Technically incorrect. This assumes that the electric signal translates linearly into sound waves, but this is not the case. Consider a diaphragm that bends and flexes during acceleration. Consider mechanical intermodulation distortion due to a single driver reproducing both low and high frequencies at the same time. There are many mechanical sources for distortion that will not directly translate back into an electric signal.
All of this demonstrates a gross misunderstanding of how any of this works. NFB in the amp never cancels mechanical vibrations.
To the speaker, the amplifier looks like short / a piece of wire. No cancellation is going on, only damping. That is, the speaker is not only damping its own vibrations mechanically but also electromagnetically.
What you seem to think is that the amp knows position/velocity/acceleration of the voice coil and then injects an inverse signal to cancel all of that and push the voil back to 0. This is not what happens.
Some active speakers do that with DSPs that have feedback loops that include the speakers.
Hm. We might want to keep in mind that Bruno's paper is ancillary to his building self-oscillating class D amplifiers that have a lot more feedback than most traditional amps- in excess of 40dB!
For those who know how valuable Gain Bandwidth Product is to an amplifier design, this is quite an achievement. But Bruno is also adamant (and rightly so) that most solid state (and all tube amps) lack sufficient feedback. Because this is so, while those amps (IOW most amps made in the last 50 years) have what looks like 'low' distortion, the simple fact is its not low enough.
The ear interprets all forms of distortion as a tonality- the 2nd is responsible for that thing audiophiles call 'warmth'; the higher orders are well-known to contribute to brightness and harshness. There is an AES paper about something called the 'Ged-Lee Metric' that presents a nice weighting system on the higher ordered harmonics. Worth a read.
Bruno points out that most amps simply don't have enough feedback; that there is a 'hump' of sorts where insufficient feedback is audibly harmful. Its pretty fair to say this has fueled the tubes vs. transistors debate for decades.
But if you can build an amp with really high feedback, now the feedback can compensate for the distortion it introduces; it can also compensate for phase shift even if the bandwidth of the amplifier isn't all that wide.
All I'm saying here is read Bruno's papers and you will see that he's not a fan of insufficient feedback (less than about 35dB). Such an amp is often bright due to higher ordered harmonics, so if you wonder how an amplifier's distortion can be audible when a speaker has so much more distortion, its because the ear interprets distortion as tonality. The ear is so sensitive to the higher orders because it uses them to sense sound pressure- and so has to have a pretty wide range. This is an evolutionary aspect no doubt, as sine waves are pretty rare in nature.
There aren't all that many designs that feature distortion as low as Bruno has been able to achieve. One of the very few that comes to mind is the Benchmark amp; notable IMO as it isn't class D. I think there were a few others but for some reason never got that successful.
Last edited:
Can't say I disagree. I do remember that Bruno, several years ago when challenged, modified a tube amp for 60dB feedback in the audio band. Said he, tongue in cheek: "It even sounded good".
I think Bruno's latest Purifi amps, which challenge even the best class A amps on transparency, prove the statement of this thread.
Jan
I think Bruno's latest Purifi amps, which challenge even the best class A amps on transparency, prove the statement of this thread.
Jan
Earl has discussed this on length on threads here AND the papers are on his website. But he calls it GedLee. Virtins multi-instrument will even calculate it. But the findings upset those who feel capacitors lift veils 🙂.
Yes. I think he also mentioned the difficulties he had with the phase margins in that tube amp...
Distortion has been low enough for a long time, but this doesn't stop audiophiles from debating about magic.
The mentioned "Ged-Lee metric" shows that you can have significant distortion including many high-order products e.g. as produced by a class-B amp and it's still not audible according to Lee & Geddes.
The "warmth" and "harshness" terminology is borrowed from instrument amps that add tons of audible distortion, color the sound deliberately, and should be considered part of an instrument's sound. But that has nothing to do with music reproduction, let alone high-fidelity.
Generally speaking, when your amp produces so much distortion that it becomes audible then it is low-fi. We also call it a nonlinear effects box, or audiophile-grade. 😛
This statement is problematic- a lot depends on what is meant by 'a long time'. If you mean 'decades' then the statement is simply false for all practical purposes. If you mean for maybe 10 or 15 years then a qualified 'maybe so', but only because certain amplifier examples existed, which is still the problem today- many solid state amps made today simply have too much distortion (although on paper it 'looks' low)- and so that distortion is audible as brightness, and since brightness sucks audiophiles wind up talking about 'magic' (which might mean 'just as much bandwidth but without the irritating brightness').
Bruno mentions this bit in particular- that audiophiles didn't like the brightness of solid state amps for a simple reason- manufacturers resorted to a simple technique called 'lying'; and sold amps with gain bandwidth product far too low to support enough feedback- as a result of poor feedback at high frequencies, brightness occurs because that's how human ears perceive higher ordered harmonics. We could all hear it; thus the tubes/transistors debate. But once you have enough feedback the tubes/transistor thing goes away.
But if we were to count the amps today that have enough feedback, it would not be a very high number. Bruno's designs would be a majority in that set...
Last edited:
Not in particular.
Its really well-known (or should be) that the ear is far more sensitive to higher ordered harmonics than it is to fundamental tones. This is really easy to demonstrate with simple test equipment.
That's been the Achilles' Heel of feedback for a very long time- since it was always never enough feedback, a result (while overall distortion was indeed lower) was higher ordered harmonic distortion. It was creating the thing to which the ear was most sensitive. So we have decades worth of bright amplifier designs out there... hence tubes are still around.
Then I'm afraid I don't understand what you mean by "The ear is so sensitive to the higher orders because it uses them to sense sound pressure"
You may not, but that won't change things. That's exactly what goes on.
If you don't get this, you can set up a simple test to demonstrate it for yourself. You'll need a sine/square wave generator, an amp, a speaker and a VU meter. Play the sine wave through the setup and set the level so the VU reads 0VU.
Now cover the meter, and turn down the volume all the way. Switch to square wave. Turn up the volume until it sounds as loud as before. Uncover the meter- what do you see?
atmasphere:
Hearing your opinion on the "high feedback" factor, where can we place your OTL amps?
Do they also feature over 40 dB of GNFB?
Hearing your opinion on the "high feedback" factor, where can we place your OTL amps?
Do they also feature over 40 dB of GNFB?
No- but our class D project does.
I realized the phase margins of our OTLs would not easily support that kind of feedback, even though we've got plenty of bandwidth. So rather than audibly screw up the sound, we ran them without feedback, and relied on other means to minimize distortion- bias points, topography and the like. This resulted in a high output impedance of course, but by understanding which speakers were going to work and which ones were not we were able to avoid tonal coloration (and FWIW, the ear has a tipping point where tonality deriving from distortion can override tonality from FR, plus no speaker is truly flat so this has worked out fairly well over the last 45 years). Since our amps express mostly a cubic non-linearity the distortion level is significantly lower than what you get with something like an SET.
In that way we avoided the 'bad area' of feedback- between 4dB up to about 35dB.
With high output impedance you will easily get peaks in the FR that are a couple of dB high with most speakers. This is plainly audible, like playing around with an EQ... except that the EQ is fixed with high output impedance.
In order to "override" this with distortion (which btw you can't) you'd need quite a bad amplifier with lots of distortion.
Also, how does it make it sense to talk of a "bad area" of feedback when the effects depend on the linearity of the design?!
In order to "override" this with distortion (which btw you can't) you'd need quite a bad amplifier with lots of distortion.
Also, how does it make it sense to talk of a "bad area" of feedback when the effects depend on the linearity of the design?!
If this is meant in response to my last post it looks to me as if that post went unread as well as Bruno’s papers central to this thread.