Head-Fi.org › Forums › Equipment Forums › Sound Science › Question about Harmonic Distortion
New Posts  All Forums:Forum Nav:

# Question about Harmonic Distortion - Page 2

This thread is old but still very informative. I guess THD+noise at 2%-4% or higher should be audible, assuming there is music in the specific frequency range and headphones don't roll off there. Not sure if THD+noise lower than 1% can be easily audible in headphones.

I heard bass distortions in my Creative Aurvana Live headphones before reviewing Innerfidelity's measurement (whose THD+noise data seems to explain what I heard). Their bass distortion was significantly reduced after we did mass-loading and damping (MarkL mod) to the drivers. I am curious what the THD+noise is after mod.

And I also have some potentially stupid questions: since even order distortions are less audible than odd order ones, is there a way to covert odd order distortions of headphones to even order ones?

And is THD+noise more audible in IEMs than full-size headphones (assuming both have the same THD+noise measurement)?
Edited by zzffnn - 5/6/13 at 3:11pm

Quote:
Originally Posted by zzffnn

And I also have some potentially stupid questions: since even order distortions are less audible than odd order ones, is there a way to covert odd order distortions of headphones to even order ones?

And is THD+noise more audible in IEMs than full-size headphones (assuming both have the same THD+noise measurement)?

No and no.

^ Thanks. I would be highly interested to know why, if you are capable of providing such reasons.
Quote:
Originally Posted by zzffnn

^ Thanks. I would be highly interested to know why, if you are capable of providing such reasons.

It's actually not as easy to explain as one would hope, but I'll try.

Distortion is most easily understood by considering the case of a pure tone, a sine wave.  A perfect sine wave consists of only one frequency, the fundamental, the pitch we hear.  If you graphed the spectrum of a perfect sine wave there would be one frequency on the graph, and no others.

We now pass that sine wave through a device that creates distortion.  Distortion is created by several types of non-linear behavior, meaning, the shape of a pure sine wave is altered somehow.  When that happens, the sine wave no longer consists of just one frequency because harmonics are generated by the distortion of the wave shape.  There are different mechanisms that cause distortion, and different kinds of distortion results.  We've been talking about even-order  harmonic distortion vs odd-order.  Even order distortion would be the fundamental of the sine wave, say 1KHz as an example, plus the second harmonic at 2KHz, plus the 4th harmonic at 4KHz, plus the 6th harmonic at 6KHz, and so on.  The specific amount of each harmonic is highly variable and depends on the mechanism that caused the distortion. But the key is, all distortion by-products are frequencies at even harmonics.

A different mechanism might cause odd-order harmonics to appear.  This kind of mechanism is again very different from the even-order one, and produces harmonics at the 3rd, 5th, 7th, 9th, and so on.

So, in answer to your question (which was NOT stupid at all!) "since even order distortions are less audible than odd order ones, is there a way to covert odd order distortions of headphones to even order ones?", is that even order harmonic distortion is caused by a different mechanism than odd. To change the kind of distortion in a headphone we'd have to change whatever caused it in the first place, and that would be a new design.

However, if you had very accurate knowledge of exactly how the distortion came to be, and exactly what mechanism caused it, it is sometimes possible to create the inverse mechanism, to create "anti-distortion" (the correct term is "pre-distortion") and compensate in advance for the distortion that is about to occur.  This won't work in all cases because some distortion mechanisms are so brutal and abrupt there no way to pre-distort for them.  But if an amplifier has a lot of gentle odd-order distortion, it may be possible to feed it from a circuit that pre-distorts the signal in an opposite way so that the end result has less distortion in it. It's not a process you can just add to an amp, it has to happen in the design itself, though.

Your second question was "would THD+N be more audible on headphones or IEMs, assuming they both had the same inherent THD."  The audibility of distortion during music is a very complex subject.  Some music will mask distortion quite well, other music will expose it.  But if your headphones and IEMs have similar frequency response characteristics, and similar THD specs, then we have to ask what would cause distortion to be more audible on one or the other?

