Preference vs. audibility - please keep them separate.

[NorthStar]

Hope you don't mind if I disagree with you there, Northstar. I suggest that a systems-based approach is the next logical step.

To the contrary, it makes for better and more informative discussions. ...Besides, you are not truly disagreeing, you are simply adding/contributing.

 

[NorthStar]

slew rate/risetime, which is a function of bandwidth.

I was waiting for that one.

Bandwidth - Rise Time relationship

And: http://www.dataforth.com/catalog/pdf/an121_RiseTime-BW.pdf

 

[NorthStar]

I'm sure, but my point is that changes at 50 KHz are inaudible. So the better place to look for changes is within the audible band.

--Ethan

Within the audible human range there are outside influences that contribute to our hearing (audibility). ...And those outside influences are way up there in the ultra high and ultra low frequencies. ...In real life as inside of our electronics (audio components).

* 50kHz translates to 5kHz | 100kHz translates to 10kHz | 200kHz translates to 20kHz (tenfold).
And: 500Hz translates to 50Hz | 200Hz translates to 20Hz | 100Hz translates to 10Hz | 50Hz => 5Hz | 20Hz => 2Hz | 10Hz => 1Hz (x 10).

 

[Atmasphere]

I'm sure, but my point is that changes at 50 KHz are inaudible. So the better place to look for changes is within the audible band.

--Ethan

I think I said that, but not the same way -although I was more interested in the why of it. In the case of that preamp, the *cause* of the symptom of brightness was at 50KHz. But the *symptoms* were in the audio passband; our ear/brain system interpreting the unnatural group delays as a brightness.

Its sort of like how so many things taste like chicken Our ears translate various HF distortions as 'brightness'.

 

[jkeny]

delightful post!

I'm sure, but my point is that changes at 50 KHz are inaudible. So the better place to look for changes is within the audible band.

--Ethan

Yes, wasn't that point already covered in that we are dealing with a stereo signal & therefore difference signals can be perceived in the audio band?

BTW, when searching for Dr Herbert Melchur, I didn't find anything until I searched further & found Dr, Herbert Y Meltzer here http://www.clp.northwestern.edu/faculty-and-staff/bio/herbert-y-meltzer-md

 

[j_j]

Within the audible human range there are outside influences that contribute to our hearing (audibility). ...And those outside influences are way up there in the ultra high and ultra low frequencies. ...In real life as inside of our electronics (audio components).

* 50kHz translates to 5kHz | 100kHz translates to 10kHz | 200kHz translates to 20kHz (multiples of ten).
And: 500Hz translates to 50Hz | 200Hz translates to 20Hz | 100Hz translates to 10Hz | 50Hz => 5Hz | 20Hz => 2Hz | 10Hz => 1Hz (x 10).

Um, how about some evidence that this is really true?

Hint: It's not. Vinyl and Tape are living proof, Vinyl is pressed to reliably do 30-18K more than once or twice, but it sounds good. Tape can do 20-20K, but only with very reduced levels at high frequencies. It sounds good. (yes, I know about the various 4-channel audio systems, and, thank you, let's look at how they survive many playbacks.)

So ...

 

[j_j]

You go for wide bandwidth in the electronics to reduce phase shift across the audio passband. We *already* have bandwidth-limited media, and bandwidth-limited loudspeakers.

Uh, not really.

Wideband amps have a higher unity gain bandwidth in general, and have much better performance at in-band frequencies, because the feedback has not disappeared at those lower frequencies.

That's the biggest mechanism.

A "fast" amp does not have THD start to roll off at 1kHz. A "slow" amp may have THD rolling off at 100Hz.

And the idea that transient speed matters is bogus as well, as long as you don't get into slope overload.

I am well aware of more than a few mistaken ideas about this issue, but for the most part, they are indeed mistaken.

 

[j_j]

Yes, wasn't that point already covered in that we are dealing with a stereo signal & therefore difference signals can be perceived in the audio band?

Only at 120+ dB SPL.

I really hope you're not listening to that.

 

[NorthStar]

Um, how about some evidence that this is really true?

I said it as my own theory; but I can search for scientific proofs. ...Give me some time please.

Hint: It's not. Vinyl and Tape are living proof, Vinyl is pressed to reliably do 30-18K more than once or twice, but it sounds good. Tape can do 20-20K, but only with very reduced levels at high frequencies. It sounds good. (yes, I know about the various 4-channel audio systems, and, thank you, let's look at how they survive many playbacks.)

So ...

...To follow...

* BTW, why are they usually applying a filter above 100kHz (or 50kHz) with DSD hi-res audio stream (from SACDs)?
Also, subsonic low frequencies of say from 0.1Hz to 5.0Hz; they don't transmit (spill) higher up in the normal human audio range (20-20)?
...Like earth's vibrations, rumbles from volcanoes, thunderstorms, jetplanes flying low (or even higher up on a cloudy day).

And same for ultrasonic high frequencies; spilling down into the human audio range.

- And if it is in our electronics (electrical current), then the same applies.

 

[jkeny]

Only at 120+ dB SPL.

I really hope you're not listening to that.

Ah, right, missed that bit of info, thanks

 

[j_j]

* BTW, why are they usually applying a filter above 100kHz (or 50kHz) with DSD hi-res audio stream (from SACDs)?

Well yes. The high-frequency noise drives amplifiers, preamps, and tweeters nuts. It has burnt out a few tweeters, at least that's what anecdotes say.

There is a LOT of high frequency noise. Bear in mind that consumer SACD is a delta-sigma convertor, the high frequency noise is that of a 1 bit (yes, really) convertor. That is, I repeat, a bloody LOT of noise.

Also, subsonic low frequencies of say from 0.1Hz to 5.0Hz; they don't transmit (spill) higher up in the normal human audio range (20-20)?

Loudspeakers choke on it, and there is a lot of it. All it would do is heat the voice coils, waste power, and push loudspeakers into compression. Remember that airplanes, wind, highways, etc, all provide a substantial amount of power in that frequency range, but it's completely inaudible, and only creates sensation when it's REALLY big, big enough to either create "wind noise" via turbulence or things blowing around. Bear in mind that every octave below 700Hz pushes the absolute threshold up by 6dB, give or take due strictly to the eardrum response. In addition, the inner ear also has reduced sensitivity. The table below is only the part due to the eardrum:

Frequency | dB down
__________________
350 -8
175 -13
87 -18
44 -24
22 -30
11 -36
5 -42
3 -48
1 -54
.7 -60

------------------

As you can see, sensation becomes unlikely via the ear at very low frequencies. Ergo, they aren't very interesting.

...Like earth's vibrations, rumbles from volcanoes, thunderstorms, jetplanes flying low (or even higher up in a cloudy day).

And same for ultrasonic high frequencies; spilling down in the human audio range.

- And if it is in our electronics (electrical current), then the same applies.

Spilling in the context you are using it means "distortion". Frequencies, in and of themselves, in a linear system, do not "spill", they can be broadened by windowing, but that's about it. High frequency distortion can happen in tweeters, etc, but is usually not a good thing.

In-band distortions CAN be a thing that sounds good. Look at a guitar speaker (electric, obviously). It is part of the instrument's spectral response. Ditto foot pedals, etc. Ditto for vinyl, tape, etc. Distortion is not always bad.

Consider: When we reproduce a signal through 2 speakers, we are reproducing 2/8ths of the original information at THOSE TWO POINTS. In order to reproduce a real, similar soundfield analytically, thousands of channels (or more) are required.

Stereo is throwing away 99.9% plus of the information in a soundfield, so there is nothing to argue that some distortion can hurt, after all we've already thrown out most of the information, and "exact" isn't even remotely possible.

 

[NorthStar]

Spilling in the context you are using it means "distortion". Frequencies, in and of themselves, in a linear system, do not "spill", they can be broadened by windowing, but that's about it. High frequency distortion can happen in tweeters, etc, but is usually not a good thing.

In-band distortions CAN be a thing that sounds good. Look at a guitar speaker (electric, obviously). It is part of the instrument's spectral response. Ditto foot pedals, etc. Ditto for vinyl, tape, etc. Distortion is not always bad.

