Unfortunately, there are such ******* about, and they tend to be loud.
When sitting on any panel, and it wasn't too many, I never even tried to identify the manufacturer of whatever model I was listening to. Instead, I concentrated on whether that sounded as it should to me, never who made it.
So, if I thought it had a recessed treble, for example, I said so. If it was too bass rich, I said so. Who made it, which model was that, what somebody else said and/or wrote about it, how much it cost is of absolutely no concern to me, I'm just listening, not buying, and if I were buying it, I'd still do it that way, although of course, in the end, we would have to get to the price.
And hearing the price usually makes me wonder why would anyone pay that much for that product, sometimes I think it's a fair price for what it offers, and on rare occasions, I think that for what it does the price is a steal. But all that is only subsidiary discussion AFTER a "verdict" has been reached.
Sorry guys to subject many of you to Ron's latest AES paper, when you may not even be able to get it yet, even if you are willing to pay for it.
Last night, I had dinner with a friendly engineer from Bose. I attempted to discuss Ron's paper with him and realized that I still did not know it well enough to be completely successful.
However, the main point that Ron emphasized in this latest paper was the effect of loading on the overall performance of the vast majority of IC's, and this can still be discussed easily.
Designers do make mistakes, and this is one that perhaps can be avoided. On the other hand, it might be a caution to IC designers to not starve their output stages just to keep the total current draw from the IC to some arbitrary spec. Yes, crossover distortion has existed in IC's since the very beginning and is still with us today. Because it often brings out higher order distortion, as well as FM distortion, perhaps it should be noted, and hopefully avoided.
Last night, I had dinner with a friendly engineer from Bose. I attempted to discuss Ron's paper with him and realized that I still did not know it well enough to be completely successful.
However, the main point that Ron emphasized in this latest paper was the effect of loading on the overall performance of the vast majority of IC's, and this can still be discussed easily.
Designers do make mistakes, and this is one that perhaps can be avoided. On the other hand, it might be a caution to IC designers to not starve their output stages just to keep the total current draw from the IC to some arbitrary spec. Yes, crossover distortion has existed in IC's since the very beginning and is still with us today. Because it often brings out higher order distortion, as well as FM distortion, perhaps it should be noted, and hopefully avoided.
This bad Côte de Beaune sample was carefully chosen by the organizers in order to promote Californian wines ?
Or by an unscrupulous exporter, to get rid of a bad lot, thinking that foreigners would not make the difference ?
My question is: Do you think you have changed your preferences if you knew the origin of each glass during this table testing ?
Specially if you have to buy some of them for your own use ?
(It is such a pleasure to find better vines for ridiculous prices because they have not a prestigious name or origin.)
Well, Blind tests do have some advantages in some situations, specially when honesty is involved. It is the only solution when it is about to verify the level of knowledge and accuracy of peoples. Even for ourselves if we have a doubt to be under influence. But is not *always* a requisite in all situations and, sometimes, to be not blind helps.
Did you fail to return a Mouton Rothschild in a restaurant when -it happens- it is oxidized ?
In this last example, it seems obvious to me that, if you are an expert in Mouton Rotchild, knowing the character of each production's years, you will be in a better position to discriminate the fault in a sample if you know the year and origin.
BTW: And I do not put me a blindfold when I try a car on the road before to buy it ;-)
Last edited:
John, this is a real question. What is the influence of harmonic distortions when they are out of our bandwidth range ?
No, it was damaged. SY and I were tasting a particularly rare (1 barrel) Mersault from a very prestigious producer, we started dropping pennies into it to try and save it from being clumsily sulfured. Just sayin' we're not particularly label conscious but have observed enough to see others do it in spades. We can be fooled too so what?
Esperado, that is an important question. Unfortunately harmonic distortion is something relatively easy to measure and model, BUT where there is harmonic distortion there is IM distortion of similar amount. What we often hear when audio harmonics get too high to hear directly are the IM byproducts that are all over the audio spectrum.
Bingo. ALL humans have lying brains, it's how we're wired. We couldn't process sensory data without it.
The difference is whether or not we're willing to admit it.
Ok. I agree, of course, with this.
As well as IM is much worse than harmonic distortion, because it adds unexpected and uncorrelated mud to the musical landscape, while little harmonic distortion, and specialty pair ones, just change a little the character of each instrument, can be in a very natural way.
Those IMs, we can measure them as well. And this measurement is more talkative than harmonics.
