Entertainment value, of course! (Seems I recall someone, somewhere saying that audio is part of the entertainment industry. Ahem...)
Some people derive their joy from listening...others from reading meters.
Grey
Some people derive their joy from listening...others from reading meters.
Grey
Aye, Nelson, but the difference being that I know I use subjective judgment in assessing audio gear. Those who allow distortion measurements to form their opinions for them are often unaware that they are doing exactly the same thing.
I have no problem whatsoever with someone who wishes to design and build a low distortion circuit as a purely technical exercise any more than I do with someone who wants to build a 5kW amp even if they'll never use all the power. Anyone who pulls off a technical feat like that is welcome to any bragging rights that might accrue.
I do have a problem with those who insist that nothing exists beyond their chosen three or four parameters, usually some combination of THD, power (only relevant for amps, obviously), S/N ratio, and IM (less common). Those tests have been known for decades and have been flogged to death. Unfortunately, the standard stuff doesn't explain things like amps with the same specs that sound different.
Something that I'd like to see--and this would be a psychological test, not an audio test--is a metric that would give an indication of how attentive to detail people are. And, yes, it would have to be defined as an audio attentiveness. (My father and both uncles worked in the textiles industry in various capacities and could discriminate shades of colors far beyond anything I can see...but didn't give a flip about music. They were terrible audiophiles.) People who had low attention to detail scores would not pick up on the sorts of subtleties that people with higher scores used to make decisions as to what's good, better, and better-est (ain't no such thing as best). I know I listen to some pretty minute little things when I'm in analytical mode. And I know from experience that a lot of people tend to paint with broader brush strokes, so to speak. If they've got the beat, the pitch, and the melody, they've got all that matters to them and they tend to look askance at those who might listen for other things in addition to the overall...shape, for want of a better term, of the music.
It's not a zero sum game. You can have the overall shape and the details, all at the same time. But to listen to some folks, you'd think they were being robbed of something. Never understood that attitude, myself.
Grey
I have no problem whatsoever with someone who wishes to design and build a low distortion circuit as a purely technical exercise any more than I do with someone who wants to build a 5kW amp even if they'll never use all the power. Anyone who pulls off a technical feat like that is welcome to any bragging rights that might accrue.
I do have a problem with those who insist that nothing exists beyond their chosen three or four parameters, usually some combination of THD, power (only relevant for amps, obviously), S/N ratio, and IM (less common). Those tests have been known for decades and have been flogged to death. Unfortunately, the standard stuff doesn't explain things like amps with the same specs that sound different.
Something that I'd like to see--and this would be a psychological test, not an audio test--is a metric that would give an indication of how attentive to detail people are. And, yes, it would have to be defined as an audio attentiveness. (My father and both uncles worked in the textiles industry in various capacities and could discriminate shades of colors far beyond anything I can see...but didn't give a flip about music. They were terrible audiophiles.) People who had low attention to detail scores would not pick up on the sorts of subtleties that people with higher scores used to make decisions as to what's good, better, and better-est (ain't no such thing as best). I know I listen to some pretty minute little things when I'm in analytical mode. And I know from experience that a lot of people tend to paint with broader brush strokes, so to speak. If they've got the beat, the pitch, and the melody, they've got all that matters to them and they tend to look askance at those who might listen for other things in addition to the overall...shape, for want of a better term, of the music.
It's not a zero sum game. You can have the overall shape and the details, all at the same time. But to listen to some folks, you'd think they were being robbed of something. Never understood that attitude, myself.
Grey
GRollins said:
is this really the main cause of distortion? don't even the most linear devices have distortion measurements several orders of magnitude higher than what could be expected by this explanation?
i would be interested in seeing some sample calculations that could explain how such a tiny delay could cause the distortion even at 0.000001% (already far better than any device i know of).
one thing that feedback also definitely does is reduce distortion and as i think nelson has pointed out, the important question is: how much feedback can we use before the distortion added begins to offset the distortion taken away by feedback.
okapi said:
If the signal is a sin wave,
But with nusic nwhich is not repeating, by the time the feedback gets back to where it is injected the music that it is a small (negative) copy of, the music it is meeting is different. It is attempting to counter the distortion in a portion of the music that has already moved further on in the circuit or the system, so in effect the feedback is added distortion -- you are mixing the music with an echo of something from the past.
dave
The reason a man or woman chooses one amp over another is personal.
They should not feel threatened if other people use different criteria to make THEIR choice.
They should not feel threatened if other people use different criteria to make THEIR choice.
planet10 said:
Seems some argue you should then make it really really fast.
