I am happy that when I inserted the link, 2nds decreased while 3rds and 4ths remain unchanged.
I've heard all sorts of reasons as to why this could not have been the case. I have postulated that there is a good reason for it. Seems NP has observed something similar and has a viable theory about it as well.
Theory says if you load the VAS distortion must go up. However, there has been little published work on how the various harmonics that go to mske up THD under dynamic conditions behave. In symmetrical amplifiers, 2 nds are supposed to cancel. But in practice, 2 nds on a symmetrical amplifier are never zero - lower than 3 rds, but not zero.
It is my contention that the relationships between the harmonics (I.e dBr) are not fixed but change under dynamic conditions.
Take a look at the Bowes test for a hint as to what might be going on.
I've heard all sorts of reasons as to why this could not have been the case. I have postulated that there is a good reason for it. Seems NP has observed something similar and has a viable theory about it as well.
Theory says if you load the VAS distortion must go up. However, there has been little published work on how the various harmonics that go to mske up THD under dynamic conditions behave. In symmetrical amplifiers, 2 nds are supposed to cancel. But in practice, 2 nds on a symmetrical amplifier are never zero - lower than 3 rds, but not zero.
It is my contention that the relationships between the harmonics (I.e dBr) are not fixed but change under dynamic conditions.
Take a look at the Bowes test for a hint as to what might be going on.
Last edited:
Regarding: Here's another analogy to pick holes in: when I was buying a TV recently, in a group of the same technology type and resolution etc there were some models that looked much more realistic to me,...?
Not surprising as you are not comparing fairly by any means. All those TV's were not adjusted to their optimum settings. Some were probably not even close, so comparing actual quality is pretty impossible.
Regards,
Greg
Not surprising as you are not comparing fairly by any means. All those TV's were not adjusted to their optimum settings. Some were probably not even close, so comparing actual quality is pretty impossible.
Regards,
Greg
Member
Joined 2009
Paid Member
I've seen this in simulations often enough - the profile is a function of output power even into a pure resistive load at the amplifier output. I noticed this whilst looking at the relationship between 2nd dominant verses 3rd dominant behaviour. Even the same topology can behave differently depending on bias levels. Of course simulations are simulations.
Its quite common for multiple distortion sources to cause cancellation of a distortion product. A common difficulty when trying to measure passive IM generation in RF componentssuch as plated connectors or attenuators. In microwave amps it is even a design technique to cancel amplifier distortions, google "predistortion".
For 2nd, if one has the capability of inverting the polarity, the effect of test equipment residual cancellations can be dissected somewhat.
Second harmonic should be the least of anyone's worries, since it takes so much to be even noticed, and then it just changes the sound somewhat. However, like everyone else, I can get into 'specs', and it is usually relatively easy to cancel out second harmonic distortion, so we do it.
After reading Cheever, I think he is on to something important. Not the ONLY factor that changes the sound quality, but his reasoning why open loop tube circuits can sound so good, even when they measure 'badly' is a breakthrough, as far as I am concerned. Still, global negative feedback appears to have something else about it that makes it sound 'off', and there is still a lot to be learned as to why something that MEASURES virtually perfectly, still has a signature sound. I am now tempted to make a zero global feedback amp and compare it to my feedback amps that are comparable to many here that are discussed, and compare its 'signature' against my best global feedback efforts.
After reading Cheever, I think he is on to something important. Not the ONLY factor that changes the sound quality, but his reasoning why open loop tube circuits can sound so good, even when they measure 'badly' is a breakthrough, as far as I am concerned. Still, global negative feedback appears to have something else about it that makes it sound 'off', and there is still a lot to be learned as to why something that MEASURES virtually perfectly, still has a signature sound. I am now tempted to make a zero global feedback amp and compare it to my feedback amps that are comparable to many here that are discussed, and compare its 'signature' against my best global feedback efforts.
Regarding "global" vs. "local" feedback: Some time back, in the Blowtorch thread, Jan D. brought Bruno P. into the discussion to clear up some confusion, especially on my part, about the difference between global and local. We know that local feedback, like emitter degeneration, emitter followers, split-load phase inverters, and anode followers, doesn't exhibit the Baxandall/Putzeys harmonic multiplication effect. But why not?
Bruno P. mentioned that *all* circuits and all stages have a high frequency rolloff, and I jumped on that like a dog on a chicken fried steak with cream gravy. I concluded that the difference must lie in the harmonic spectrum; above the high frequency pole distortion components have a 20dB/decade rising response, and local feedback means a higher pole frequency.
On later reflection I see that this can't be right, because harmonic multiplication is independent of frequency. But rereading the "F-word" article, I think I see a better explanation. He shows the mechanism of SID as being an error *outside* of the feedback loop. Bingo! "Local" for feedback means to be so intrinsic to the device or stage that it operates differently than "global" feedback.
