What on earth is intermodulation distortion?
Guitarist on youtube
Listening to IMD is all the evidence you need which support the statement that IMD is always worse than a fairly high amount of THD.
Hi ! thanks a lot that is what i wanted to know ... how bad can be IMD.
And above all a very easy test to do with basic tools ... strange that it is not more popular ...
Hi ! thank you sincerely for your kind and very valuable explanation. Honestly all this is much beyond my ability to understand.
I can only capture basic concepts
My point was simply that some transducers seem to add other signals to the test tones. They create something not present in the original signal. They are not true to the source.
Moreover choosing rightly the tones it is possible to push a driver to its limits.
I smile a little when i hear of incredible bass out of a 5" woofer ... which spl ? which distortion ?
I used to listen to a small pair speakers with a 6" woofer and be happy with the bass they were producing.
Then i listened in a larger room to a pair of 15" Tannoy dual concentric
... there was more bass.
Last edited:
mr. linkwitz used three tone tests for imd and hd measurements of drivers, what do you think of this method?
Midrange distortion test
Midrange distortion test
Hi ! thank you very much indeed for the link to these great papers
So much valuable information ... so it looks like distortion measurements are of some help in evaluating drivers/speakers.
Good ! Looks like a really good small-format 1" dome tweeter if you have the ability to fix the frequency response.
The IMD measurements have been added to SB34NRXL75-8 data:
SB Acoustics SB34NRXL75-8 | HiFiCompass
The measurements of the ScanSpeak 32W/8878T11 woofer have been added:
ScanSpeak 32W/8878T11 | HiFiCompass
SB Acoustics SB34NRXL75-8 | HiFiCompass
The measurements of the ScanSpeak 32W/8878T11 woofer have been added:
ScanSpeak 32W/8878T11 | HiFiCompass
I don't like that the modulation products coincide. It makes it that much harder to pull apart the orders and leaves you with fewer difference and over-tones. In this sense, I prefer sticking with only two input tones to manage overlap.
WOW!
-that extra money for the Scan Speak is actually going to something extra!
Not sure it's worth x2.5 the SB though..
Last edited:
WOW!
-that extra money for the Scan Speak is actually going to something extra!
Not sure it's worth x2.5 the SB though..
Are you referring here to the <100hz HD measurements? At a quick glance, that is the only serious edge the scanspeak seems to have... But maybe I didn't look closely enough? If the price is 2.5x, it doesn't seem much of an edge.
Absolutely, it is after all: a bass driver.
Scale the power and look at the freq.s below 40 Hz (nearfield of course). Scan Speak's results here are atypical.
I am trying to choose a 6 - 7" mid range driver with a good 300-2800 hz area sound quality. I was looking at these measurements and to my surprise the Purifi PTT6.5W04-01A seemed to win the twice as more expensive ScanSpeak Ellipticor 18WE/4542T00 in harmonic distortion. I also noticed that Satori could not compete with those two.
Just wanted to verify, that did I look at the graphs right? Also, has any driver measure better than the Purify so far? I wonder if a 8 ohm version exists or not.
Just wanted to verify, that did I look at the graphs right? Also, has any driver measure better than the Purify so far? I wonder if a 8 ohm version exists or not.
Yes, the overall THD is lower for the Purifi PTT6.5 than for any other speaker in that frequency range. It is mostly due to the very low 2nd order distortion. But particular behavior of 3, 4 and 5 orders a bit better for the 18we and Satori.
You should take a look at SB17NBAC35 too.
Anyway, ultimate choosing the midrange speaker basing on distortion figures only is not a great idea.
I hadn't heard about Purifi 8 Ohm version so far.
Because it's difficult to represent the data in a way that is easy to interpret as having an additional tone adds an extra dimension to the data.However it is very important for me to understand the importance of the IMD measurements.
And again i cannot help but notice that they are very rare. Why ?
A harmonic distortion sweep can be represented as a 2-D plot. This is easy and intuitive to interpret (at least I think so
), great!If you were to do a 2-tone IMD sweep the plot would become 3-D as there would be two frequency axes which the data was plotted on so you could look up the resulting level of IMD for any combination of two frequencies. So you'd have overlapping 3-D surfaces for however many orders of IMD you want to plot and that just becomes incredibly confusing to interpret as they visually obscure each other.
If you do IMD measurements at only specific frequencies then the data becomes less informative as a peak in distortion can hide between two discrete measurements. E.g. a peak in distortion when 1kHz is played could be completely missed if you only make measurements with tones at 500Hz and 2kHz. To be sure this never happens, IMD measurements must be swept.
One thing I have tried is a dual swept IMD where one tone's frequency is a non-integer multiple of the other. The non-integer multiple is required such that the IMD falls on frequencies which are different than harmonic distortion products.
By sweeping with a fixed multiple between the two tone frequencies, the sum of the resulting IMD tones for each order can be plotted exactly as a HD sweep would on a 2-D plot. You'd just add two scales for the x-axis to represent the frequencies of each of the tones.
Unfortunately room effects and measurement noise floor are much more of a problem than with HD sweep measurements. Firstly because there are up to four significant distortion products for each resulting distortion order (= 4x worse noise floor). Secondly because you must use a wide gate to measure subtractive distortion products occurring at much lower frequencies than the two fundamental tones. Lastly because decaying noise from sweeping fundamentals earlier on in the sweep occupies the same frequencies as subtractive distortion products occurring later in the sweep. HD sweeps don't have the latter problems because you're always measuring HD at higher frequencies than has been previously played in the sweep, and because your gate only has to be wide enough for the fundamental. To implement this method for IMD measurement successfully you need a really big room, and sweep really slow.
Le(x) and Bl(x) measurements are good predictors of IMD performance, the data can be easily plotted/interpreted and you don't need to worry about making a measurement which is blind to a narrow-band spike in distortion at a certain frequency.
Last edited:
Hi thanks a lot for the very kind and helpful explanation that i cannot understand entirely for my intellectual limits
As i said before very sadly i am bad at math and therefore not that smart.
But i understand that in general there is an ideal behaviour and a real one.
The ideal driver when a double tone is sent in outputs the same double tone without any intermodulation products (i guess they are called like this)
The less IM products are present in the output the more ideal the behaviour of the driver and so the better.
Of course ideal is not real ... a compromise must be accepted.
What i really cannot accept is that the drivers selection is done by ear. This happens only in audio fidelity.
I have the feeling that the more ideal the driver the best also must be its sound.
Often i hear that material cone is important for the sound. Paper cones sound different from plastic, ceramic, metallic ones ... i would bet they have different distortion spectra
Moreover not always what sounds more pleasant is the more true to the original sound.
But high distortion is not a sign of fidelity at all.
If i remember well some manufacturers like JBL used to provide distortion figures for their professional drivers and in the case of compression drivers indicating the horn model used during the testing
No need to say that their drivers are very low in distortion and sound excellent ... i would put them against so called high end speakers for a comparison. Just out of curiosity.
I am particularly interested in low distortion woofers of 8" and 6.5" for an active speaker that i would like to put together ... actually two.
I am very intrigued by metal cone woofers ... as i am using 24dB/octave filtering the high Hz spikes could be tamed a little.
I like their stiffness. And i also like concave woofers. Like these for instance
https://images-na.ssl-images-amazon.com/images/I/61WHLAYljeL._SL1018_.jpg
so much that i am thinking to cannibalize them ...
As i said before very sadly i am bad at math and therefore not that smart.
But i understand that in general there is an ideal behaviour and a real one.
The ideal driver when a double tone is sent in outputs the same double tone without any intermodulation products (i guess they are called like this)
The less IM products are present in the output the more ideal the behaviour of the driver and so the better.
Of course ideal is not real ... a compromise must be accepted.
What i really cannot accept is that the drivers selection is done by ear. This happens only in audio fidelity.
I have the feeling that the more ideal the driver the best also must be its sound.
Often i hear that material cone is important for the sound. Paper cones sound different from plastic, ceramic, metallic ones ... i would bet they have different distortion spectra
Moreover not always what sounds more pleasant is the more true to the original sound.
But high distortion is not a sign of fidelity at all.
If i remember well some manufacturers like JBL used to provide distortion figures for their professional drivers and in the case of compression drivers indicating the horn model used during the testing
No need to say that their drivers are very low in distortion and sound excellent ... i would put them against so called high end speakers for a comparison. Just out of curiosity.
I am particularly interested in low distortion woofers of 8" and 6.5" for an active speaker that i would like to put together ... actually two.
I am very intrigued by metal cone woofers ... as i am using 24dB/octave filtering the high Hz spikes could be tamed a little.
I like their stiffness. And i also like concave woofers. Like these for instance
https://images-na.ssl-images-amazon.com/images/I/61WHLAYljeL._SL1018_.jpg
so much that i am thinking to cannibalize them ...
Last edited: