MarkS
Major Contributor
- Joined
- Apr 3, 2021
- Messages
- 1,078
- Likes
- 1,514
It's purely hypothetical.I'd like to see a cite or evidence for that.
It's purely hypothetical.I'd like to see a cite or evidence for that.
Right, but there’s an assertion that a *particular type and level of distortion* will make it sound “harsh”. We can at least try that level of distortion and see if it can be heard at all. I think that is the hypothesis to be tested, not the question of whether the Adcom is audibly different.A comparison of artificially doctored and undoctored signals is not equivalent to a comparison of actual amps. The doctoring is far too simplistic for this to be valid.
FWIW, I'm in the "all amps sound the same" camp, and have been for decades, because human hearing is just not very good at detecting the miniscule flaws in the vast majority solid-state amps.
But I still don't think a doctored-signal test is a valid substitute for comparing actual amps.
Wouldn't determining if a "particular type and level of distortion" is audible be the follow up test?Right, but there’s an assertion that a *particular type and level of distortion* will make it sound “harsh”. We can at least try that level of distortion and see if it can be heard at all. I think that is the hypothesis to be tested, not the question of whether the Adcom is audibly different.
Both assertions have been made, one is easier to test. @atmasphere has made the claim about types of distortion, and used the Adcom as an example.Wouldn't determining if a "particular type and level of distortion" is audible be the follow up test?
It seems to me we should first determine IF there are audible differences between amps. If so, then you look at distortion and other characteristics to discover the reason for the difference.
It's a capital mistake to determine the next experiment before knowing the outcome of the first one.Wouldn't determining if a "particular type and level of distortion" is audible be the follow up test?
It's a capital mistake to determine the next experiment before knowing the outcome of the first one.
So in this post, and the ones following, @atmasphere lays out the case that higher-ordered harmonic distortion (HOTHD) sounds harsh, when not masked by lower order. The responses, which started all this kerfuffle, suggest that the higher-order harmonics in most amplifiers are inaudible, and therefore this is a specious assertion. An example is requested, which is when the Adcom came in - as an amp with audible HOTHD but no lower order to mask it. But the *original* assertion is the harshness of unmasked higher-order harmonics.I forgot to mention that the ear has a masking principle that is at play when dealing with higher ordered harmonics. Its well known that the 2nd and 3rd harmonics can mask the presence of the higher orders if in sufficient quantity (this is why SETs sound so smooth despite having more higher ordered harmonics than any solid state amp). In the case of a 'properly designed' SET (which will be perceived as 'musical' sans feedback) it is imperative that the higher ordered harmonics fall off on an exponential curve. In this fashion they are masked.
There are two other inconvenient facts regarding the ear. The first is that it uses higher ordered harmonics to sense sound pressure (easily proven with simple test equipment). The second is that it assigns tonality to any form of distortion (harmonics, IMD or aliasing). Higher ordered harmonics if unmasked are assigned 'harsh and bright'; that latter bit is the part we've known since the 1930s.
Quite often in solid state amps which employ feedback (most SETs do not), the feedback is fairly successful at suppressing the innate distortion of the amplifier. But due to several problems with the application of the feedback, its application results in generation of higher ordered harmonics. This problem has been around for a very long time (see the writings of Norman Crowhurst back in the late 1950s). It comes from 3 different issues:
1) the feedback node isn't linear and so the correction signal becomes distorted before it can do its job.
2) Gain Bandwidth Product in the design might be insufficient to support the feedback at all frequencies in the audio band. This problem results in increasing distortion with frequency since at the turnover point (usually above 1KHz) the feedback decreases and so distortion climbs.
3) the phase margin of the amplifier might be insufficient to allow for higher amounts of feedback without stability issues. Phase margin is that value where at some frequency, the feedback has so much phase shift that it is positive rather than negative. Positive feedback results in oscillation so usually this is an upper limit of sorts. Of course if the feedback loop is properly designed this problem can by mitigated quite a lot.
So in such an amp with the problems listed above (which is the majority of solid state amps make since the inception of the transistor) there will be a perception of 'harsh and bright' at some volume setting (and this is why tube amplifiers are still made; you don't have to know anything technical to understand this fact; its economics). You can mitigate this to some degree by reducing treble output in the speaker but since the problem is caused by distortion rather than a frequency response error this will be found to be an inadequate solution.
Self oscillating class D amps do provide a solution to this rather classic problem. In a self oscillating amplifier, so much feedback is applied that the amp goes into oscillation as soon as its powered up. The feedback loop is then designed so that the amp can only find one oscillation frequency; that is used as the switching frequency. Finally, the feedback is mixed with the incoming signal outside of the active signal path (in the same way its done in opamps) so the correction signal has far less distortion when it arrives to mix with the incoming signal to do its job. This all means you can have much more feedback than was previously practical, without Gain Bandwidth Product problems and using a more accurate feedback signal. The Purifi module is currently the best example of this approach.
