Ok.. talk is cheap. Take this test and see how sensitive you are to the major nonlinear distortion mechanisms in dynamic loudspeakers.
http://www.klippel-listeningtest.de/lt/default.html
There are some good info about the subject at the site. Read it if you haven't.
I don't know the exact performance of the driver choosen or how the calculation is done, but a least it would be interesting to see how much we differ in our ability to detect distortion.
Hoping for an intresting discussion.
/Peter
http://www.klippel-listeningtest.de/lt/default.html
There are some good info about the subject at the site. Read it if you haven't.
I don't know the exact performance of the driver choosen or how the calculation is done, but a least it would be interesting to see how much we differ in our ability to detect distortion.
Hoping for an intresting discussion.
/Peter
I reached -30dB testing L, -18dB testing Bl+C+L and -3dB for the Bl.
If Klippel choosed a medium performance driver of about -55dB @90dB THD in the midrange (dominantly 2nd and 3rd order harmonics) this means that I would need a driver with -75dB THD in order to not be able of hearing nonlinear artifacts.
A little simplified reasoning perhaps but it mirrors earlier findings the way I see it.
/Peter
If Klippel choosed a medium performance driver of about -55dB @90dB THD in the midrange (dominantly 2nd and 3rd order harmonics) this means that I would need a driver with -75dB THD in order to not be able of hearing nonlinear artifacts.
A little simplified reasoning perhaps but it mirrors earlier findings the way I see it.
/Peter
So here are my results with JVC earbuds at my "comfortable listening level" - might be a bit low for somebody
6inch
Pop:-18 dB
Pop - Bl: -9 dB
Pop - L: -12 dB
Octave: -33 dB
Octave - Bl: -30 dB
Octave - C: -9 dB
Octave - L: -36 dB
Terz: -18 dB
Twintone: -33 dB
telecom:
Vocals: 0 dB
Octave: -24 dB
Terz: -3dB
Twintone: -21 dB
That was my first try. Some of the distortions are quite easy to detect.
6inch
Pop:-18 dB
Pop - Bl: -9 dB
Pop - L: -12 dB
Octave: -33 dB
Octave - Bl: -30 dB
Octave - C: -9 dB
Octave - L: -36 dB
Terz: -18 dB
Twintone: -33 dB
telecom:
Vocals: 0 dB
Octave: -24 dB
Terz: -3dB
Twintone: -21 dB
That was my first try. Some of the distortions are quite easy to detect.
poptart said:
I am not sure that I know what you mean by basic harmonic distortion and THD?
THD in this test is just a number, that is the total distortion from the respective phenomena - spectrums are most probably different - as are different the sensitivities for average man. I gues that some kind of spectrum is more audible than other.
poptart said:
When you measure THD of a speaker you are more or less measuring the 2nd and 3rd harmonic distortion of a sinewave. This itself may not be very harmful, interesting or whatever BUT it is an indication of the nonlinear transfer curve of the speaker.
If a device has "THD" it will also have IMD when playing wide bandwith signals, makes sense?
The Klippel test (if we trust it being correct and representative for a typical decent performance product) is very telling. It shows that the distortion levels need to be lowered by about 30dB for a speaker to not color the music by its nonlinear transfer function.
THD is an indication of nonlinearity. You never hear only THD from a nonlinear device listening to music, you will always hear a mix of harmonic distortion and non harmonic intermodulation distortion.
The Klippel test is excellent IMO.
If you want to listen to pure harmonic distortion all you have to do is download some freeware tone generator and mix signals yourself and adjust levels between the fundamental and the choosen overtone/harmonic. Doing that you'll see that you can have pretty high levels of 2nd harmonic but as you go up the scale you will hear lower and lower levels.
/Peter
cbdb said:
Most decent raw drivers has about 0.3%-0.5% 2nd and 3rd order harmonics at aprox. 90dB/1m. Some SOTA designs has 0.03%-0.1%, not many have listened to such designs though and therefore have come to wrong conclusions about the current SOTA regarding hifi and studio monitor speakers, something that gets me going once in a while! ;-)
/Peter
Right, I understand and have done the Klippel test before and read about Earl's work on the Gedlee metric and the complete lack of correlation between THD and subjective impression. I thought this test was too specific to issues in the magnetic circuit of speakers to apply to your very broad comments in the other thread about microphones, amplifiers and speakers requiring low THD.
"Speakers that have about 0.1%THD (or below) at 90dB SPL tend to sound very clean, natural and relaxed."
That is very likely a true statement, but it implies that a speaker with several percent THD would *not* sound clean, natural and relaxed and this may not be true if that distortion is simple 2nd harmonic. As we all know 0.1% 9th harmonic is more objectionable than several % 2nd harmonic so your statement, and THD as a measurement, are meaningless. The Klippel test is very specific and obviously meaningful, but it does not connect to your general statements about THD. That's why I asked for a general THD test so people could see for themselves that low THD numbers don't necessarily mean better sound.
Nobody has ever claimed that we can't hear any form of distortion, only that THD is a useless figure to quantify the audibility of distortion.
"Speakers that have about 0.1%THD (or below) at 90dB SPL tend to sound very clean, natural and relaxed."
That is very likely a true statement, but it implies that a speaker with several percent THD would *not* sound clean, natural and relaxed and this may not be true if that distortion is simple 2nd harmonic. As we all know 0.1% 9th harmonic is more objectionable than several % 2nd harmonic so your statement, and THD as a measurement, are meaningless. The Klippel test is very specific and obviously meaningful, but it does not connect to your general statements about THD. That's why I asked for a general THD test so people could see for themselves that low THD numbers don't necessarily mean better sound.
Nobody has ever claimed that we can't hear any form of distortion, only that THD is a useless figure to quantify the audibility of distortion.
If we could see the spectrum of the distortion from the test, then we would most probably see strong 2nd and 3rd harmonic which should be relevant to most loudspeakers.
Either I get it wrong or there IS a strong corellation between THD and listening experience, if you take into account that this THD figure is dominated by 2nd and 3rd harmonic THD is about the simplest number that can let us compare two loudspeakers. I do not believe that there are many woofers with higher order harmonics in the distortion. 1 % THD is -20 dB, 0.1 % THD is -40 dB, right? I could easily hear -30 dB THD with their model speaker - and I would bet it was mostly 2nd and 3rd harmonic. Nobody in the world can persuade me that distortion does not matter.
Or the test is not relevant and contains more higher order harmonics
Either I get it wrong or there IS a strong corellation between THD and listening experience, if you take into account that this THD figure is dominated by 2nd and 3rd harmonic THD is about the simplest number that can let us compare two loudspeakers. I do not believe that there are many woofers with higher order harmonics in the distortion. 1 % THD is -20 dB, 0.1 % THD is -40 dB, right? I could easily hear -30 dB THD with their model speaker - and I would bet it was mostly 2nd and 3rd harmonic. Nobody in the world can persuade me that distortion does not matter.
Or the test is not relevant and contains more higher order harmonics
pelanj said:
There is plenty of data that shows that 2nd and 3rd harmonics are largely ignored (to a point) by the brain...higher orders, especially odd orders, are another ball of wax. With loudspeakers, I would assume that IMD is the biggest contributor to detecting distortion since it has no harmonic relationship to the music. Of course, you need harmonic distortion to generate IMD, but it isn't a linear relationship...design, materials, construction, etc., all play a role in its generation.
poptart said:
Can we take one thing at a time and start somewhere?
What do you believe is the main cause of distortion in speakers? What do you suggest we test for instead of the nonlinear Compliance, Bl (power factor) and L (voice coil inductance)?
Of course this test was specific to speakers.. when you want to find out something about speakers it makes a whole lot of sense testing speakers don't you think?
There are other sources of distortion also but the above is always there.
And this test is not only about magnetic circuits, it's also about mechanical distortion from the suspension and spider as well as the voice coils inductance.
Correct, studies and trials indicate this is the case and so did this Klippel test. I would never bring a speaker into my house that has several percent distortion above 100Hz.
And here is that common misunderstanding again. There are no audio devices that only has 2nd order distortion playing music, it's a myth build on incompletete understanding. I did already mention that in this thread. Before you have a grip on such basics there's no meaning arguing about the rest, really.
Not at all if you have understanding of the device and what is measured. It doesn't tell the whole story but together with other information it tells a lot. It's not like you will find -55dB 9th harmonics in a speaker designed by someone with a brain.
Which I never claimed, but still the connection is there.
I don't know what a general THD test is. Maybe we can design a test in order to find the answer to some specific question? I don't really understand your point though. And yes, everything equal, lower HD measurements will results in better sound until the distortion products are below the threeshold of audibility. My definition of better is being more true to the source.. I am not interested in what some guy find subjectively pleasing.
It's not useless if you have baisc understanding of the device at hand. OTOH it tells a lot about the device. BUT, that should not be read as I think that any type of THD reading on any type of audio gear can be compared, I never said that.
/Peter
I agree - THD can be measured (AFAIK) with just a sinewave at a certain level - but I think a series of measurements like this can give a hint about total distortion performance / nonlinearity. You can see what I mean for example in this datasheet: http://www.seleniumloudspeakers.com/site2004/catalogo/pdf/12W10P.pdf
You are right about IMD - with two tones, I could easily hear the modulation products (kind of a tremolo effect). Guitarist use this to tune a guitar - you play the same harmonic on two strings - the faster the sound pulsates, the more the strings are out of tune.
By the way, have you noticed that in the octave and two tone test the distorted one seems as it is "out of tune", a bit lower than the other?
You are right about IMD - with two tones, I could easily hear the modulation products (kind of a tremolo effect). Guitarist use this to tune a guitar - you play the same harmonic on two strings - the faster the sound pulsates, the more the strings are out of tune.
By the way, have you noticed that in the octave and two tone test the distorted one seems as it is "out of tune", a bit lower than the other?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.