(i like anything big. Boring, i know.)
planet10 said:
music is multiple sin waves added together. other than that you are simply repeating grollins point. my point is that this delay is likely not a significant cause of the distortion we see as a result of feedback. Also, feedback does in fact eliminate some of the distortion that exists as a result of the non linearity of the device even in a musical wave form.
fredex said:
i certainly don't dispute someone's choice in amplifiers but i think if we are going to make progress in creating a device that more accurately duplicates it's input signal then we need to have tools to do this. Although observation (listening) is one of those tools it cannot be the only tool because it is often biased. therefore, i think that observation used with measurement is essential. in other words, the key is to capture the observations in measurement so that they can be quantified and then effectively communicated. just saying something sounds better and expecting everyone to agree tends to get us just where we are - arguing subjective viewpoints.
I didn't expect to have trouble with the concept that test signals and music are different.
The point that's being missed is that music does not consist of a continuous, repetitive signal, regardless of waveform. The idea that you can deconstruct a waveform using Fourier analysis, although useful for other things, is irrelevant for present purposes. It is precisely the complex nature of real music that defeats steady state tests. An impulse--the attack of the waveform--passes. It does not repeat. The feedback arrives too late to correct the blunted peak. That's the entire point.
Might I suggest spending some time watching music...not test tones...on an oscilloscope? It's a hashy mess, not the tidy, orderly thing that many assume it to be.
Grey
The point that's being missed is that music does not consist of a continuous, repetitive signal, regardless of waveform. The idea that you can deconstruct a waveform using Fourier analysis, although useful for other things, is irrelevant for present purposes. It is precisely the complex nature of real music that defeats steady state tests. An impulse--the attack of the waveform--passes. It does not repeat. The feedback arrives too late to correct the blunted peak. That's the entire point.
Might I suggest spending some time watching music...not test tones...on an oscilloscope? It's a hashy mess, not the tidy, orderly thing that many assume it to be.
Grey
GRollins said:
just my tuppence worth.
i would think it's an issue of the instantaneous heat modulation, rather than the long term.
60 dB changes, due to audio changes of 60dB
myhrrhleine said:
I agree. There was an amp (produced I think) where all the small signal transistors were mounted on the main heatsink which ran hot. The idea is to heat them to a higher temp than they would be normally thus 'swamping' any thermal effects due to signal variations. As almost all semi conductor characteristics change with temperature it seems a very reasonable theory to me.
An impulse--the attack of the waveform--passes. It does not repeat. The feedback arrives too late to correct the blunted peak. That's the entire point.
Can a delay be measured if there is one ?? Surely there must be away to do this even if its .0001 seconds and has it been done ?? This way the NGFB argument would be settled. This is all wishfull thinking until proven. If it has been done please put the reference here, I for one will stop bothering looking at FB designs.
I have built a NGFB amp too see what all the hype was about, a simple design from the net, with mixed results, in some areas its better and some worse than one with feedback. I compared this to a highloop gain very low distortion amp of mine, THD of anything above h3 is at least at -120db level. My conclusion is that feedback does introduce new higher order distortion but if used correctly these distortions can be so low that noone can hear them.
Can a delay be measured if there is one ?? Surely there must be away to do this even if its .0001 seconds and has it been done ?? This way the NGFB argument would be settled. This is all wishfull thinking until proven. If it has been done please put the reference here, I for one will stop bothering looking at FB designs.
I have built a NGFB amp too see what all the hype was about, a simple design from the net, with mixed results, in some areas its better and some worse than one with feedback. I compared this to a highloop gain very low distortion amp of mine, THD of anything above h3 is at least at -120db level. My conclusion is that feedback does introduce new higher order distortion but if used correctly these distortions can be so low that noone can hear them.
fredex said:
That's more or less what I was describing above with the power resistor adding heat when the device cooled. To work, you'd have to run the system at a temperature somewhat above the "normal" operating temperature of the transistor. Whether it would work or not remains to be seen, but at least it's conceptually simple.
homemodder said:
...And if you could stop those pesky rogue cells from dividing, you'd have a cure for cancer. It's easy to wave your hands and describe, in the abstract sense, how to fix a problem. It's another matter entirely to get results in the real world.
Bummer.
I'd also like to note in passing that cancer research is a lot better funded than audio research. Not saying it shouldn't be--audio isn't a life or death matter, whereas cancer is--it's just one of those facts of life.
People have argued about why negative feedback does what it does for years. The inherent time delay is the best hypothesis I've heard yet, but I'm open to other suggestions. Those who believe that negative feedback is a gift from the gods will still swear that it's manna from heaven, no matter what measurements are made. Any who suggest otherwise are clearly infidels and should suffer the fate of all unbelievers.