For your consideration and correction,
Chris
Bruno P. mentioned that *all* circuits and all stages have a high frequency rolloff, and I jumped on that like a dog on a chicken fried steak with cream gravy. I concluded that the difference must lie in the harmonic spectrum; above the high frequency pole distortion components have a 20dB/decade rising response, and local feedback means a higher pole frequency.
On later reflection I see that this can't be right, because harmonic multiplication is independent of frequency. But rereading the "F-word" article, I think I see a better explanation. He shows the mechanism of SID as being an error *outside* of the feedback loop. Bingo! "Local" for feedback means to be so intrinsic to the device or stage that it operates differently than "global" feedback.
For your consideration and correction,
Chris
No feedback "sound better" because an undistorted sound is lifeless, ask the musiscians here and there...
Also contrary to what is assumed by many people relative phase has an importance, a speaker that is reverse connected, assuming the rest of the chain is not inverting the signal, will make the music sound dynamicless..
Actualy this problem extend to the whole reproduction chain, some recordings are forcibly not adequate and should be phase inverted.
Also contrary to what is assumed by many people relative phase has an importance, a speaker that is reverse connected, assuming the rest of the chain is not inverting the signal, will make the music sound dynamicless..
Actualy this problem extend to the whole reproduction chain, some recordings are forcibly not adequate and should be phase inverted.
Chris, harmonic distortion is just a way to quantify nonlinearity. The actual effect of the nonlinearities includes intermodulations which happen all over the spectrum. Lowpassing a signal may affect a hd measurement by hiding harmonics, but not much help against the signal impairment the nonlinearity causes.
My point being, limited bandwidths don't have a huge beneficial effect on nonlinearities.
My point being, limited bandwidths don't have a huge beneficial effect on nonlinearities.
Last edited:
Greetings to PDX! Thanks for your comments. I think what struck me about the high frequency pole was the idea that every stage has one. Conventional modern amplfiers have a pole way down in the audio range, or even below, and for nominally good reasons. This means that distortion components have a rising frequency response with feedback. But local feedback often allows the pole to be near or above audio frequencies, so the distortion spectrum is not "differentiated".
But that turns out to not be related to distortion multiplication. With my pea brain, it just seemed to be.
Much thanks,
Chris
All power to your efforts re open loop linear John Curl. The best sounding headphone amp I ever built was an open loop tube circuit, CC followed by CF. The sonic signature of Papa Nelsons amps is clearly due to his choice of fets, as you have made this choice done in your own preamps.
I am not familiar with the Putzey harmonic multiplication effect, but if
you are referring to Baxandall's famous curve (attached) then you will
find that local feedback (including degeneration) creates the same
phenomenon.
Attachments
Yes, and in the famous Boyk and Sussman paper about IM distortion, all the examples are local feedback.
As indicative of a nonlinearity, something that produces second will produce sum and difference components from two tones. These energies may fall in regions of high aural acuity, and with equal temperament music may be unrelated harmonically, and thus on both counts be more noticeable.
This is why Putzeys considers second objectionable I believe. Clearly not a big deal for harmonic distortion, but significant for IM.
bcarso, precisely. You will only see harmonic distortion only with a single sinusoid input. As soon as you add a second tone you will get intermodulation products.
Now take a look at the spectrum of actual music...
In reproduction of music any significant harmonic distortion product sounds bad, regardless if it is 2nd or 3rd order.
Now take a look at the spectrum of actual music...
In reproduction of music any significant harmonic distortion product sounds bad, regardless if it is 2nd or 3rd order.
Last edited:
Chris, I think I follow where you were going now. And I think it makes sense of a kind. Not in higher product generation maybe, but it seems it would limit the ability of the feedback to reduce upper frequency products (as they don't get fed back above the bandwidth, particularly if the open loop bandwidth is much below the audio bandwidth).
There s tons of odd harmonics in almost any instrument sound, without thoses harmonics the sound is dull even if there s a lot of even harmonics.
Your point? No argument about instruments here. The issue is faithful reproduction.
My point is that the level of distorsion that are discussed as being sound quality influencial do not make sense at all.
Speakers distorsions is orders of magnitude higher, whoever pretend that he can hear say 0.1% distorsion from an amplifier with a musical signal is just talking non sense as this would mean that his ears automaticaly discriminate the amp distorsion from the speaker distorsion.
Faithfull reproduction doesnt need ppm level of distorsion but a good pair of speaker, and as long as this element is far below amps in matter of linearity amplifiers distorsion audibility will be a never ending and inconclusive debate..
That amplifiers with varying caracteristics can be equally good is an indication that the difference lies elsewhere than in the amplifier, and that s why people keep asking themselves how this "perceived" difference can be accurately measured, this would require to measure the relevant parameters, and as said i doubt that they are to be found in the amplifier.
Last edited: