Hmm... as I could clearly hear that out-of-band THD peak caused by the sound of the circuit board rattling in the subwoofer, I'd beg to differ with that statement 
.
For bass frequencies though, it seems that we are a lot less sensitive to distortion which would otherwise be considered excessive. Do you know if anyone's done any testing with taking the same alignment and adding different "THD profiles" to see if the differences are actually audible? Most of the tests I've seen concern THD for a specific frequency, or the same THD applied across a frequency band, not a band of frequencies with varying THD applied across the range.
.For bass frequencies though, it seems that we are a lot less sensitive to distortion which would otherwise be considered excessive. Do you know if anyone's done any testing with taking the same alignment and adding different "THD profiles" to see if the differences are actually audible? Most of the tests I've seen concern THD for a specific frequency, or the same THD applied across a frequency band, not a band of frequencies with varying THD applied across the range.
Well, if THD isn't an issue, why bother with going with high Xmax drivers at all? Just use the cheapest driver you can find in the size that you want to use, use EQ or DSP to achieve the frequency response you want to achieve, power it it up with an amp that provides enough power to reach the driver's RMS rating and enjoy the results .
.
Your second statement is basically correct.
THD is a measure of nonlinearity, but perceptually it is a poor one. There are much better ways to analyze driver nonlinearity than THD, and all those suggest that woofer nonlinearity is not an issue ... until it is. That is in a well designed driver of course.
And I could also make one that sounded like crap at low levels. With low THD to boot.
THD is a measure of nonlinearity, but perceptually it is a poor one. There are much better ways to analyze driver nonlinearity than THD, and all those suggest that woofer nonlinearity is not an issue ... until it is. That is in a well designed driver of course.
And I could also make one that sounded like crap at low levels. With low THD to boot.
I've found what you are saying about THD to be true with CD/horns, or rather higher frequencies, but empathicly not true for subs.
I believe anyone can easily hear THD in subs, if they know what to listen for.
As I said, I can make them audibly distorted or not. I prefer not. My system does not audibly distort, at any frequency I suppose, at tolerable listening levels. Is that to say that I could not run up someone else's until I heard nonlinearity. Of course not. You have to put things in perspective.
"I believe anyone can easily hear THD in subs, if they know what to listen for." Absolutely, but this perception will not correlate with THD as a measure.
"I believe anyone can easily hear THD in subs, if they know what to listen for." Absolutely, but this perception will not correlate with THD as a measure.
Last edited:
I think we're going around in circles.
Here's what I said in my original post:
"The question really should be if the distortion is appreciable (measured) and noticeable (subjective). 10% measured THD at low bass frequencies could be considered appreciable, but whether or not it's noticeable is another matter - that's basically up to the listener and what frequency range we are talking about."
...and...
"In general, I've found that different subwoofer alignments have different THD profiles, which generally have a lot to do with how the driver is loaded. 4th order bandpass alignments have the advantage of cutting a lot of the distortion generated in the passband by passing it through an acoustic filter, which results in pretty unique "THD profile" that perhaps accounts for part of the reason that it sounds "different". This THD profile could give the impression that it could be played louder before distortion becomes noticeable."
In summary, the audibility of distortion depends on what frequency range we're talking about (it's not really all that noticeable at bass frequencies until it's pretty high), and an acoustic bandpass alignment can move that point of perception a bit higher in SPL for a given driver. I have heard too many stories of people blowing their bandpass subwoofers because they did not hear any noticeable signs of the driver being driven beyond its electrical or mechanical limits.
I'm interested in the better ways to analyze driver nonlinearity that might be available to DIY hobbyists like myself. Sounds interesting...
Here's what I said in my original post:
"The question really should be if the distortion is appreciable (measured) and noticeable (subjective). 10% measured THD at low bass frequencies could be considered appreciable, but whether or not it's noticeable is another matter - that's basically up to the listener and what frequency range we are talking about."
...and...
"In general, I've found that different subwoofer alignments have different THD profiles, which generally have a lot to do with how the driver is loaded. 4th order bandpass alignments have the advantage of cutting a lot of the distortion generated in the passband by passing it through an acoustic filter, which results in pretty unique "THD profile" that perhaps accounts for part of the reason that it sounds "different". This THD profile could give the impression that it could be played louder before distortion becomes noticeable."
In summary, the audibility of distortion depends on what frequency range we're talking about (it's not really all that noticeable at bass frequencies until it's pretty high), and an acoustic bandpass alignment can move that point of perception a bit higher in SPL for a given driver. I have heard too many stories of people blowing their bandpass subwoofers because they did not hear any noticeable signs of the driver being driven beyond its electrical or mechanical limits.
I'm interested in the better ways to analyze driver nonlinearity that might be available to DIY hobbyists like myself. Sounds interesting...
There aren't really. My point is that I want many subs, not fewer. So a little boost in efficiency in a band pass was not really needed and the limited bandwidth turned out to be a problem. So I gave up all my bandpass subs, had three, and now use all closed box, with a lot of low end EQ on each one. The bandpass were two limiting.