There's an interesting parallel between high negative feedback and early digital. No, I am not saying that they are the same, so don't try to go there. It's an analogy, nothing more. Or maybe there is something there, who knows? It goes like this...early to mid-period Red Book recordings had an unnatural silence between notes. Digital boosters pointed to this as a "benefit" to the digital recording medium. Those who listened said it was too dead. It wasn't a signal to noise ratio issue, it was a moment in the music when there should have been reverberation in the hall, even if there were no notes being played. Instead the notes disappeared. Dead. Gone before their time. The decay of the notes in the hall was truncated. Battles were fought and the ears won (finally). It turned out that the analog to digital conversion process was dropping low level information. Once the signal dropped below the least significant digit, it went to zero, not some arbitrary value in between the way it would on a system with more bits. One thing they did to fix the problem was rather paradoxical; to keep that LSB from turning off prematurely, they injected noise into the signal. (And, yes, more bits helped later on...) Ta da! It was a band-aid, but it worked.
Okay, so what's this got to do with feedback?
One of the characteristics that you tend to see in high feedback designs is a deep, dark, velvety silence. A little too silent, if you follow what I'm saying. It's not as dramatic as the digital problem, but it is reminiscent. Now, one of the things feedback does is lower noise. Fair enough. You want it to do that. But what if it (different mechanism, but with the same end result) is suppressing low level information such as ambience and subtle cues to instrumental timbre? By definition, hall ambience approaches uncorrelated noise. In fact, one of the problems you run into when building a digital delay is to turn what would normally be a sharp, slap echo into a gentle, gliding sussuration. In other words, you have to destroy the correlation inherent in the digital echo. In short, what if feedback sees ambience and other low level information as noise...and obliterates it?
Ponder on that one. I don't insist that it's true. It's just a combination of my listening observations and a lot of head scratching.
Non-feedback amplifiers have advantages and drawbacks. A decent circuit tends towards better imaging (which is of interest to me) and more natural instrumental timbre. On the downside, the damping factor is generally sorta low, so that tends to limit what you can accomplish in the low end...or at least you have to choose your speakers carefully. With that in mind, it's not a cure for all ills.
In my case, I'm working towards servo subwoofers (hence feedback) and low-to-no feedback from the midrange up. Woofers? Hmmm... The jury is still out on that one. I'll make up my mind when I get there. Right now what I need is time to build the circuits I have in mind. Unfortunately, time is in short supply for me.
Grey
GRollins said:
Whoops, looks like I stood on the toes of one of my heroes! Grey, I'll always be indebted to you for the Aleph-X thread. Perhaps my tone was too strong.
I see you lump the THD believer in with the DBT believer, into one True Believer. Surely they are completely different; the first believes measurements are everything, the second that listening is everything. Let's keep them separate, for that is what they are.
The DBT believer is in a subset of "those who listen", although obviously a different subset from you, Grey. I cannot resist arguing that they are in the more discriminating subset.
The agony and the anger of "those who listen" who are anti-DBT arises when uncontrolled listening leads them to some pretty exciting perceptions, which the DBT crowd cannot replicate with controlled listening tests.
For some reason the next step of the uncontrolled listeners is to attack controlled listening as somehow flawed.
Whatever flaws there may be in controlled listening test methods, there is no doubt that the flaws in uncontrolled listening are thousands of times (not tens, not hundreds) greater. So great as to render them invalid. Ask any scientist who has done the hard yards, what value they would put on the outcomes of an uncontrolled test.
GRollins said:
Show me any audio system from any era with 0.0001% distortion at the listener's ear.
P.S. what relevance or significance do "those who listen (uncontrolled)" assign to NP's article that started this thread? LOW, because it deals with distortion measurements? Or HIGH, because it post-rationalises their conclusions from uncontrolled listening experiences?
The thermal rise time of an IRF240 is about .1 seconds. In a pure
Class A amplifier, the maximum thermal rise time will occur somewhere
around 1.5 Hz, and the temperature change will be proportional to
the dissipation. The temperature variation due to an AC signal should
decline proportionally to the inverse of frequency above that.
Is it measurable? Yes. Is it big? No. Is it important? Who knows.
😎
Class A amplifier, the maximum thermal rise time will occur somewhere
around 1.5 Hz, and the temperature change will be proportional to
the dissipation. The temperature variation due to an AC signal should
decline proportionally to the inverse of frequency above that.
Is it measurable? Yes. Is it big? No. Is it important? Who knows.
😎
This may help to keep all semiconductors at a more constant temperature and thus stabilizing operating points.fredex said:
Unfortunately it has no effect on the modulation of the die temperature due to the music signal (i.e. frequencies above some Hz), as the die still has its own thermal constant and the same thermal resistance to the case.
One thing which helps is to run relatively high currents (compared to the current modulation due to music signal) through all stages and not only the output stage.
Current is good 😀
Running hot is good
Tino
homemodder said:
Coming from a digital circuits background, I had a hard time trying to understand NFB. How does it work? we're half-way to the next clock tick by the time the signal comes around, and the train already left the station.
clearly I overestimated how well it worked, hence my confusion.
zinsula said:
Don Catino,
boils down to the same large thermal inertia thought.
(how about we shave half of the semiconductor case off to expose the dies and submerge the entire circuit in thermal oil ? Oh dear, me and my submariner stuff again
)