# When low HD is low enough?

#### flex2

So your perceptions contradict the facts. Then you have to make a choice; I don't have to spell it out for you.
The earth is the center of the universe, everyone can clearly see that.
Those astronomers must be wrong!

Nice talking to you, all the best,

Jan

So you call me an ignorant idiot.
Expected more from a grown up.
Your opionion is by the way not facts. They are opinions.

Have a nice evening Jan, but we simply have different experiences.

#### tomchr

Paid Member
@flex2 - Nobody called you anything. All Jan did was to hold up a mirror.

Hopefully you've found time to read his article by now. I think he makes a pretty strong argument. Enough so that I thought of replying with a follow-up, until I realized that I really couldn't poke holes in his theory. It's refreshing to see someone who backs their opinion with data and established theory, in particular in a thread like this.

Tom

2 users

#### abstract

High distortion in the input stage with clean output is a good sign. It means that almost all amplifier distortion is removed from the output.
Feedback works by - ideally - only have the distortion at the input stage, in appropriate phase.

Lets run some numbers. Assume you have an amp that has a gain before feedback of 100,000 and an output of 1V.
The input to that amp is then only 10uV.
Also assume it has 1% distortion, and we close the feedback loop for a gain of 20, and an input signal of 1V/20 = 50mV.
(But remember that signal-wise, the amp itself sees still the 10uV between its inputs, or input and ground, which is the difference between the source signal and the feedback signal!).
We then 'must have' a distorted input signal of 1% of 1V /20 = 100uV at the amp terminal.
That swamps the 10uV input signal and looks like a huge input stage distortion. Which is good.

Jan
But in order to get that 100k gain, you would have to stack a few stages. The more there is going on, the more potential for "unknown unknowns" to throw surprises.
Apart from the amplifier + speaker interactions that I mentioned earlier, which would probably be #1 to fix in terms of total distortion, a few ideas come to mind:

An LTP stage, where cancellation of even harmonics "drifts across zero". With a +/-50mV signal, which is almost all common-mode, the tail CCS would be susceptible to drift from temperature cycling, leading to drift in the total gain, so that's a possible source of imbalance. So, with a temperature sensitive CCS, the typical LTP seems like a perfect setup for even harmonics to undergo amplitude modulation between near-perfect PPM cancellation and some unbalanced higher value.

There's an irony there, too. By adding harmonics, say, with a tube pre-amp stage, total modulation seems likely to actually go down. 0.5% +/-0.05% (of H2, etc.) may be subjectively less offensive than a rather more drastic 0.0005% - 0.05% (H2 with amplitude modulation of 100:1). So, what are we actually doing here? Adding harmonic distortion, or reducing amplitude modulation?

Incidentally, I've heard that LTPs tuned by ear may actually end up being tuned far from what the measurement equipment might suggest.

#### tomchr

Paid Member
But in order to get that 100k gain, you would have to stack a few stages.
Maybe. Gains can be pretty high in the IC world. 80 dB would be a lot for a single stage, but 40-50 dB isn't out of the question, even with discrete devices.

Either way, the theory holds just as well for 40 dB gain as it does for 80 dB.

Tom

#### TheGimp

A couple of observations:

When discussing HD, the relative amplitudes of successive harmonics is as important as the THD. For example, if 2nd is below 3rd, and 5th,there is a propensity to be harsh. Monotonically reducing harmonic series tend to be discerned as more pleasant.

As far as levels that are detectable, there was a comparison between a straight wire and a OP-Amp with very low distortion, using matched levels, and played with FUBAR 2000. I don't remember who posted the files. I participated in the test and was astonished when I was able to tell the difference between the files with over 90% confidence. My hearing sucks, and I was only able to do this with an O2 headphone amp and Sennheiser HD600 headphones. I could not tell the difference with my stereo system (Klipsch Heresy 1977, Onkyo TX-NR838). My understanding was that the Op-Amp THD was well below -100dB.

YMMV.

1 user

#### Markw4

When discussing HD, the relative amplitudes of successive harmonics is as important as the THD.
Not only that IMHO, but at LF where humans tend to be sensitive to phase then the relative phase of harmonics may matter too. IOW, the actual time-domain waveform of distortion may be more of a factor at lower frequencies. Why would phase matter? Maybe because its related to crest factor. A distortion with a higher crest factor may have a higher peak SPL.

