Hello
Here's a post about analysis of the spectrum of distortion of intermodulation of loudspeakers (IMD) using my multitone signal of 14 non harmonic frequencies in a wav file.
Loudspeakers have two types of distortion: linear and non linear. The non linear distortion are the one we will test and are made of new frequencies in the acoustic signal of loudspeakers output, those frequencies was not in the original audio signal.
A way to measure it consists in sending two non harmonic frequencies, but only two frequencies don't represent a typical musical signal at all, and the normal use of the loudspeaker.
Another way to measure the IM distortion consists in replicating a more complex signal, and to measure the supplementary frequency level appeared in exit. The advantage of a multitone nonharmonic tones measure of the intermodulation distortions is that the signal test is large, more complex than the pure frequency used for the harmonic distortion or the two frequencies IMD measure. This distortion of intermodulation is therefore more representative of a typical musical signal, and of the normal use of the loudspeaker.
The signal used for the test is designated under the term of multitone, because it consists in a non harmonic frequency collection. They correspond to the big stripes on the picture of an analysis of the spectrum of distortions or distortion of intermodulation. Everything that is between these stripes is the distortion of intermodulation and the harmonic distortion; it is these undesirable signals that it is necessary to reduce to the minimum. Indeed, as for the harmonic distortion, the more the distortion of intermodulation is weak, the more the sound is clean and transparent. Contrary to the harmonic distortion the level of distortion of intermodulation is very correlative to the subjective sound quality.
To make this multitone signal test a low distortion signal is necessary, I have synthesized a low distortion multitone signal of 14 non harmonic frequencies in a wav file, (here's the frequencies;
22, 30, 52, 83, 123, 199, 316, 499, 802, 1248, 1998, 3152, 4997, 7997 hertz) and placed this file in this site web, so you can download it:
http://www.filedropper.com/g22hza8khzmultitone
To make the analysis of the spectrum of distortion of intermodulation while using my multitone signal of 14 frequencies, you can make a cd of it and use a cd player of the best quality, or you can use the wav file and an excellent sound card in full duplex mode and a software of analysis of distortion spectrum: Arta, SpecLab, Visual Analyser, ect... to see this distortion of intermodulation.
For the analysis of the distortion of intermodulation of a loudspeaker you also need a microphone of the best quality, you place the loudspeaker in a room that has the least possible reverberations, you can use a carpet and a thick curtains to decrease the reverberations.
I also include an image of the IMD of a very good loudspeaker using my test.
Bye
Gaetan
Here's a post about analysis of the spectrum of distortion of intermodulation of loudspeakers (IMD) using my multitone signal of 14 non harmonic frequencies in a wav file.
Loudspeakers have two types of distortion: linear and non linear. The non linear distortion are the one we will test and are made of new frequencies in the acoustic signal of loudspeakers output, those frequencies was not in the original audio signal.
A way to measure it consists in sending two non harmonic frequencies, but only two frequencies don't represent a typical musical signal at all, and the normal use of the loudspeaker.
Another way to measure the IM distortion consists in replicating a more complex signal, and to measure the supplementary frequency level appeared in exit. The advantage of a multitone nonharmonic tones measure of the intermodulation distortions is that the signal test is large, more complex than the pure frequency used for the harmonic distortion or the two frequencies IMD measure. This distortion of intermodulation is therefore more representative of a typical musical signal, and of the normal use of the loudspeaker.
The signal used for the test is designated under the term of multitone, because it consists in a non harmonic frequency collection. They correspond to the big stripes on the picture of an analysis of the spectrum of distortions or distortion of intermodulation. Everything that is between these stripes is the distortion of intermodulation and the harmonic distortion; it is these undesirable signals that it is necessary to reduce to the minimum. Indeed, as for the harmonic distortion, the more the distortion of intermodulation is weak, the more the sound is clean and transparent. Contrary to the harmonic distortion the level of distortion of intermodulation is very correlative to the subjective sound quality.
To make this multitone signal test a low distortion signal is necessary, I have synthesized a low distortion multitone signal of 14 non harmonic frequencies in a wav file, (here's the frequencies;
22, 30, 52, 83, 123, 199, 316, 499, 802, 1248, 1998, 3152, 4997, 7997 hertz) and placed this file in this site web, so you can download it:
http://www.filedropper.com/g22hza8khzmultitone
To make the analysis of the spectrum of distortion of intermodulation while using my multitone signal of 14 frequencies, you can make a cd of it and use a cd player of the best quality, or you can use the wav file and an excellent sound card in full duplex mode and a software of analysis of distortion spectrum: Arta, SpecLab, Visual Analyser, ect... to see this distortion of intermodulation.
For the analysis of the distortion of intermodulation of a loudspeaker you also need a microphone of the best quality, you place the loudspeaker in a room that has the least possible reverberations, you can use a carpet and a thick curtains to decrease the reverberations.
I also include an image of the IMD of a very good loudspeaker using my test.
Bye
Gaetan
Attachments
This is a re-packaging of Jon Risch's "Phi tone test". It is not popular with "marketing" or with audio neophytes because it does not produce a single number to point at (and brag about). It does, however, produce a clear-if-difficult-to-interpret graphical presentation of types of distortion that we definitely hear but which are not fully exposed by simpler harmonic distortion tests, particularly the broadband effect of zero-crossing distortion in amplifiers and the very real overall "grunge" that results from cone breakup and intermodulation in loudspeakers.
This is a type of test that manufacturers (and some audiophiles) broadly reject not because it is a bad test but because it so mercilessly exposes the flaws of bad equipment.
This is a type of test that manufacturers (and some audiophiles) broadly reject not because it is a bad test but because it so mercilessly exposes the flaws of bad equipment.
That's quite a claim. Can you point to any hardware that was audibly improved by the use of this testing? If it's that superior, it should produce superior products, right?
Tests don't "produce" products . . . they merely expose flaws that the actual producer should, but not always does, correct. And when not used they don't do anything at all . . .
In my own experience this is the only type of test that correlates well with "amplifier sound" (Class A amps, even those with relatively high harmonic distortion at rated power, generally test and sound "clean" using "spectral contamination" tests of this sort, especially at low power. As a first approximation you can attribute that to the absence of zero-crossing distortion.). What you see coming from loudspeakers can be downright shocking . . . especially if you believe that simple harmonic distortion tests of the drivers tell the whole story . . .
Please don't get pedantic about something we all know. We all know tests don't make products.
My point was, if this is such an important tool, it should have been proven somehow. I imagined that if a manufacturer used it they would have a tool to produce a better product. It sounds like it's just a test that produces interesting numbers that you don't know if it proves anything yet.
For instance, using this test to compare different op amps in headphone amplifiers. You use this test, and pick an op amp, and everyone goes "wow"
If we don't have something like that, it's just a mathematical exercise. I'm open to other uses, but this is the type of example I would expect to see before anyone calls it a worthwhile or revealing test.
My point was, if this is such an important tool, it should have been proven somehow. I imagined that if a manufacturer used it they would have a tool to produce a better product. It sounds like it's just a test that produces interesting numbers that you don't know if it proves anything yet.
For instance, using this test to compare different op amps in headphone amplifiers. You use this test, and pick an op amp, and everyone goes "wow"
If we don't have something like that, it's just a mathematical exercise. I'm open to other uses, but this is the type of example I would expect to see before anyone calls it a worthwhile or revealing test.
Hello
Here's few articles about the subject:
https://www.klippel.de/measurements/nonlinear-distortion/multi-tone-distortion.html
http://www.marshallchoong.co.uk/pdfs/intermod_testing.pdf
http://www.neumann-kh-line.com/neum...9025D8C4F126AD8C12578B2003A71E9?Open&term=TIM
Bye
Gaetan
Here's few articles about the subject:
https://www.klippel.de/measurements/nonlinear-distortion/multi-tone-distortion.html
http://www.marshallchoong.co.uk/pdfs/intermod_testing.pdf
http://www.neumann-kh-line.com/neum...9025D8C4F126AD8C12578B2003A71E9?Open&term=TIM
Bye
Gaetan
Here are some burst shaped tests by linkwitz lab
Issues in speaker design - 2
Toneburst Test Signal CD
while be very care full on higher frequency while your tweeter gets overloaded.
just my
hp
Hi,
I like using Phi tones defined by Jon Risch but modifying their amplitudes so the spectrum matches long term average music spectral balance. Attached is an example of testing an old Audigy card with the signal.
IEC 268-1, 1985 is a standard signal whose mean power spectral density is meant to be representative of a wide variety of music types. It’s basically flat from 100 Hz to 2 kHz, approx 6 dB down at 5 kHz and 12 dB down at 10kHz
Peter Chapman at B&O published a paper (AES preprint 4277) comparing a large number of CDs to the standard, while also looking at peak data. I was pretty close in my peak/rms estimates. Symphonic averaged 17 dB, pop, rock and hip hop from 8 to 10 dB. Average powers of most music followed the spec closely, except for hip-hop, which was almost flat from 300 Hz to 10 kHz, with 6 dB prominence at 60 Hz.
I like using Phi tones defined by Jon Risch but modifying their amplitudes so the spectrum matches long term average music spectral balance. Attached is an example of testing an old Audigy card with the signal.
IEC 268-1, 1985 is a standard signal whose mean power spectral density is meant to be representative of a wide variety of music types. It’s basically flat from 100 Hz to 2 kHz, approx 6 dB down at 5 kHz and 12 dB down at 10kHz
Peter Chapman at B&O published a paper (AES preprint 4277) comparing a large number of CDs to the standard, while also looking at peak data. I was pretty close in my peak/rms estimates. Symphonic averaged 17 dB, pop, rock and hip hop from 8 to 10 dB. Average powers of most music followed the spec closely, except for hip-hop, which was almost flat from 300 Hz to 10 kHz, with 6 dB prominence at 60 Hz.
Attachments
Am I misinterpreting what you or that referenced paper are saying? Of the greater than 6000 music tracks that I've personally analyzed and remastered to date, they don't look like that B&O paper or IEC standard. Rather they follow a 1/f curve much more closely in my experience, with a -16 to -18 dB/decade slope from about 100 Hz to 20 kHz, with some pop and other non-classical genres consistently accentuating the 2-8 kHz band.
European classical recordings (digital or analog) and solo instrument recordings usually have a large 400-800 Hz spectral peaks with decreasing spectral loudness across the tracks on both the HF and LF. This also appears to be placed there at mastering time, since when inverse EQ applied to those classical tracks to yield a much more 1/f trend (mentioned above)--it results in a much more realistic playback performance.
Deviations from that trend all appear to be related to "Loudness War" practices that have been around since at least the early 1960s (i.e., the introduction of stereo on phonograph records) and those practices of attenuating bass over and above the RIAA curves are probably as old as phonograph records themselves.
Also, all my data show that analog recordings have about the same or higher dynamic range (crest factors) as the original CD music tracks of the 1980s, unlike what the Chapman paper is reporting. See "Loudness War" Dynamic Range Compression & The DR Database - Observations - diyAudio for the gathered statistics based on DR Database values that have been crowdsourced from about 30,000 albums--roughly 300K tracks.
Chris
Last edited:
Chris,
I spent 30 mins going through my speaker paper library and just couldn't find the paper again, sorry, I must have lent it out. Which is too bad, its a good paper.
Thanks for sharing your results but I can't explain why they differ from the other papers. Pure conjecture without having the Chapman paper at my disposal, but perhaps the differences are due to differences in all or some of the following attributes of the measures: averaging, hangover, attack/decay time of envelope detectors?
Not much to hang a hat on, but I really need the papers to be able to answer this.
I spent 30 mins going through my speaker paper library and just couldn't find the paper again, sorry, I must have lent it out. Which is too bad, its a good paper.
Thanks for sharing your results but I can't explain why they differ from the other papers. Pure conjecture without having the Chapman paper at my disposal, but perhaps the differences are due to differences in all or some of the following attributes of the measures: averaging, hangover, attack/decay time of envelope detectors?
Not much to hang a hat on, but I really need the papers to be able to answer this.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.