Distortion, recall, is additional harmonic frequencies added to the original signal by some form of nonlinearity in the system.  We hear distortion because we hear those new harmonics.  We hear their frequency and level.  We already know we can't change the frequency of the distortion harmonics, and if the frequency response of a headphone or IEM doesn't change their level, then the distortion wouldn't sound any different on either one.  But if you have some headphones that have radically different frequency response, and they have the unfortunate ability to boost response around  3KHz, and you have a recording of a solo flute playing notes around 1KHz, and there's no other music content to mask the distortion harmonics, then you might hear the distortion more.  But that's a whole lotta "if"s and "ands".  Similar sounding headphones won't make distortion any more or less audible.

^ Thank you very much for making things so easy to understand. I highly appreciate your detailed and helpful explanations.

Also thanks for going extra miles in writing paragraphs #1 - #4 as introduction, paragraphs after #5 are good enough for me, but paragraphs #1 - #4 make it easy for most readers.

Sorry for not carefully defining my questions. There would be lots of "if's" involved. I did know that my question #1 was a bit crazy.

You cleared up my confusion in my question #2. I thought that distortion may be more audible in IEMs (assuming same FR) due to:

1) generally more noise isolation (less masking/interference) ; and

2) closer distance of drivers (to ears).

Now that I think about it:

Re 1) it would depend on what kind of noise masking which distortion (e.g., low frequency noise may mask bass distortion, if the specific IEMs can block low frequency noise to start with);

Re 2) FR and distortion graphs of IEMs are measured with a much closer distance (than full-sized headphones) to artificial ears. So once we match volumes, distance of drivers should not make difference in terms of distortion perception.

As a follow-up question: lots of dynamic headphones have significant THD+noise below 40 Hz. What causes that? Is it due to:

1) drivers' inability to move at < 40 Hz (e.g., high acoustic impedance, or not enough driving/magnetic power)?

2) driver's mechanical noise (e.g., rattling)? Mass-loading and damping should reduce distortion here (e.g., see MarkL mod of Denons), correct?

3) other reasons?

Thanks again!

Edited by zzffnn - 5/9/13 at 8:31am
Quote:
Originally Posted by zzffnn

^ Thank you very much for making things so easy to understand. I highly appreciate your detailed and helpful explanations.

Also thanks for going extra miles in writing paragraphs #1 - #4 as introduction, paragraphs after #5 are good enough for me, but paragraphs #1 - #4 make it easy for most readers.

Sorry for not carefully defining my questions. There would be lots of "if's" involved. I did know that my question #1 was a bit crazy.

You cleared up my confusion in my question #2. I thought that distortion may be more audible in IEMs (assuming same FR) due to:
1) generally more noise isolation (less masking/interference) ; and
2) closer distance of drivers (to ears).

Now that I think about it:
Re 1) it would depend on what kind of noise masking which distortion (e.g., low frequency noise may mask bass distortion, if the specific IEMs can block low frequency noise to start with);
Masking distortion with noise and masking distortion with music are completely different, not exactly comparable. You need a ton of noise to mask music, basically more noise than music, but a single musical note can mask another close in frequency but lower in level completely...it's actually the basis for .mp3 and the like. So the ability or inability of a headphone design to block external noise wouldn't be a factor in distortion audibility.
Quote:
Originally Posted by zzffnn

Re 2) FR and distortion graphs of IEMs are measured with a much closer distance (than full-sized headphones) to artificial ears. So once we match volumes, distance of drivers should not make difference in terms of distortion perception.
Right.
Quote:
Originally Posted by zzffnn

As a follow-up question: lots of dynamic headphones have significant THD+noise below 40 Hz. What causes that? Is it due to:
1) drivers' inability to move at < 40 Hz (e.g., high acoustic impedance, or not enough driving/magnetic power)?
2) driver's mechanical noise (e.g., rattling)? Mass-loading and damping should reduce distortion here (e.g., see MarkL mod of Denons), correct?
3) other reasons?
Other reasons, but you're close with #1. The perfect transducer would move its diaphragm exactly in proportion with the driving signal. But we have materials with elasticity, and each diaphragm has limits as to how easily it flexes, how far, and how fast. It just simply cannot perfectly follow the input signal. Each design has different physical constraints, but they all result in some form of nonlinear behavior, and distortion is the end result.
Quote:

Quote:
Originally Posted by zzffnn

As a follow-up question: lots of dynamic headphones have significant THD+noise below 40 Hz. What causes that?Is it due to:
1) drivers' inability to move at < 40 Hz (e.g., high acoustic impedance, or not enough driving/magnetic power)?
2) driver's mechanical noise (e.g., rattling)? Mass-loading and damping should reduce distortion here (e.g., see MarkL mod of Denons), correct?
3) other reasons?

Other reasons, but you're close with #1. The perfect transducer would move its diaphragm exactly in proportion with the driving signal. But we have materials with elasticity, and each diaphragm has limits as to how easily it flexes, how far, and how fast. It just simply cannot perfectly follow the input signal. Each design has different physical constraints, but they all result in some form of nonlinear behavior, and distortion is the end result.

Thank you for the enlightenment.

I thought mass-loading dynamic drivers can tighten bass by reducing distortion? Please kindly comment. How does mass-loading  tighten bass then? We did MarkL mod and did hear tightened bass. Details of MarkL mod can be found here:

Quote:
Originally Posted by zzffnn

Thank you for the enlightenment.

I thought mass-loading dynamic drivers can tighten bass by reducing distortion? Please kindly comment. How does mass-loading  tighten bass then? We did MarkL mod and did hear tightened bass. Details of MarkL mod can be found here:

Mass loading changes the resonant frequency of the transducer.  I cannot comment on the results of that specific mod.  There were no before and after measurements, only subjective opinion, which could express a preference without any real change present, or indicate an actual improvement.  We don't know what MarkL heard, we only know what he said he heard.  I don't mean to discredit him in any way, there are plenty of reasons his mod would change something, but there's no documentation, only opinion of one person.

Thanks. Fair enough. I also lean heavily on the objective side and can understand your view.
I heard the bass improvement of DIY MarkL mod though. I used Blutak instead of Dynamat there and modded my headphones. To my ears, their flappy bass is no longer there. Please take my statement with a grain of salt though, as my ears are pretty insensitive to silver cables, iPhone DACs or LAME V0 mp3 vs lossless vs hi-Rez :-p
Edited by zzffnn - 5/11/13 at 10:31am

When you say heavier drivers, what do you mean? Does the mod even touch the diaphragm?

Sorry for my mistake. Please forgive my morning without coffee / tea. Mass-loading or MarkL mod does not touch the diaphragms.

So could it be that mass-loading reduces driver resonance which happens to be in bass region? Can that driver resonance be revealed in bass region of the THD+noise graph?

Look at the impedance plots. Most headphones show a distinct impedance peak in the bass region (phase = 0°), which is where the resonant frequency is.

^ Thanks.

http://www.innerfidelity.com/images/DenonAHD5000.pdf
http://www.innerfidelity.com/images/CreativeAurvana.pdf

The above headphones are targets of the mentioned mass-loading / MarkL mod. I did see impedance peak in bass region (CAL's peak is milder than D5000).

So your answer is yes to my above questions? Sorry, I am a biologist PhD and no EE. I would highly appreciate it if you can use layman's language to explain it for everyone's understanding. Thanks again!
Edited by zzffnn - 5/11/13 at 11:13am

Yes and no. The mod could change the resonant frequency, but that frequency does not tell you anything about THD+N or vice versa. Assuming you're using an amp with low output impedance, the peak in the impedance curve does not even influence the frequency response.

Edited by xnor - 5/11/13 at 12:35pm
^ Thanks. So such mod change (mass-loading reducing of flappy bass) may be revealed in CSD plots, assuming that CSD plots are reliable around ~40Hz?
New Posts  All Forums:Forum Nav:
Return Home
Back to Forum: Sound Science