What was behind my question is about class A and class B.
When a class AB amp is very fast (high slew rate etc.) the residual crossover distortion produce, on my experience, less IM in the audible range.
When all those distortions are under the audibility threshold (-90dB ?), and covered by the mask effect, what can explain the difference we can still feel with the same amp in Class A ?
some engineers have been "up to speed" on this century old (as applied in electronics) math since our distant University days
have posted here a decade ago:
Just a preliminary view of AM vs FM intermodulation products, Am modulation (green) produces sum and difference products at 10 +/- 1 Hz = 9 and 11 Hz – not truly representative of feedback distortion, 100% AM modulation has removed the 10 Hz “carrier” and in a feedback loop that 10 Hz would be interacting with the products in the nonlinearity to produce additional distortion products that would look like the series of products in the FM modulation fft
The FM fft example (red) shows a series of distortion products at 10 +/- n* 1 Hz which decrease in amplitude by the n_th power, the green AM products @ 9&11 Hz overlay the red FM products @ 9&11 Hz and are essentially the same amplitude
I plotted the products by differencing out the FM fundamental, as you can see the amplitude and character of AM and FM modulation products are very similar (the higher order FM products are included but not discernable by eyeball in the time plot)
If you examine the 9&11 “first order” difference products, they have a relative phase of 180 degrees in both the AM and FM cases; for AM 9&11 Hz are +/-0 90 degrees relative to the (missing) fundamental, in the FM case they are 0, 180 degrees relative to the fundamental
As I pointed out this doesn’t represent the distortion generated under different feedback conditions but it does let you see the defining essential of AM vs FM IM distortion products - the relative phase of the distortion products
The amplitude of the products is determined by the depth of the modulation, (1/10 in this example) the frequencies of the products are the f_0 +/- n*f_mod
The claim that FM distortion can’t be easily measured seems strange, the (complex valued) fft result does contain the phase information of the component frequencies, and since AM and FM distortion products are in quadrature, no cancellation can occur between the products, of course since they are in quadrature and add (vectorially) there seems to be little point in resolving the amplitudes of a particular IM pair into separate AM and FM components when we can use high feedback to squash the summed amplitude of both
For the high feedback camp the argument is basically that if there is no IM product amplitude, then the relative phase of the distortion products can’t matter – no distortion product, no FM distortion
The consternating crux for me is the insistence that FM distortion is somehow worse than higher levels AM distortion – when measured by the amplitude of the distortion products, and that there is some subtle flawed trace left in the signal when the amplitude of the distortion products is reduced below the audibility limits if they were presented on their own in a anechoic chamber, much less in conjunction with the masking of the undistorted source signals, room reverberation, ambient noise and loudspeaker distortion (including Doppler-FM distortion)
Nothing new John tends to discount input from certain quarters. I think to this day he has never bothered to understand the first order crossover removal in the AD797 even though I have posted the images from the AES preprint. Just last month he repeated his opinion that there is nothing there but a folded cascode copied from Harris.
Ok. On the other side, despite all the very fast and sophisticated measurements instruments used on their both cars, what makes the difference between the world champion of Formula One and his second, is most of the time, his ability to 'feel' the road, and to propose more accurate tunings. I believe his brain is not lying too much ;-)
Same reason why armies continue to use human pilots in their fighters. More and more assisted, it is true, to can concentrate on their 'feelings'. Like an audio designer with his measuring instruments ?
Anyway, all this hifi game is just an attempt to... fool our senses. A pure lie.
Last edited:
But is it not the point of the exercise to fool the brain in to perceiving a analog of a real musical performance ?
Yes, but that's a different question than "Can I actually hear the difference between A and B?" If you can, then you can move on to which sounds "better" to you. If you can't, then you can move on to issues which actually matter to the sound. Or you can become a high end huckster.
Well, congratulations Scott, for the AD797. I think that this is a great measuring IC, especially with normal loading. However, many here, and certainly in the outside world use other IC's that might not fare as well, especially when loaded with 500 ohms or less.
My apologies for insulting you in some way. I do think that the original circuit topology is derived, first from Harris, then PMI, but I agree that you did significantly more to make it an 'ideal' IC part compared to most. I have found another part that is as quiet and sounds as good when loaded properly, and less than 1/2 the price. However, it may not measure quite as well, and this could be important in making accurate instrumentation.