Not willy nilly (I do a lot of those in my professional life) but just the opposite- trying to plan future experiments before knowing the outcome from the first ones. I used to make my teams laugh when I'd chart the research plans; "OK we do XYZ, then see what happens," and terminating the Gannt chart right there.I guess that Geiger and Marsden’s experiment was a capitol mistake by Rutherford then?
The way I heard is was that Rutherford had them do the experiment again measuring all back the way back in angle to near 180 degree, because I he was busy and wanted the two young fellows occupied.
I am not against a capitol offence, but I am not sure that doing an experiment too soon, or willy nilly, qualifies.
Unfortunately, that isn't actually the case. The dominant harmonic is, surprise surprise, the third.An example is requested, which is when the Adcom came in - as an amp with audible HOTHD but no lower order to mask it. But the *original* assertion is the harshness of unmasked higher-order harmonics.
Feedback only increases higher order distortion if there’s not enough. Read the F-word by B Putzey. Most solid state amps have more than enough reducing all distortion to much lower levels than a SET (which I consider an effects box, a distortion generator.) Let me guess, you sell tube amps.I forgot to mention that the ear has a masking principle that is at play when dealing with higher ordered harmonics. Its well known that the 2nd and 3rd harmonics can mask the presence of the higher orders if in sufficient quantity (this is why SETs sound so smooth despite having more higher ordered harmonics than any solid state amp). In the case of a 'properly designed' SET (which will be perceived as 'musical' sans feedback) it is imperative that the higher ordered harmonics fall off on an exponential curve. In this fashion they are masked.
There are two other inconvenient facts regarding the ear. The first is that it uses higher ordered harmonics to sense sound pressure (easily proven with simple test equipment). The second is that it assigns tonality to any form of distortion (harmonics, IMD or aliasing). Higher ordered harmonics if unmasked are assigned 'harsh and bright'; that latter bit is the part we've known since the 1930s.
Quite often in solid state amps which employ feedback (most SETs do not), the feedback is fairly successful at suppressing the innate distortion of the amplifier. But due to several problems with the application of the feedback, its application results in generation of higher ordered harmonics. This problem has been around for a very long time (see the writings of Norman Crowhurst back in the late 1950s). It comes from 3 different issues:
1) the feedback node isn't linear and so the correction signal becomes distorted before it can do its job.
2) Gain Bandwidth Product in the design might be insufficient to support the feedback at all frequencies in the audio band. This problem results in increasing distortion with frequency since at the turnover point (usually above 1KHz) the feedback decreases and so distortion climbs.
3) the phase margin of the amplifier might be insufficient to allow for higher amounts of feedback without stability issues. Phase margin is that value where at some frequency, the feedback has so much phase shift that it is positive rather than negative. Positive feedback results in oscillation so usually this is an upper limit of sorts. Of course if the feedback loop is properly designed this problem can by mitigated quite a lot.
So in such an amp with the problems listed above (which is the majority of solid state amps make since the inception of the transistor) there will be a perception of 'harsh and bright' at some volume setting (and this is why tube amplifiers are still made; you don't have to know anything technical to understand this fact; its economics). You can mitigate this to some degree by reducing treble output in the speaker but since the problem is caused by distortion rather than a frequency response error this will be found to be an inadequate solution.
Self oscillating class D amps do provide a solution to this rather classic problem. In a self oscillating amplifier, so much feedback is applied that the amp goes into oscillation as soon as its powered up. The feedback loop is then designed so that the amp can only find one oscillation frequency; that is used as the switching frequency. Finally, the feedback is mixed with the incoming signal outside of the active signal path (in the same way its done in opamps) so the correction signal has far less distortion when it arrives to mix with the incoming signal to do its job. This all means you can have much more feedback than was previously practical, without Gain Bandwidth Product problems and using a more accurate feedback signal. The Purifi module is currently the best example of this approach.
In the Adcom? So his example doesn’t fit his hypothesis?Unfortunately, that isn't actually the case. The dominant harmonic is, surprise surprise, the third.
Yes.In the Adcom?
Even better some of the Stereophile writers back in 1989 mentioned how very well the Adcom was able to portray the Cowboy Junkies Trinity Sessions.
Well, I mean... there you go, then!Even better some of the Stereophile writers back in 1989 mentioned how very well the Adcom was able to portray the Cowboy Junkies Trinity Sessions.
Well, I mean... there you go, then!
Imagine their reaction if they'd have tried Jazz at the Pawnshop?
Hope so; it a great education on how insensitive human ears are to small amounts of realistic mixes of distortion.So where is atmasphere all of a sudden? Maybe he is playing with Distort. Would be good if he is I suppose.
Yup but really boring music.An absolute stunning recording.