Subs & Group Delay


Any opinions as to the value of this spec vs. distortion specs. At any given price it seems that you can choose to optimze one or the other - or maybe compromisse a bit on each. Even the pricey JL 113 can't touch the $600 SVS for distortion at 95db (anechoic), but it shows about half the group delay. I'll try to audition both, but in the meantime, any opinions on the relative merits of these measures would be appreciated.

Marty
martykl
I'd go group delay first as it is no use having low distortion if there is a whole load of resonance or ringing from your sub - remember your room is going to resonate too! So less resonance or the sooner the sub stops ringing (overdamped) the better IMHO. Ideally you want both low group delay AND low distortion AND high SPL output but it is expensive to get all three...

a good rule of thumb is less than 25 ms group delay at 20 Hz This corresponds to 20 feet added distance between something you hear at 20 Hz and its upper harmonics at 80 Hz (which may only have a 2 msec delay). It is probably not audible at this level so bass will be nice and tight and sound integrated with the rest of the music (kick drum is real tangible and there). Delays of 50 msec are probably becoming audible (nearly a full cycle) but this is all hotly debated. The problem is that the tweeter sound of the kick drum beater hitting the skin may last a mere 2 msecs - so if the deepest sounds arrive delayed by 50 msec then it is going to sound like delayed room or concert hall reverberation rather than the instrument itself - so while this may be an added and pleasing effect on dinosaur footsteps it may cause the kick drum to sound distant and less "integrated"or lacking "punch". (You can look up Hass Effect to understand how our brain separate sounds that arrive much later - we simply interpret these late arrivals as being reverberant reflected sounds and NOT the primary sound.)

As a rule of thumb, ported subwoofers generally have twice the phase shift and twice the group delay of sealed subwoofers and oscillate twice as long after the signal stops. That is why I claim that "as a rule of thumb" sealed subs will be tighter and better sounding even if they have poor SPL output. (Of course a ported sub that is tuned way way low at 10 or even 5 HZ is likely to sound pretty damn good and close to sealed sub - so it is only a "rule of thumb")
Low-order harmonic distortion is benign. I'm aware of a controlled listening test in which 30% second harmonic distortion was statistically undetectable - so don't go nuts pursuing low harmonic distortion specs.

Group delay typical of a vented woofer system has been shown to be detectable on test tones, but to the best of my knowldege not conclusively so on music program material - thus the "hotly debated" reference in Shadorne's post.

In my opinion the in-room frequency response makes a bigger difference.

Duke
dealer/manufacturer
Hi Duke, you said, "In my opinion the in-room frequency response makes a bigger difference."

How close of a freq response do you think is "needed", for low freq say from 20 to 100hz to sound good? My thought is +- 3db with one to two peaks and nulls up to 5 to 6db will be close enough for good playback, in most rooms.
What are your thoughts on this?

Bob
I'm aware of a controlled listening test in which 30% second harmonic distortion was statistically undetectable

Second harmonic is fairly benign but 3rd harmonic is bad - so it depends on the type of distortion. The other issue is that 1% distortion at the 3rd harmonic of a 20 Hz fundamental will sound nearly equally as loud as the fundamental note itself due to your increased hearing sensitivity at 60 Hz compared to 20Hz (see Equal loudness contours). So in the end, a distorting subwoofer simply populates the bandwidth where your hearing is much more highly sensitive from 60 to 180 Hz with audible response. This is may be quite the opposite of your original intention when buying the sub - which was to augment the bottom octave rather than overwhelm the 60 to 180 Hz range.
Unless you listen almost exclusively to Baroque Organ, I can't imagine much 20hz content. For HT use, where the 'musical' properties of a sub are.....less needed......it may even be a non-starter when listening to 'effects'.
Also, if Crossed-over low enough, that mitigates the effect of 20hz 3rd harmonic. I cross over at between 40/45hz and even my vented HSU produces pretty tight stand up bass. So the amplitude of 3rd/5th harmonic 20hz is pretty well attenuated.
A couple of follow up questions.

1)An SVS 13" Ultra with its ports sealed shows app 5% THD at 20hz at 100db anechoic. In a real world listening room, that's presumably a crushing spl, so could one safely assume distortion is no issue?

2) The same sub stays below 20ms group delay at 20hz. Is there any benefit to looking for a lower # here? Also, is the goal in GD minimizing the value, or matching it to the gd of your main speaker near the x-over point? (Does that question even make sense?)

3)Magfan - how does a low x-over attenuate 3rd order harmonic distortion. I assume that the sub generates this distorted output as a by-product of its response to a 20hz inpu signal. It seems that the x-over point - which attenuates inputs signals - shouldn't affect this - Am I mistaken?

Thanks in advance.

Marty
1)An SVS 13" Ultra with its ports sealed shows app 5% THD at 20hz at 100db anechoic. In a real world listening room, that's presumably a crushing spl, so could one safely assume distortion is no issue?

It is pretty impressive yes but a far cry from a crushing SPL due to the fact we can barely hear 20 Hz. For comparison it would sound as loud as a 40 db SPL note at 1 Khz...a little above the rustling of leaves. The good news is that if your sub can achieve 115 db SPL then it suddenly becomes much more audible - as 120 db SPL will sound like 80 db SPL at 1 Khz - so clearly audible even above a conversation. Also there is not much at 20 Hz anyway so the biggest benefit is the lack of distortion at higher harmonics rather than any bone crushing sound.

The reality is that a great sub will NOT sound devastating on a 20 Hz pure tone at 100db SPL - is will be mostly inaudible if it is a good high quality sub!!!

Is there any benefit to looking for a lower # here?

probably not but lower is always better - it just may not be audibly perceptible.

I assume that the sub generates this distorted output as a by-product of its response to a 20hz input signal.

Exactly - in most subs playing a 20 Hz tone most of the sound you are actually hearing is distortion.
I've never bothered about the science, never had the time, when my ears were still good, but experimented happily using my ears and from what they told me, I'd agree completely with what Shadorne has said, namely the lower the better and even if you cannot --and shouldn't-- hear it,( if you did, it was somehow funny, not right) you could feel it.
Regarding group delay, this may be of interest: http://www.trueaudio.com/post_010.htm

I believe as Duke has said elsewhere, group delay reseach is lacking.
Shadorne,

Your point is taken regarding sensitivy vs. frequency - do you miss your "loudness" control?

However, when I run test sweeps on the SMS-1, the lowest audible tones generate a VERY loud rumble at 90db and up. Whatever frequencies we're talking about here (it appears to be app 25hz from the readout), it should be low enough to fall into the "insensitive" range of perception. This doesn't seem consistent with "rustling leaves" and I wonder what the explanation is. I'd also note that A0 on a piano (typically 27.5hz) registers as LOUD! well below the levels implied by your math (the room is not shaking). Any thoughts as to what I'm missing here?

Beyond this, my intended point re:crushing spl is that (it would intuitively seem that) the 100db anechoic test output should register much higher than 100db in a typical listening room. Obviously, there's lots of things going on that affect the change in output at low frequencies between anechoic and real world environments, but my assumption is that the same signal in a 15' x 20' room should be perceived as much louder than in the (quasi) anechoic measurement environment used by HTshack. Do you believe that this is a good assumption (at least below the primary cancellation frequency)?

Finally, do you have a sense as to which test sweep (they go in 5db increments from 90db to 115db) in those subwoofer tests might be the best proxy for LOUD listening - (music) in a real room. Obviously, there are too many variables for a precise answer, but, as a practical matter, there should be a reasonable answer (or range of answers) to that question. Just wondering which sweep you might look to in evaluating a sub.

Thanks in advance.

Marty
I'd also note that A0 on a piano (typically 27.5hz)
registers as LOUD!

You will hear the 2nd and higher harmonics of the piano loudly.

Now for the weird part.... you can completely remove the fundamental and play
just the harmonics and your ears will hear the correct note (but timbre will be
off)....hearing is a weird science!!!!

If you notice above I mentioned a pure 20 Hz tone or test tone (not a note from
an instrument as any instrument will have higher harmonics too)
but my assumption is that the same signal in a 15' x
20' room should be perceived as much louder than in the (quasi) anechoic
measurement environment used by HTshack. Do you believe that this is a
good assumption (at least below the primary cancellation frequency)?