Consider: When we reproduce a signal through 2 speakers, we are reproducing 2/8ths of the original information at THOSE TWO POINTS. In order to reproduce a real, similar soundfield analytically, thousands of channels (or more) are required.

Stereo is throwing away 99.9% plus of the information in a soundfield, so there is nothing to argue that some distortion can hurt, after all we've already thrown out most of the information, and "exact" isn't even remotely possible.

Yes, thank you, that's exactly what I meant; distortion from the outside human range manifesting audibly. I didn't express myself correctly; you did.

____________________

To me, it sounds more safe to play music from an analog playback system (TT, R-2-R) than a digital one (hi-res audio files, DSD, ...);
because with analog you have less chances to blow things up (tweeters, subwoofer's overdriving, ...).

Yes, distortion exist in both, from various manifestations; needle inside the grooves (friction), motor's speed, belts from transports (TTs, R-2-R decks, CD/SACD players), IMH, THD, jitter, RIAA non-adjusted curve, phase anomalies, various harmonics (digital 1st, 2nd, 3rd, 4th, ...), airborne vibrations, driver's feedback, subsonic, etc., and within the audio range, and outside of it.
...Plus, from our own electrical supply (grid), 60Hz, electrical appliances (fridge, halogen lights, radios, switches, ...), wires (them too introduces distortion by receiving radio frequency waves), etc.

From amplifiers, preamplifiers (volume control, balance control, tone controls, ...), loudspeakers (drivers, crossovers, enclosures, vibrations, resonances, ringing, ...), sources like DVD/BD players, phono preamps, surround sound processors, tonearms, cartridges, ...

* Audio is more complex than just the music that is playing. The sum of all the parts plays an important role at eliminating all kind of undesirable noise, and some of that noise is pleasant for some of us as you just said.
This hobby is like a compromise between two very different worlds; preference and audibility.

For some silence is the ultimate peace, others prefer the smooth comfort of live music (classical chamber music, solo piano or cello, smooth jazz, ...), and others like Rock&Roll and Heavy Metal music with all the high decibels of distortion that normal human beings cannot abnormally support. ...Live or/and reproduced. ...Be it Blues, Rock, Jazz, World, Electronica, New Age, you name it.

_________________

Questions: Are some electronic audio components (including mechanical loudspeakers) better at reproducing some specific music genre?
And is there such system(s) that's good/best for all?
Last, should we get familiar with all type of distortion, audible and non-audible? ...And what true relationship measurements have on audibility and preference?

 

[Groucho]

Wideband amps have a higher unity gain bandwidth in general, and have much better performance at in-band frequencies, because the feedback has not disappeared at those lower frequencies.

That's the biggest mechanism.

A "fast" amp does not have THD start to roll off at 1kHz. A "slow" amp may have THD rolling off at 100Hz.

And the idea that transient speed matters is bogus as well, as long as you don't get into slope overload.

I am well aware of more than a few mistaken ideas about this issue, but for the most part, they are indeed mistaken.

j_j (and anyone else)

Could you explain a little more about this issue, because it certainly confuses me, for one.

I can see that an amplifier that is 'fast' when open loop is a good thing, as you say for maintaining high levels of feedback at higher frequencies (I think). But closed loop we can tailor the amplifier's high frequency response to be what we like. Would you allow the amp to pass 300kHz at full (closed loop) gain or would you begin to roll off at a lower frequency?

What would be the main reason for this frequency response tailoring? (e.g. Improved stability?, less likelihood of passing RFI?)

What are the relative merits of implementing high frequency roll off within the amp's feedback loop vs. doing some of it before the input and making the amp's closed loop bandwidth wider?

I can see that attempting to exceed the amp's intrinsic slew rate limit is never going to be a good thing, but having a high frequency component simply low pass filtered by the amp and its closed loop frequency roll off doesn't sound like much of a problem. If an intrinsically fast (open loop) amplifier can do 30kHz at full power (closed loop), and our source is bandwidth limited to 22 kHz, say, is the amplifier not always going to be 'fast' enough anyway?

Thanks!

 

[j_j]

j_j (and anyone else)

Could you explain a little more about this issue, because it certainly confuses me, for one.

