"Our" meaning the amp marketing department?
That's one of those marketing tactics when you look through audio electronics mag or retail websites. Yeah, even though the frequency extension is beyond our hearing and the distortion level is way below the audible threshold, they look good in the advertisement. Plus, there are people who fall for those, thus the business lives on.
An amp built to those standards would be capable of doing a reasonable job of reproducing sound, better than some modern fashion systems. You need to be aware that they were not saying "do this and it will be hi-fi" but rather "this is what we think you need to do as a minimum". The definition of hi-fi has not changed; the requirements we think you need to achieve it have been tightened up a little. Hi-fi is not supposed to sound amazing; it is not supposed to sound at all.kylej1050 said:
Which are audible is objective and lots of studies have been published on this subject.
Which are objectionable or not completely depends on taste. Modern dance music styles often use very high amounts of harmonic distortion combined with samplerate reduction and aliasing to create the sounds used, often in used in multiple layers and mixed together. If your adventurous try some raw style video's on utube for examples.
I've never encountered this in classical recordings. But even there "processing" (in the form of parallel compression, which adds harmonic distortion) is the norm. Don't forget the huge amounts of edits, more than 1000 edits is not uncommon. Although its not actual distortion in the technical sense. Life events of these artists can be "surprising".
In fact I don't know any recording without added linear or non linear distortion. Mics being the main source of distortion even in "clean" recordings.
We better face it, distortion is great. But for some not in the reproduction chain.
Yes, low even order has minimum shape damaging effect.
At least the 2nd, 10% not a big deal. I posted software http://www.diyaudio.com/forums/software-tools/296814-harmonics-tone-generator.html .
http://www.gedlee.com/downloads/The Perception of Distortion.pdf
Look what I found!
Pretty much proves lower order THD (and therefore IMD) is much less audible than higher order nonlinearities.
It's also very enlightening that nonlinear distortion is most audible at low sound intensity! So if an amplifier has monotonic distortion and plays clean at lower power levels with simple harmonic products.. very likely it will be audibly transparent.
Sounds like a class A SET Amp 😎
I think the most interesting thing though, is that diffraction distortion are highly audible!
Look what I found!
Pretty much proves lower order THD (and therefore IMD) is much less audible than higher order nonlinearities.
It's also very enlightening that nonlinear distortion is most audible at low sound intensity! So if an amplifier has monotonic distortion and plays clean at lower power levels with simple harmonic products.. very likely it will be audibly transparent.
Sounds like a class A SET Amp 😎
I think the most interesting thing though, is that diffraction distortion are highly audible!
Any bend in the transfer function (even if it is a bend that produces pure 2nd harmonic) is going to produce IM products when you mix tones. You may not notice much (and it may even sound subjectively better) with simple music where not a lot is going on, but the IM products will bet there whenever two or more tones mix, and they will not be harmonically related to the tones that are doing the mixing.
I built an open-loop push-pull amp that had an AC balance adjustment in the phase splitter and I played around with it a bit. Subjectively, what I noticed was that when I had a lot (~0.2% @ 1W) 2nd harmonic, "S" sounds in the vocals didn't sound quite right with really complex music. I'm talking about modern electronic music. Nulling out the 2nd harmonic distortion (down to 0.02% @ 1W) made the "S" sounds much cleaner in those songs.
I built an open-loop push-pull amp that had an AC balance adjustment in the phase splitter and I played around with it a bit. Subjectively, what I noticed was that when I had a lot (~0.2% @ 1W) 2nd harmonic, "S" sounds in the vocals didn't sound quite right with really complex music. I'm talking about modern electronic music. Nulling out the 2nd harmonic distortion (down to 0.02% @ 1W) made the "S" sounds much cleaner in those songs.
Last edited:
Indeed this is the likely cause with a speaker that sounds bad when you turn it up.
It makes sense in this context. Crossover distortion was identified as a greater problem.Wavebourn said:
One goes with the other.SpreadSpectrum said:
I wonder if any one mentioned that (almost) all of those who test harmonic
and intermodulation distortion are using an 8 Ohm Resistor (or 4 Ohms or 16 Ohms,
but they are using a Resistor).
Most loudspeakers are not 8 Ohms at all frequencies.
Even worse, they are Reactive at most frequencies.
That means that harmonic and intermodulation distortion
can vary widely versus frequency (including the strength
of the harmonics relative to each other, i.e. falloff rate).
This problem exists, whether negative feedback is used,
or whether no negative feedback is employed.
It is a problem whether the amp is single ended or push pull.
Take a triode curve and put a Resistive load line on it,
from 0 volts on the grid to 2X the grid bias volts.
Now, draw an ellipse that encloses the resistive load line,
over those same end points. The distortion is different as the signal voltage
increases (more toward 0 Volts on the grid) than it is when the signal
voltage decreases (more toward 2X bias on the grid).
And consider that the eclipse is different at different frequencies,
similar to different load Resistors which have different load lines.
The loudspeaker is resistive, capacitive, inductive, capacitive plus resistive,
and inductive plus resistive, depending on frequency.
The eclipse will have different slopes, and different widths (spread wider
or narrower).
Now you have very complex factors that affects harmonic and intermodulation
distortion versus frequency.
How about some harmonic and intermodulation testing that use real loudspeakers
as the load?
You need to have an anechoic chamber (your ears), a bursted signal source (to prevent
speaker burnout), and an analyzer that can work with bursted signals.
Steady state signals at medium and high power levels will burn out tweeters and midrange loudspeakers.
and intermodulation distortion are using an 8 Ohm Resistor (or 4 Ohms or 16 Ohms,
but they are using a Resistor).
Most loudspeakers are not 8 Ohms at all frequencies.
Even worse, they are Reactive at most frequencies.
That means that harmonic and intermodulation distortion
can vary widely versus frequency (including the strength
of the harmonics relative to each other, i.e. falloff rate).
This problem exists, whether negative feedback is used,
or whether no negative feedback is employed.
It is a problem whether the amp is single ended or push pull.
Take a triode curve and put a Resistive load line on it,
from 0 volts on the grid to 2X the grid bias volts.
Now, draw an ellipse that encloses the resistive load line,
over those same end points. The distortion is different as the signal voltage
increases (more toward 0 Volts on the grid) than it is when the signal
voltage decreases (more toward 2X bias on the grid).
And consider that the eclipse is different at different frequencies,
similar to different load Resistors which have different load lines.
The loudspeaker is resistive, capacitive, inductive, capacitive plus resistive,
and inductive plus resistive, depending on frequency.
The eclipse will have different slopes, and different widths (spread wider
or narrower).
Now you have very complex factors that affects harmonic and intermodulation
distortion versus frequency.
How about some harmonic and intermodulation testing that use real loudspeakers
as the load?
You need to have an anechoic chamber (your ears), a bursted signal source (to prevent
speaker burnout), and an analyzer that can work with bursted signals.
Steady state signals at medium and high power levels will burn out tweeters and midrange loudspeakers.
In my previous post, I should have said 'a Family of Triode Curves',
not 'aTriode Curve'.
Now if you are thinking of testing distortion of the amp when loaded
with a loudspeaker, what model loudspeaker do you use?
Just some food for thought.
not 'aTriode Curve'.
Now if you are thinking of testing distortion of the amp when loaded
with a loudspeaker, what model loudspeaker do you use?
Just some food for thought.
The only reason a resistor is nice is because everyone has one. You can easily compare one amplifier to another using a standard source and someone else can test it with another resistor and likely get the same result.
When you start testing with a loudspeaker pretty much any and all comparative ability goes out the window. Everyone has different speakers and even crossover networks vary between speakers of the same make and model between years. Now, if you could build a speaker simulator that has some inductive, capacitive, and resistive value that others could recreate you'd be on to something.
When you start testing with a loudspeaker pretty much any and all comparative ability goes out the window. Everyone has different speakers and even crossover networks vary between speakers of the same make and model between years. Now, if you could build a speaker simulator that has some inductive, capacitive, and resistive value that others could recreate you'd be on to something.
A load equivalent with different resistance values is more than enough when you know what you are doing. However, for learning purposes why not to observe results loading an amp on different loudspeakers?
Problem, yes or no?
The distortion of an amplifier is generally dependent on the load.
This thread was regarding distortion versus frequencies
(i.e. as shown in the curves in this thread). These curves were done with a resistor load.
Many other factors other than the load are responsible for the increasing distortion at very low frequencies, and at very high frequencies. But the load still does affect the distortion.
One factor that often is forgotten is the loudspeaker that is connected to the amp.
Most loudspeakers present a different load versus the signal frequency.
It can be an ecliptic load line, or a straight load line, depending on frequency.
The distortion of an amplifier versus frequency can be shaped to a greater or lesser degree, depending on the amplifier and the load resistor or loudspeaker load.
An amplifier and loudspeaker need to be looked at as a system, not just individual
components.
No, testing this way is not easy, but failing to test something does not change the results
you get.
Perhaps I am barking up the wrong tree, and this is only a very small inner layer that is very near the center of the onion. But it is a layer of the onion.
The distortion of an amplifier is generally dependent on the load.
This thread was regarding distortion versus frequencies
(i.e. as shown in the curves in this thread). These curves were done with a resistor load.
Many other factors other than the load are responsible for the increasing distortion at very low frequencies, and at very high frequencies. But the load still does affect the distortion.
One factor that often is forgotten is the loudspeaker that is connected to the amp.
Most loudspeakers present a different load versus the signal frequency.
It can be an ecliptic load line, or a straight load line, depending on frequency.
The distortion of an amplifier versus frequency can be shaped to a greater or lesser degree, depending on the amplifier and the load resistor or loudspeaker load.
An amplifier and loudspeaker need to be looked at as a system, not just individual
components.
No, testing this way is not easy, but failing to test something does not change the results
you get.
Perhaps I am barking up the wrong tree, and this is only a very small inner layer that is very near the center of the onion. But it is a layer of the onion.
This is easier to do than you think with modern test tools to capture the amplifier signal when connected to a real speaker.
Well designed amplifiers keep a straight signal no matter if the load is inductive or resistive by keeping a low impedance output at all frequencies.
Actually I have some amplifiers which have lower distortion and less inter-modulation under a real speaker load!
Well designed amplifiers keep a straight signal no matter if the load is inductive or resistive by keeping a low impedance output at all frequencies.
Actually I have some amplifiers which have lower distortion and less inter-modulation under a real speaker load!
Last edited by a moderator:
If you want to see how much or little amp distorts at different load and frequency, you test it at different load and frequency. Why do you call that a problem?
Please tell me how you tried it and found out that it's not easy.
Who failed? Names please.
True, but it can be straightened out. It's debatable whether this should be necessary but why not? SET amps might need it more, but by conjugating the impedance I won't allow it to be an excuse not to build one.
Woofer resonance is a little more difficult to deal with (not much) but either way, damping can be shared with the cabinet (volume). It's good knowing the system as a whole. Does it become indivisible this way? Not necessarily, these are simple considerations, I think.