The THD and IMD probably do have the highest "numbers" for loudspeakers, but the numbers don't mean anything.
This is an old topic that I have discussed at length and published books and papers on. THD and IMD simply do not correlate to perception. .1% THD in an amp can sound horrific, while 15% THD in a loudspeaker can be inaudible. How can one possibly use a scale like that to talk about "distortion"!!??
Also, as is usual with discussions like this, people are not all talking about the same things. What I was talking about is "nonlinear distortion" not "linear distortion" like frequency response.
Linear distortion perception is a very complicated matter as one has to now take the room and goal into consideration. All rooms are different and so are most goals making convergence on any reasonable answer problematic.
Linear distortion perception is a very complicated matter as one has to now take the room and goal into consideration. All rooms are different and so are most goals making convergence on any reasonable answer problematic.
Member
Joined 2009
Paid Member
it must be true then
Simplistic measurements gives meaningless numbers but are required by the marketing department because the customer want it.
The customer is you.
Yes, with high potential timing, spatial and power response anomalies being of importance.
And harmonic distortion.. I've built a (M/T range) valve amplifier, for low distortion of course but it isn't as low as my SS bass amp yet the simplicity of the distortion produced makes it a non-issue.. Same with speakers, I often find.
it must be true then
While you can argue with Earl, and people have been doing so since I was on PE Tech Forum 13 years ago, he has impeccable academic credentials and follows clear scientific research criteria.
Just because I don't want to agree with Earl doesn't mean that he's not right and I'm actually in left field.
I STOPPED USING POINT SOURCE SPEAKERS AWHILE AGO BECAUSE OF INHERENT DISTORTION.
When you ask one speaker, no matter how well behaved, to handle the whole of a range, and then play it loud, the speaker's FR will degrade, and the IMD will increase.
The only solution is a line array. When you ask 20 small speakers to handle the three octaves of the midrange, and 7 ribbon tweeters to handle the 2 octaves of the upper ranges, you will always have less IMD and a flatter frequency response.
And in the bass areas, its not only harder to hear the distortion, but most people actually prefer it.
When you ask one speaker, no matter how well behaved, to handle the whole of a range, and then play it loud, the speaker's FR will degrade, and the IMD will increase.
The only solution is a line array. When you ask 20 small speakers to handle the three octaves of the midrange, and 7 ribbon tweeters to handle the 2 octaves of the upper ranges, you will always have less IMD and a flatter frequency response.
And in the bass areas, its not only harder to hear the distortion, but most people actually prefer it.
Look up CEA-2010 testing standards. It defines a standard to test subwoofer's distortion performance based on psychoacoustic audibility of distortion. It is not a hard THD number, but a threshold based on each harmonic component of the distortion. Because of the masking effect, earlier order distortions is allowed a higher amount of distortion before hitting their passing threshold. So, if a subwoofer passes the CEA-2010 at a certain output at a certain frequency, it will sound subjectively clean to listeners.
What gedlee is referring to is that for bass frequencies, if the 15% THD is mostly dominated by the 2nd order harmonic distortion, then due to the masking effect, this 15% distortion is nearly inaudible to the human ear. Obviously, 15% distortion is unacceptable for anything higher than bass, but it is rather difficult to hit 15% distortion anyways with midrange.
Note, CEA-2010 only test distortion, it doesn't factor in much higher audibility problems like cabinet rattling or driver noise. So it is possible for a subwoofer to pass CEA-2010 at a certain SPL but still sound bad because of distortion sounds that doesn't come from harmonic distortion.
Last edited:
So how did the gedlee people compare the 15 percent distortion bass to the zero distortion bass (double blind test?) to come to the conclusion the 15 percent distortion bass is not audible?
The lowest distortion drivers I have measured are large format 4" diaphragm compression drivers. That being in the upper midrange and treble.
Go to his website. All his articles and white papers are there. They are not very exciting reading, but Earl goes through all the data methodically. I gave up after I fell asleep twice.
So how did the gedlee people compare the 15 percent distortion bass to the zero distortion bass (double blind test?) to come to the conclusion the 15 percent distortion bass is not audible?
The lowest distortion drivers I have measured are large format 4" diaphragm compression drivers. That being in the upper midrange and treble.
Easy, get a much bigger subwoofer. Say you have a ported 18" subwoofer that produces like 1% distortion at 100dB at 40Hz. Then you can compare that to a sealed 10" subwoofer that is distorting 15% at 100dB at 40Hz.
If it doesen’t matter then why don’t they all sound the same? Why aren't all speaker cabinets loaded with the same cheap drivers?
While some are saying it’s all just a scam the flip side is as likely true that this claim is agenda driven by those who don’t wish to spend on higher quality drivers.
If for one just don’t buy it.
Barry.
While some are saying it’s all just a scam the flip side is as likely true that this claim is agenda driven by those who don’t wish to spend on higher quality drivers.
If for one just don’t buy it.
Barry.
Because it is really easy for a normal speaker's bass drivers to hit 15% distortion when reproducing bass frequencies even at reasonable volumes. And there's much more to a subwoofer's performance than distortion. Thermal, port, and motor/suspension compression become very significant at higher playback volumes. What happens in the time domain is significant as well, and that is affected by the amount of cone control the motor has (start and stop). It is difficult to absorb low frequencies well, and there is an art to effective absorption. Things like port resonance, inductance, cabinet vibration and resonance all play a small but audibly noticeable difference. There are other things too, but the point is that there is much more to subwoofer performance than simply distortion.
Last edited:
No one is claiming "they all sound the same." The claim is that HD alone is not an adequate criterion for judging the quality of a driver, especially in lower frequency ranges. I concur.
However, I also believe that low level of HD does become more important as frequency gets higher. So a woofer with five or more percent distortion might not be audible, but a midrange with 5% might be awful, and a tweeter with 5% could be lethal. But perhaps it is still not the only or even the best factor to determine quality of a driver. Two tweeters, both with fractions of a percent HD at a "significant" frequency, can certainly sound different. Probably neither will sound bad, but each listener will prefer one over the other.
So you can't go shopping by the numbers. You can eliminate bad candidates, but you can't definitively select the "best" driver for you simply from looking at graphs or brand names.
Peace,
Tom E
However, I also believe that low level of HD does become more important as frequency gets higher. So a woofer with five or more percent distortion might not be audible, but a midrange with 5% might be awful, and a tweeter with 5% could be lethal. But perhaps it is still not the only or even the best factor to determine quality of a driver. Two tweeters, both with fractions of a percent HD at a "significant" frequency, can certainly sound different. Probably neither will sound bad, but each listener will prefer one over the other.
So you can't go shopping by the numbers. You can eliminate bad candidates, but you can't definitively select the "best" driver for you simply from looking at graphs or brand names.
Peace,
Tom E