But here is where I think we probably agree and that is people way underestimate the amount of LF air movement that is required to go low in frequency. 95% of the systems that I see have bass limitations IMO. My point is that one can get to the point where THD numbers just don't mean anything. I think the phrase is "Farting in the wind." Operate your bass source too close to its limitations and you will always have a problem.
But here is where I think we probably agree and that is people way underestimate the amount of LF air movement that is required to go low in frequency. 95% of the systems that I see have bass limitations IMO. My point is that one can get to the point where THD numbers just don't mean anything. I think the phrase is "Farting in the wind." Operate your bass source too close to its limitations and you will always have a problem.
If I could scatter multiple sealed subwoofers around the living room, I'd probably opt for that. Alas...
Different situations require different solutions....
For example:
1. In my car, I used sealed subwoofers - two 12" drivers in a low-Q alignment. That cabin gain makes them a great choice to get down to single-digit frequencies at pretty decent SPL levels with pretty low THD. Think "God whispering"
2. In my living room, my SO will beat me over the head with her broom if I fill it up with more boxes. It's mostly use for HT duties now. Bass duty is provided by one high-Xmax 10" driver in a vented alignment tuned to 21 Hz and EQ'd to get the best response in the usual listening location (the couch).
3. For my desk, where funnily enough I do most of my listening these days (work from home and all that), I have an 8" driver in an offset TL design that goes down to 32 Hz and reaches up far enough to blend with the iddy-biddy full range drivers I'm using on the desk, and one 8" subwoofer driver is more than enough to keep up with them
In the plans - two 12" drivers in small (~26L) passively-assisted 30 Hz-capable sealed alignments that will be located in in the living room near to the couch for nearfield subwoofer duty. As they are going to take the place of two small tables, and are going to be disguised as tables, I'm hoping that SWMBO won't mind.
Different situations require different solutions...
. For example:
1. In my car, I used sealed subwoofers - two 12" drivers in a low-Q alignment. That cabin gain makes them a great choice to get down to single-digit frequencies at pretty decent SPL levels with pretty low THD. Think "God whispering"
2. In my living room, my SO will beat me over the head with her broom if I fill it up with more boxes. It's mostly use for HT duties now. Bass duty is provided by one high-Xmax 10" driver in a vented alignment tuned to 21 Hz and EQ'd to get the best response in the usual listening location (the couch).
3. For my desk, where funnily enough I do most of my listening these days (work from home and all that), I have an 8" driver in an offset TL design that goes down to 32 Hz and reaches up far enough to blend with the iddy-biddy full range drivers I'm using on the desk, and one 8" subwoofer driver is more than enough to keep up with them
In the plans - two 12" drivers in small (~26L) passively-assisted 30 Hz-capable sealed alignments that will be located in in the living room near to the couch for nearfield subwoofer duty. As they are going to take the place of two small tables, and are going to be disguised as tables, I'm hoping that SWMBO won't mind
.Hi Earl,
You use the subs till ~25Hz, so whats your opinion on PR tuned very low to 25Hz (or maybe a bit lower). Even thats a small box with an octave of extension as compared to sealed that you currently use?
Thanks and Regards,
WA
Here's why i think THD does correlate with perception....for subs.
When i said 'I believe anyone can easily hear THD in subs, if they know what to listen for."....
I should have instead said what to listen for, which is a clear tonal shift upwards.
The 2nd and 3rd harmonics of anything in the sub range (I'll use up to 100Hz), are very distinctly higher in perceived tone to my ears and simply shift a sub's sound tonally upwards.
I don't know why i would throw that information away into a nondescript undefinable non-linear distortion bucket, when THD appears to describe what i am hearing.
I totally agree with another of your posts, saying 95% of systems have bass limitations. And that leads to folks driving their subs into non linear levels.
How to solve bass limitations is an entirely separate issue imo, from whether sub THD is a valid measurement correlating to perception.
Here's how knowing the subwoofer's THD profile helps.
That same THD profile for the Enigma that I published earlier, that showed it was fine for use in a small room, but when it was repurposed to be used in a bar at a higher SPL level, after reviewing its THD profile, I used DSP to engage a steep HP filter at around 30 Hz to significantly reduce the possibility of it being driven into noticeable distortion (and the driver possibly damaged in the process) if it ever saw a high-level input signal at such low frequencies (not that that will happen very often, but still). The THD profile basically took the trial and error and guesswork out of the equation. Now, I could have used the HP filter at a higher frequency to allow even more SPL from the subwoofer before noticeable distortion, but at that point the driver would be seeing much more than its rated power anyway - something I addressed via another tweak.
