an overall bandwidth of 2 MHz..
So it depends on how much loop gain you need/want at what frequency.
Depends on what gain you are talking about. When talking about SS amps there is usually a substantial amount of feedback to keep distortion within reasonable limits. For this you need loop gain. Now if you want 40 dB loop gain at 20 kHz and the amp should have a gain of 20 dB, you will need 60 dB open-loop gain at 20 kHz. This translates in a GBP of 20 MHz and our amp will have an overall bandwidth of 2 MHz..
So it depends on how much loop gain you need/want at what frequency.
For a first order rolloff, this phase characteristic results in a declining group delay with increasing frequency until the group delay is halved at the -3db point compared to frequencies near DC. If a network was added to linearize this group delay variation and perhaps partially correct the frequency response droop below cutoff, how much more would this amp's sonic characteristic resemble a much wider bandwidth amplifier's, in peoples' opinions?
Hannu, yes that is right. But so would your ears. You can only detect frequencies up to 20kHz so if you listened to the pulse through a perfect 20kHz system and then through a perfect 200kHz system you wouldn't be able to tell the difference (provided your dog wasn't in the room that is).
According to Mr. Fourrier a music signal can be modelled as the sum of a collection of sinewaves, none of who's frequencies need be higher than 20kHz. This is how CD works - they do their best to remove anything over 20kHz before recording and after playback. And CD's sound damned good. So if you have a CD-based system your amp should not be receiving much at all above 20kHz anyhow.
In my view, amplifying spurious noise above 20kHz can only be a bad thing - at best a waste of power.
traderbam said:
Well, first there are individual variations. Most of us can only hear
up to some 20kHz, and the upper limit decreases quite heavily
with age for people in the Western civilization, probably due to
noise and stress etc. However, there are individuals who can hear
well above 20kHz. I think certain illnesses can cause such extended
frequency range, but there are also people having it without any
such reason. I have read of a study that claimed that broadcasting
technicians very often had a range extending above 20kHz, the
most extreme of these cases was 27kHz. I also know first-hand
of one person who claims (I couldn't verify it, of course) that he
has repeatedly heard over 30kHz in hearing tests.
OK, that is just a minor point about what frequency is a reasonable
limit. Replace 20kHz with 40kHz and we should still be safe
regarding these individuals.
The following contains a lot of speculationsHowever, how does human hearing
actually work? Do we just detect frequencies? I am no expert on
the anatomy of hearing, but we have the tapered cochilea with
sensory cells distributed along its inside walls. Presumably the
tapering of the cochilea has some function as a filter, by its
varying dimensions somehow relating to the different wavelenghts
detected. AFAIK the sensory cells are sensitive to particular
frequencies so the further into the cochilea, the higher the
"tuning" frequency of the sensory cells. This should mean first that
either some people have a cochilea that is longer and thinner in
the end, or we all have the physical capability to hear above 20kHz,
but for one reason or another cannot do that. Secondly, it means
that contrary to how measurement equipment works, we do have
some kind of distributed real-time Fourier analysis. I don't know
how this works, or if anybody does, but let's speculate a little
bit to get outside the standard thinking. What if the sensory cells
are not specialized in recognizing a particular frequency but rather
a rise/fall time rate. If it were so, it could for instance be possible
that we could distinguish between a sine and a square at say
20kHz, but some filter, probably in the brain, prevents us from
hearing fundamentals above 20kHz. I am not saying it works like
this, but just want to point out that maybe we oversimplify how
the human hearing works by making erroneous analogies to
measurement equipment? Maybe I am missing something here,
but I don't think this hypothetical hearing model would violate
Fourier results. We must not forget that the physical ear is just
one part of the hearing. We also have an astonishing signal
processor between our two ears that can play all kinds of tricks
with the information it receives from the ears.
I just realized I could probably cut outsome of the alternative hypothesis
and get to the same point anyway, but I'll keep it. Actually, even
if the sensory cells just detect frequencies, but we have such
cells capable of hearing well above 20kHz (which obviously some
people have) it could still be that the only difference between
these individuals and us others is that the brain imposes a
limit on what frequencies it bothers to accept as fundamentals.
Maybe we all hear the 3rd harmonic of 20kHz but if appearing
alone as a 60kHz tone the brain filters it out as irrelevant, but
does not do so in the presence of a 20kHz tone which could thus
be a fundamental. Different peoples brains could have different
limits for what fundamentals are considered relevant.
I did a little research, and it looks to me like a first order rolloff at, say 30khz may have its phase and amplitude effects largely compensated for up to 15khz by adding a second order LP filter with a Q=1 with a cutoff frequency of about 35khz.
IMO, if valid musical information is present in a recording above 20khz, then reproducing it will add to the realism of the sonic event. This is even though the sine waves will not be audible because they can still excite the middle ear's nonlinearities and are expected by the ear-brain mechanism to most fully emulate a lifelike reproduction of music that includes cymbals, brass etal.
IMO, if valid musical information is present in a recording above 20khz, then reproducing it will add to the realism of the sonic event. This is even though the sine waves will not be audible because they can still excite the middle ear's nonlinearities and are expected by the ear-brain mechanism to most fully emulate a lifelike reproduction of music that includes cymbals, brass etal.
traderbam said:
I do not know if this makes much sense but this statement made me wonder:
say you have an istrument that is actually producing something with a pulse that is nice and square at 10khz . when you listen to it your ears will do whatever ears do to limit the response. so your ears "squashed" the pulse a bit. and thats what it sounds like to you.
now listen to the same thing through an amplifier that also squashes the pulse due to a 20khz rolloff. the 10khz pulse has been squashed once by the amplifier, then is squashed again by your ears. (and CD player but lets not get into that
)I might be interesting to take one of these pulse waves and listen to it after it has been trough 1, 2 and more stages of 20khz rolling off. then you might be able to hear a difference.
I'm happy to extend the upper audible frequency for special humans who have K9 genes or complex fillings.
Nonetheless, a CD's output is heavily truncated just above 21kHz or so. It has to be due to the Nyquist limit of the digitizing process. So if you accept CD as a realistic rendition of the original music and if you use CD as your source, how can it be beneficial for your amp to reproduce frequencies much above 20kHz?
Now, I agree that the implementation of some amps necessitates higher bandwidth (such as those that use global feedback), but I argue that this is not a requirement of music reproduction.
Nonetheless, a CD's output is heavily truncated just above 21kHz or so. It has to be due to the Nyquist limit of the digitizing process. So if you accept CD as a realistic rendition of the original music and if you use CD as your source, how can it be beneficial for your amp to reproduce frequencies much above 20kHz?
Now, I agree that the implementation of some amps necessitates higher bandwidth (such as those that use global feedback), but I argue that this is not a requirement of music reproduction.
"Well, first there are individual variations. Most of us can only hear
up to some 20kHz, and the upper limit decreases quite heavily
with age for people in the Western civilization, probably due to
noise and stress etc. However, there are individuals who can hear
well above 20kHz. "
My audiologist (who also teaches at Stanford) says otherwise. Even in non-industrialized societies 20kHz is tops and that is predominately limited to females no older than 20. He claims there are a number of independant field studies confirming this. However, baring disease it is true that hearing loss is greater in industrialized environments. The worst of these tend to be emerging third world areas where workplace health regulation are not affordable.
In "our world", North America and Europe, the upper range for males over 30 diminishes with age fairly steadily. 16kHz as tops is is pretty typical. Whatever one's personal upper limit is it is will be down from "flat" by at least a few db.
There were experiments conducted a few years ago that seemed to point to people being able to indentify sounds above 20kHz. Follow up experiments concluded that it was intermodulation artifactsbelow 20kHz that were heard. As long as the recorded media included the artifacts is was indistigishable from the case where the +20kHz were included. On the other hand, from an audiophile point of view it is probaly not desireable to have similar artifacts generated post playback when they are not resent on the recorded media - i.e., as interactions ocurring in the listing environment.
up to some 20kHz, and the upper limit decreases quite heavily
with age for people in the Western civilization, probably due to
noise and stress etc. However, there are individuals who can hear
well above 20kHz. "
My audiologist (who also teaches at Stanford) says otherwise. Even in non-industrialized societies 20kHz is tops and that is predominately limited to females no older than 20. He claims there are a number of independant field studies confirming this. However, baring disease it is true that hearing loss is greater in industrialized environments. The worst of these tend to be emerging third world areas where workplace health regulation are not affordable.
In "our world", North America and Europe, the upper range for males over 30 diminishes with age fairly steadily. 16kHz as tops is is pretty typical. Whatever one's personal upper limit is it is will be down from "flat" by at least a few db.
There were experiments conducted a few years ago that seemed to point to people being able to indentify sounds above 20kHz. Follow up experiments concluded that it was intermodulation artifactsbelow 20kHz that were heard. As long as the recorded media included the artifacts is was indistigishable from the case where the +20kHz were included. On the other hand, from an audiophile point of view it is probaly not desireable to have similar artifacts generated post playback when they are not resent on the recorded media - i.e., as interactions ocurring in the listing environment.
This reminds me of a test that a onetime technical editor of the deceased Stereo Review Magazine performed (back in 1973 before 'perfect sound forever' made it a punishable offense to admit that energy above 20khz contributed in any way to the listening experience). He took two separate oscillators, one set to 23khz and the other set to 24khz driving two separate amplifiers connected to his left and right speakers. IOW, he ensured that there was absolutely no way for these two signals to interact in such a way as to produce audible difference products before reaching the listener.
He wrote that when he sat in his centered listening chair, lo and behold, he heard a distinct 1khz tone. His conclusion: nonlinearities of the human middle and inner ear easily create quite audible difference tones when presented with only frequencies beyond 20khz.
He wrote that when he sat in his centered listening chair, lo and behold, he heard a distinct 1khz tone. His conclusion: nonlinearities of the human middle and inner ear easily create quite audible difference tones when presented with only frequencies beyond 20khz.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.