I can see that an amplifier that is 'fast' when open loop is a good thing, as you say for maintaining high levels of feedback at higher frequencies (I think). But closed loop we can tailor the amplifier's high frequency response to be what we like. Would you allow the amp to pass 300kHz at full (closed loop) gain or would you begin to roll off at a lower frequency?

What would be the main reason for this frequency response tailoring? (e.g. Improved stability?, less likelihood of passing RFI?)

What are the relative merits of implementing high frequency roll off within the amp's feedback loop vs. doing some of it before the input and making the amp's closed loop bandwidth wider?

I can see that attempting to exceed the amp's intrinsic slew rate limit is never going to be a good thing, but having a high frequency component simply low pass filtered by the amp and its closed loop frequency roll off doesn't sound like much of a problem. If an intrinsically fast (open loop) amplifier can do 30kHz at full power (closed loop), and our source is bandwidth limited to 22 kHz, say, is the amplifier not always going to be 'fast' enough anyway?

Thanks!

Man, that's a whole PhD thesis in a paragraph.

The problem is that "it all depends" and on rather a lot of things to do with the amp design.

 

[andy_c]

I can see that an amplifier that is 'fast' when open loop is a good thing, as you say for maintaining high levels of feedback at higher frequencies (I think). But closed loop we can tailor the amplifier's high frequency response to be what we like. Would you allow the amp to pass 300kHz at full (closed loop) gain or would you begin to roll off at a lower frequency?

What would be the main reason for this frequency response tailoring? (e.g. Improved stability?, less likelihood of passing RFI?)

What are the relative merits of implementing high frequency roll off within the amp's feedback loop vs. doing some of it before the input and making the amp's closed loop bandwidth wider?

Regarding the frequency response tailoring, the purpose is to keep the amplifier stable while trying to get as much feedback as possible. What constitutes a "fast" open-loop amplifier is not what one might think at first. It seems at first that "fast" should be defined in terms of the bandwidth of the open-loop amplifier, but it's actually determined by its gain-bandwidth product. There's lots of information on this subject in this article by Bruno Putzeys (PDF).

For example, take an amplifier with an open-loop bandwidth of 20 kHz and a DC open-loop gain of 1000. Its gain-bandwidth product (GBW) is 20kHz * 1000 = 20 MHz. It's also possible to have an amplifier with a DC open-loop gain of 1 million and an open-loop bandwidth of 20 Hz (yes, 20 Hz). Its GBW is also 20 MHz. When the same feedback network is placed around both amps, giving the same nominal gain for both amps, the closed-loop bandwidth of both amps will both be the same in the absence of the usual input low-pass filter and output inductor/resistor network. The latter amplifier will have the same amount of feedback at frequencies above 20 kHz as the first one, but much more than the first one at frequencies below that. See the linked article for more information.

 

[Phelonious Ponk]

I'm completely confused. It sounds like you guys are saying that supersonic amplifier extension has to do with getting lots of feedback, yet this all started with Ralph (atmasphere) who believes amps should have as little feedback as possible? Did I get that wrong? Lord knows I could have. So, very basic question for an uber-civilian with a pretty highly evolved skeptic response?

Is any of this clearly audible or is this another high-end solution to a non-problem?

Tim

 

[Orb]

Bruno also mentions it in his Master classes presentations (beyond negative feedback), I think may possibly be found on the internet.
Tim, what Bruno says that either you have lots of negative feedback or very little, inbetween creates the problems.

Cheers
Orb

 

[Orb]

Just to add, I thought one reason for wide bandwidth JFETS/circuitry was to also improve or relates to stability/noise/reduce parasitic oscillation in some designs - appreciate I may be very wrong.

Cheers
Orb

 

[andy_c]

For another perspective on feedback (from a tube amp designer) see this interview with Scott Frankland. He does a good job with the issue Tim brings up, but without resorting to complicated math.

 

[Gregadd]

Perhaps a better analogy might be acceleration. A car can maintain a 60 mph speed with about 12 horsepower. Where it comes in is reaching 0 to 60 in 5 seconds may require horsepower in the 100's. That may be the difference. You also want to stay out of the redline less other bad things happen.