That same THD profile for the Enigma that I published earlier, that showed it was fine for use in a small room, but when it was repurposed to be used in a bar at a higher SPL level, after reviewing its THD profile, I used DSP to engage a steep HP filter at around 30 Hz to significantly reduce the possibility of it being driven into noticeable distortion (and the driver possibly damaged in the process) if it ever saw a high-level input signal at such low frequencies (not that that will happen very often, but still). The THD profile basically took the trial and error and guesswork out of the equation. Now, I could have used the HP filter at a higher frequency to allow even more SPL from the subwoofer before noticeable distortion, but at that point the driver would be seeing much more than its rated power anyway - something I addressed via another tweak.
The issues with perception vs. THD are complex.
If I take a given woofer and put in a pure sine wave and raise the level I will begin to clearly hear the higher harmonics AND this will correlate very well to the measured THD. But with music the audible level will be orders of magnitude different. This is problem 1 - isn't music the proper source for perception?
Now take a different woofer, one that has a "buzz or rub". These artifacts will be very high order and will not be a significant contribution to the measured THD especially as the level is increased. So let's say that the nonlinearity is such that the THD vs. level is the same in the two drivers. The second will sound terrible on music while the first will sound fine, yet both have the same THD. This is the 2nd problem - one cannot translate one drivers THD audibility to another unless the nonlinearities are identical - which never happens. Different forms of nonlinearity yield different perceptions and this is not captured with THD.
No one has every said that nonlinearity is not audible only that THD levels on music over a wide array of nonlinearities shows no correlation. It's the nature of the nonlinearity that matters not the THD that is created. But none of this gets captured with THD.
I look at the shape of the BL flux linkage vs. x and does the driver have a shorting ring. Look at the ac6ual construction of the driver and is it likely to produce higher order nonlinearity or lower order. Low orders are not nearly as audible as higher orders. THD tells us none of this.
If I take a given woofer and put in a pure sine wave and raise the level I will begin to clearly hear the higher harmonics AND this will correlate very well to the measured THD. But with music the audible level will be orders of magnitude different. This is problem 1 - isn't music the proper source for perception?
Now take a different woofer, one that has a "buzz or rub". These artifacts will be very high order and will not be a significant contribution to the measured THD especially as the level is increased. So let's say that the nonlinearity is such that the THD vs. level is the same in the two drivers. The second will sound terrible on music while the first will sound fine, yet both have the same THD. This is the 2nd problem - one cannot translate one drivers THD audibility to another unless the nonlinearities are identical - which never happens. Different forms of nonlinearity yield different perceptions and this is not captured with THD.
No one has every said that nonlinearity is not audible only that THD levels on music over a wide array of nonlinearities shows no correlation. It's the nature of the nonlinearity that matters not the THD that is created. But none of this gets captured with THD.
I look at the shape of the BL flux linkage vs. x and does the driver have a shorting ring. Look at the ac6ual construction of the driver and is it likely to produce higher order nonlinearity or lower order. Low orders are not nearly as audible as higher orders. THD tells us none of this.
Last edited:
Very true. But at no point does THD correlate to perception across a wide array of systems and source material. Only in very specific cases, as you mention, is this the case and it cannot be generalized to a general situation.
All your points are well taken, and easy to find agreement with.
I'm only going to quibble with this one a little.
It's not just audible with pure sine waves, or even pink noise,....it's with music too.
I do a lot of outdoor listening that naturally requires subs being driven harder to make sufficient SPL. So I've measured THD via sine sweeps and CEA2010, on bass-reflex and FLH subs.
As SPL goes up, the bass-reflex are quicker to start showing a rise in THD than the folded horns.
And the music sound matches that.
At equal SPL, lower freq transients and tone take on a sharper/punchier/grittier sound with the bass-reflex.
These comparisons are at levels well below x-max, well below the onset of port chuffing, and without mechanical rub or buzz issues.
Listening to music through the subs only, the sound of the FLH's (Danley's Labhorns and JTR orbitshifters) are remarkably "more low frequencies only" when low-passed identically to BR's.
This phenomenon starts occurring at a surprisingly moderate SPL, and only gets worse as cranked up.
So yeah, i have to think THD matters with subs...
But that said....nowhere else !
I used to think your comments about THD and CD's didn't make sense.
And so I made the same THD measurements for CD's, but without any conclusions other than shite got loud
But then i tried Art Welter's well done CD THD experiment/test/demo, and couldn't hear any differences. So now I'm a believer
Thx for the continued sharing..
Agreed. As I'd mentioned before perception of distortion at bass frequencies appears to happen only at fairly significant levels. And particularly when playing music.
Yes, but now you're talking about a faulty driver, and one would not usually need a THD test to detect that
.I use the subwoofer's HD at across its passband not only to compare it to other subwoofers of similar frequency response, but to also give me an idea of how much it can be pushed before distortion becomes noticeable, and if there's anything that can be done (better high-pass filtering, for example), to squeeze even more of it (not all of us can fill our rooms with multiple subwoofers
). Unfortunately it's not only the driver's construction that contributes to distortion - the alignment that it's used in, and how that alignment is implemented, will also contribute. A badly-designed TH alignment for example will do a great job of amplifying any distortion components produced by the driver.