1 user

#### ginetto61

Paid Member
I think it's important to also note that what really matters is the threshold of someone being able to hear the distortion--not just being able to measure distortion, as the OP doesn't make that distinction very clear.
Hi and thanks a lot for the valuable advice I have a kind of conflict in mind
I notice that there is a need to show extremely low THD figures for the various projects
Meanwhile many comment on the fact that below a certain levels of THD further improvements will not impact the listening experience
i guess that in general low THD could be a sane goal ... but how low is not clear to me
In the case of THD, I assume that threshold of audibility numbers are found using headphones or loudspeakers playing either pure tones or perhaps tonal sequences that mimic some musical instrument tone, then they add some form of harmonic or modulation distortion on top of the generated test tone for detection of human threshold levels.In most cases, I would think, these very low level distortion levels for finding hearing thresholds are much more revealing than typical music, in which the distortion levels would need to be much, much higher for humans to detect which tone has added distortion and which does not--like Keith Howard did a couple of decades ago with FMD: Red Shift: Doppler distortion in loudspeakers. I believe Mr. Howard has a web site where he posted some apps that do the job nicely.
this is very interesting I deem headphones the most revealing transducer considering the price In some cases headphones are so much revealing that even very good speakers Problem is that they are not useful to check soundstage
But i have the feeling that they could be great tools to check the sound quality of the part of the system upstream them
My old idea is to get a nice sound at least at headphones level and then amplify it to drive speakers
To use headphones like doctors use stethoscopes
The main problem is that the signal that drives well a HP will overload most of power amps
And the selection of the HP becomes fundamental

#### uriy-ch

When you listen to music on headphones, you eliminate the influence of the room from the equation. No matter how good your amplifier and speaker system are, the room contributes to the sound you hear. The room and the speaker system are a single instrument for reproducing music. Until you take into account the influence of the room on the sound, you will not be able to get a sound better than in headphones.

The THD of the amplifier is the beginning of the beginning, if you look at good sound from the microphone in the listening area, then the THD of the amplifier is a very small part of the main problem of achieving good sound in the listening area.

1 user

this is very interesting I deem headphones the most revealing transducer considering the price In some cases headphones are so much revealing that even very good speakers Problem is that they are not useful to check soundstage
I think the value of headphones is that you're basically replacing listening room acoustics with the "over ear" or "in ear canal" acoustics. Bob Carver (of Sonic Holography fame) mentioned that this isn't actually better, but worse...

One of the downsides of very high efficiency loudspeakers (i.e., fully horn-loaded) is that, in small rooms, they are microphonic, i.e., they receive back the acoustic energy reflections (shifted in time) from the room and pass it back to the amplifier's outputs in the form of back EMF signals. What the amplifier does with this is typically a function of the amplifier's output impedance characteristic.

If you're using an amplifier with output impedance that is close to the input impedance of the loudspeakers, then you will actually get a sort of very low level reverberation effect as the amplifier reacts to these returning back-EMF energy waves that are coming back from converted in-room reflections of acoustic energy. Of course, the loudspeakers have to be extremely high efficiency, the listening room rather small, and the amplifier has to have relatively high output impedance, and the reverberation effect is down in the -60 dB range (i.e., still audible, but difficult to hear clearly). I believe this is the effect that the horn enthusiasts think is the "magic" from tube-type amplifiers with high output impedance, like class A SET amplifiers, coupled to very high efficiency loudspeakers - like fully horn loaded loudspeakers.

Of course, if your amplifier has output impedance that is 10x higher than the input impedance of the loudspeaker (such as perhaps a headphone amp, or a First Watt F1/F1J or F2/F2J amplifier), then the back EMF from the loudspeaker is effectively blocked (because you're using a transconductance amplifier--a current-source amplifier), and you don't hear the effect, just like amplifiers having extremely low output impedance.

JMTC.

Chris

Last edited:
1 user

So when you say that the soundstage is gone when using headphones--I think that Mr. Carver explained why that occurs: multiple eardrum-headphone diaphragm acoustic bounces (due to the now-closed path between the two) scramble the human hearing system's processing of acoustic images.

