Yeah, but I hate it to see BS being spread. Especially when it is totally unnecessary because we know the stuff since 100 years or so 🙄
Jan
Jan
I recall Thimios doing distortion tests with and without actual speakers attached on an amp. Distortion measurements actually improved with the speaker attached. I can't find the post right now but if I remember correctly it was Ostripper's Spooky input stage (Leach amp) coupled to Vzaichenko's non-switch output stage.
Last edited:
I found the post. Vzachenko's AX Deluxe input stage with his NS-OPS output stage. Revisiting some "old" ideas from 1970's - IPS, OPS
Totaly wrong. You have no idea how feedback works. The feedback does not add signal to the input. It compares the 2 signals (in and out) and corrects the output to match the input. So in your example the amp will keep the output at zero by absorbing the speaker energy, damping it.
you have also missed the point that the ltp feedback is also connected to a very low impedance output transistor net work that will short out signals coming back from speakers.
Because loudspeaker distortion creates higher order distortion products from distortion produced by the amplifier. A second order distortion product from the amplifier gets transformed into a fourth order distortion product by the second order distortion of the loudspeaker.
Higher order distortion products are more audible than lower order distortion products. Hence, the more upstream in the audio chain, the more important low distortion is.
Hey, don`t take my post #1647 too seriously.
More technically speaking, negative feedback has negligible impact on the cone behavior at low frequencies and motional feedback is extremely tricky to implement properly. Negative feedback makes a mess of phase relationships and should be avoided in audio amplifiers. Loudspeakers should be controlled by the output stage alone.
sser2,
insightfully put, except for the fact that motional feedback is indeed highly efficient in reducing cone resonances at low (sub-bass) frequencies. It is not just a matter of diaphragm rigidity. You should urgently revise your current stance.
More technically speaking, negative feedback has negligible impact on the cone behavior at low frequencies and motional feedback is extremely tricky to implement properly. Negative feedback makes a mess of phase relationships and should be avoided in audio amplifiers. Loudspeakers should be controlled by the output stage alone.
sser2,
insightfully put, except for the fact that motional feedback is indeed highly efficient in reducing cone resonances at low (sub-bass) frequencies. It is not just a matter of diaphragm rigidity. You should urgently revise your current stance.
No, not at all. This is a great example of the fallacy of ignoring that fact that speakers act in 3 spatial dimensions. If you rapped on a speaker cone and the cone was absolutely rigid, moved all as one surface, no flex no wobble, just movement in the same axis as you rapped it. And the speaker did this over the entire hearing bandwidth.... Which of course brings us to, well, ...there's nothing even remotely close to that. Then maybe the rap sound would be silenced.
I'm not sure of the exact conditions in the chart, but as you said a few pages back 😉 it's 2020. We have very capable DSP available, particularly for the DIYer, so frequency response aberrations caused by output impedance are easily solved. In relation to this thread's discussion of NFB and fidelity, a high fidelity system should already be using DSP to fix the room at least below 200hz, so adding it across the spectrum should not be a reach. On the other hand, DSP is not really up to the task of nulling non-linear speaker distortions, while current drive (i.e. high output impedance) has been shown to do so above LF. Of course, high output impedance does not necessarily mean low GNFB, and it also increases susceptibility to back EMF and RFI. But I don't think the low GNFB crowd is necessarily blind, or enjoying distortion (even if they think they are), or whatever else. They might be enjoying the reduced speaker level distortions coming with a higher output impedance, and even at LF the realized frequency response of varying damping factors is room dependent. Current drive also reduces frequency response variation over temperature, with the downside of increasing the chances of driver death.
I'm not defending those who eschew measurements as incapable of capturing sonics. Instead I'm saying the most prominent headline measurements only capture a portion of the story for the accuracy of the signal at the listener's ear. And the word accuracy begs the question of what the reference is (I can't imagine Bob Pollard and co were listening back on Revels when they recorded Bee Thousand...), but that's a topic for another place...
As to another comment I saw in the thread re: low order amp harmonics creating higher harmonics and IMD once reproduced by the speakers: I used to worry about this but I really think it is many levels down the list of concerns for two reasons: Speakers at low levels exhibit low levels of non-linear distortion, so a -80dB second harmonic is unlikely to cause significant artifacts; and the distortions caused by the signal itself will always swamp the distortions of distortions so signal related distortions are the first disease to cure--a 1% distortion of signal is at -40dB, while a 1% distortion of a .1% amp distortion is at -100dB, and in reality the distorted portion of the signal will be distorted less by the speaker than the pure signal portion due to its lower level.
The one caveat I'll make is I'm only comparing amp distortions that are fairly close to speaker distortions, like general THD and IMD. Crossover distortion, "TIM", etc. are in a different category, audibly vulgar, and NFB of all sorts tends to improve them.
THD on its own is not particularly useful, although a good way to ensure repeatability during manufacture.
More useful is the amplitude of the first six or seven harmonics. If you can achieve a monotonic, decreasing level of the harmonics, regardless of highish THD the amp generally will sound good.
Global feedback does influence the profile of the harmonics.
This thread has lasted and lasted, it really has me intrigued........
HD
More useful is the amplitude of the first six or seven harmonics. If you can achieve a monotonic, decreasing level of the harmonics, regardless of highish THD the amp generally will sound good.
Global feedback does influence the profile of the harmonics.
This thread has lasted and lasted, it really has me intrigued........
HD
Same old same old regurgitated over and over, the audio equivalent of acid reflux, I've not read anything new here yet, have you? I wonder why......There are many people afraid of feedback because the don't understand it, that's not going to change.😉
This blanket statement only reinforces mythical and magical thinking. Any kind of harmonic distortion will result in intermodulation with signals that are not just pure sine waves. As such, a high THD amp can sound like crap despite it having audiophile-approved "decreasing levels of harmonics." As for "sounding good", the point of high-fidelity is transparent reproduction. The amp is not supposed to add audible distortion, which however is not that hard anyway since speakers are the weakest link and produce distortion that is magnitudes higher.
That's a point often ignored. Some second harmonic can sound euphonic but only with simple music, girl and guitar for example 😉
Yes but then you would have to do that for each and every speaker type on the market you want to use with that amp, for they all have their own curve. Not practical.
The very purpose of trying to make an amp as an ideal voltage source, Zout = 0, is that it works with any speaker because speakers are designed to be driven from a voltage source. (yes I know there are exceptions).
So of course when you design an amp with appreciable Zout, be it by low or no feedback or just plain bad design, almost every speaker will sound different with it.
The scenario mostly goes like, you switch in a high Zout amp, the sound is quite different, so it must be better.
Jan