My apologies for insulting you in some way. I do think that the original circuit topology is derived, first from Harris, then PMI, but I agree that you did significantly more to make it an 'ideal' IC part compared to most. I have found another part that is as quiet and sounds as good when loaded properly, and less than 1/2 the price. However, it may not measure quite as well, and this could be important in making accurate instrumentation.
Hi,
Another of these little lies is to believe odd order harmonic distortion is "unnatural". It is not.
Distortion is what it is. Any kind of distortion should be avoided. And I really mean any kind.
So far, IMHO. the easiest way to avoid most kinds of distortions is to use valves. Pick the ones that can deliver some decent current and run them in class A. (Which moves the problem to the OPT but so be it)
Or, use the same kind of valve, run it class AB and discard the OPT altogether.
(Which puts the problem elsewhere but it can be solved)
Excessive amounts of GNFB never is a solution as there always is a time delay.
Cheers,
Another of these little lies is to believe odd order harmonic distortion is "unnatural". It is not.
Distortion is what it is. Any kind of distortion should be avoided. And I really mean any kind.
So far, IMHO. the easiest way to avoid most kinds of distortions is to use valves. Pick the ones that can deliver some decent current and run them in class A. (Which moves the problem to the OPT but so be it)
Or, use the same kind of valve, run it class AB and discard the OPT altogether.
(Which puts the problem elsewhere but it can be solved)
Excessive amounts of GNFB never is a solution as there always is a time delay.
Cheers,
Last edited:
Who said that ?
All musical instrument produce harmonics, odd and even. Just odds are more aggressive and distinguishable (Brass section). Some can even find them out of tune (7th, 9th).
And so the wheel turns for the n-th time.
The problem with distortion in general is not that it simply exists, but also its behavor. I have heard power amps with relatively high distortion factors quoted in their spec sheets, but with a "good" decay, 3rd being lower than 2nd, 4th from 3rd, and so forth. When compared with similar amps from other reputable sources, which measured that same distortion an order of magnitude lesser, the first amp sounded better, "more natural", so to speak, easier on the ear.
I belive this is the key reason why we can't agree what the threshold of hearing distorion is. The play-it-safe people will simply dump it all into one bag and say tat we just need to keep it below say -90 dB level, and technically, that's probably right, but the problem is that such devices still may not be able to sonically match other devices which measure much worse, but still sound better.
Before unleashing the ack-ack fire on me, consider tube devices. They consistently measure sometimes catastrophilally bad compare to solid state devices, yet manage to sound better, sometimes even much better. Yet, they measure like 3% THD and 3% IM. So, obvuiously, we don't know it all yet, so we still have to choose what we like.
Ditto for op amps. They look the same, they generally do the same, but they just don't sound the same. How come? And the comment I have heard on a number of occasions, that the op amp designers usually know zilch about audio is too stupid to even comment on.
But we do NOT know it all from measurements alone. Or, perhaps we are not measuring the right stuff yet, or fail to correlate our findings properly, or something. If it were not so, everybody would be making perfect amps.
The problem with distortion in general is not that it simply exists, but also its behavor. I have heard power amps with relatively high distortion factors quoted in their spec sheets, but with a "good" decay, 3rd being lower than 2nd, 4th from 3rd, and so forth. When compared with similar amps from other reputable sources, which measured that same distortion an order of magnitude lesser, the first amp sounded better, "more natural", so to speak, easier on the ear.
I belive this is the key reason why we can't agree what the threshold of hearing distorion is. The play-it-safe people will simply dump it all into one bag and say tat we just need to keep it below say -90 dB level, and technically, that's probably right, but the problem is that such devices still may not be able to sonically match other devices which measure much worse, but still sound better.
Before unleashing the ack-ack fire on me, consider tube devices. They consistently measure sometimes catastrophilally bad compare to solid state devices, yet manage to sound better, sometimes even much better. Yet, they measure like 3% THD and 3% IM. So, obvuiously, we don't know it all yet, so we still have to choose what we like.
Ditto for op amps. They look the same, they generally do the same, but they just don't sound the same. How come? And the comment I have heard on a number of occasions, that the op amp designers usually know zilch about audio is too stupid to even comment on.
But we do NOT know it all from measurements alone. Or, perhaps we are not measuring the right stuff yet, or fail to correlate our findings properly, or something. If it were not so, everybody would be making perfect amps.
- Status
- Not open for further replies.