This is just like the effect of stereo loudspeaker soundstage collapsing the stereo image between the loudspeakers due to cross-channel effects of the two loudspeakers in-room. Sonic holography addresses this by offering an out-of-phase midrange/tweeter channel on the opposite stereo loudspeaker to cancel the effect of this cross-channel interference of stereo signals reaching both ears. Polk and Carver capitalized on this type of "soundstage correction of stereo" in the late 1970s-early 1980s, Polk with their SDA monitors, and Carver with his Sonic Holography preamplifier that simply mixes in slightly attenuated out-of-phase midrange/tweeter signals in the opposite stereo channels.

Chris

Last edited:
1 user

#### ginetto61

Paid Member
So when you say that the soundstage is gone when using headphones ....
Hi and good morning
I would like to state that my old dream is a pair of headphones that could provide sounds in front of me
Headphoines in some cases is the only way to listen to music without disturbing and be disturbed by other people
I am pretty sure that with DSP we will be able to experience a great headphone experience in the future
There are some test tracks for headphones where a barber walks around the head of the customer It is sad how i can feel the sound on a side than the sound disappears (i cannot locate its source) to appear again on the other side
The front sounds are completely not existing there is like a silent black hole just in front of me
This is completely unnatural and unbearable
As i said above i am quite sure that some kind of digital processing of the L and R signal can cure that
This would be a godgift to all the people who have to live in a flat with difficult neighbours
I would prefer a closed headphone that isolate me from the rest of the world
the are so many annoying noises

Paid Member
Try x-feed?

//

1 user

#### ginetto61

Paid Member
Hi the day that a X-feed will work it will win the videogame market
The videogames market is huge and it is pushing this kind of DSP technology
It is only a matter of time The future is 90% virtual and 10% real
Virtual vision is already at a good level
It will be exciting to watch an event without being there but with a live feeling

#### mr_zener

Why don't people see for them selves? PKHarmonic is a harmonics generator VST so allows you to add distortion as one pleases.
https://distortaudio.org/pkharmonic.html

3 users

#### tmuikku

See david griesinger Sonic Focus . Headphone sound localizing inside head is HRTF related thing I think. As per the Sonic Focus frontal localization should happen if you calibrate them for your personal hearing. Haven't tried it yet.

edit. here

1 user

#### torgeirs

@flex2 - Nobody called you anything. All Jan did was to hold up a mirror.

Hopefully you've found time to read his article by now. I think he makes a pretty strong argument. Enough so that I thought of replying with a follow-up, until I realized that I really couldn't poke holes in his theory. It's refreshing to see someone who backs their opinion with data and established theory, in particular in a thread like this.

Tom
(Very off topic) Thats funny. Your picture beside the article you are commenting

1 user

#### torgeirs

One relationship between a audio signal and HD that maybe is not accepted by everyone is that every audio signal can be made by its fundemental + the harmonics of that fundemental frequency. https://en.wikipedia.org/wiki/Fourier_series
Just the same as the HD. It is harmonics of a fundamental.
So if a signal gets distorted by HD it only adds to the excisting harmonics or adds harmonics if the original signal is band limited.
So 1 % distortion gives simplified 1 % of the fundamental as added volume to the harmonics. No wonder why it is difficult to hear in real music.

It is often said that a small bandwith peak filter with a Q of less than 0.5 dB can be difficult to hear. 0.5 dB increase is about 5% increase for comparation

Last edited:

#### torgeirs

It is also interesting that the gain curve of a circuit can ,at least picewise, be expressed as an power series.
That can be expressed as

Compare that with a0 is DC, a1 is the fundemental. a2 is the amount of second harmonics, a3 is the amount of third harmonics and so on.
Så linear circuit has only a0 and a1. Unlinear has also the harmonic components a1,a2,a3 +++

So THD measures a2, a3 +++
To measure IMD, x is replaced with two components.
If linear system the two comonents are only multiplied with a1 and summed with a0.
But unlinear circuit will at least generate a term that is the two signals multiplies x 2 + the original signals. (x+y)^2 = x^2 +2xy + y^2. The 2xy term is more easily heard as it is nonharmonic to either of the input signals. (as the input signals are nonharmonic)

#### Markw4

...every audio signal can be made by its fundemental + the harmonics of that fundemental frequency.
Not really. Some audio signals are composed of multiple non-harmonically related frequencies.

Also, phase of each harmonic has some effect on the resulting time domain waveform. In some cases the phase relationships can be audible, which is to more or less in keeping with the idea of a threshold of audibility for group delay.

#### torgeirs

Yes, agree there are exceptions. So almost every
And then the signals are combined, but can still be decomposed as seperate signals regarding harmonic distortion.