Absolutely - you will get at least a 3 to 6 db boost and more or less at room
mode peaks/troughs.

Finally, do you have a sense as to which test sweep
(they go in 5db increments from 90db to 115db) in those subwoofer tests
might be the best proxy for LOUD listening - (music) in a real room.

I can't say but I think you need roughly 110 db SPL in the parking lot at 30 Hz
to convincingly represent peaks from real live instruments (like a drum set) in
a large sized room. This will have you covered for rock and dance club tracks
but classical can be handled with 10 to 15 db less. You can probably get away
with 100 db SPL or less at 20 Hz - as I think I mentioned I'd rather not have
20Hz at all if it was at all distorted.

One or two JL F113's would have it covered for most people in a large room
(if you ask around I think you will find several people own two of
these monsters and are very happy)
Hi Shadorne, I dont know where to go with my reply to your post above (six above). But I must say it has me "confused". Where you wrote,

"It is pretty impressive yes but a far cry from a crushing SPL due to the fact we can barely hear 20 Hz. For comparison it would sound as loud as a 40 db SPL note at 1 Khz...a little above the rustling of leaves. The good news is that if your sub can achieve 115 db SPL then it suddenly becomes much more audible - as 120 db SPL will sound like 80 db SPL at 1 Khz - so clearly audible even above a conversation. Also there is not much at 20 Hz anyway so the biggest benefit is the lack of distortion at higher harmonics rather than any bone crushing sound.

The reality is that a great sub will NOT sound devastating on a 20 Hz pure tone at 100db SPL - is will be mostly inaudible if it is a good high quality sub!!!"

I am actually having a hard time digesting this. If you would like to try re-explaining this to me I would be obliged. I am not sure how I want to counter to those statements. Perhaps we can dissect it at some point and discuss it that way. Not trying to be argumenative, but I do belive you need to perhaps rethink this.

Or perhaps I DO!

Bob
Shadorne,

Thanks again.

Your final suggestion goes to the heart of my questions:

1) I'm pretty confident that a pair of 113s will do it right (and look good doing so), but distortion specs are significantly better on the SVS ultra 13 (even with the ports sealed to minimize group delay). Will I hear this?

2) The SVS ofers an appealling 20ms Group Delay, but the JL stays below 10ms (for the most part). Will I hear this?

3) Will I hear improvement IN MY ROOM at MY LISTENING LEVEL with either choice versus my dual SPLR 800s which, honestly, don't seem remotely stressed when listening now.

4)If I spend $2500 for the pair of SVS subs, will I feel less silly than if I spend 3X that on pair of 113s only to find that the answer to #3, above, is "NO!".

5) If I can hear a difference, and the difference is distorion rather than group delay will I still feel silly because the cheaper SVS will get you virtually all of the distortion benefits and most of the group delay benefits at half the cost of the SVS Ultras?

You've been tremendously helpful, but I suspect I'm at the point where tossing it around further becomes pointless. I'll make my decision and report back.

Marty
I am actually having a hard time digesting this. If you
would like to try re-explaining this to me I would be obliged. I am not sure
how I want to counter to those statements.

My comments are entirely based on equal loudness contours. A deep bass
pure tone at 20 Hz (with no harmonics) is quite soft sounding at 100 db SPL
(in fact you may hear rattling of windows etc. louder than the pure tone - i.e.
you are more sensitive to harmonics or vibration induced by the 20 Hz tone).
Basically we hear 1 KHz tones with some 60 db better hearing sensitivity. As
you get louder and close to 130 db SPL at 20 Hz then the equal loudness
curves flatten significantly and that pure tone would sound as loud as 100 db
spl at 1Khz ( a mere 30 db drop in sensitivity as our hearing response
"flattens" as you go louder) As you can see from those curves - even a
tiny bit of harmonic distortion (say 5%) from your sub on a 20 Hz pure tone is likely to be more audible than the test tone itself - just because of our poor hearing sensitivity at 20Hz.
Hi Shadorne, Now I am with you! But I think the numbers and their implications are a bit exaggerated.

It looks to me you are using the Robinson Dadson cure as opposed to the Fletcher Munson. I see that at 80db the R-D curve is up 30db (not 40db) at 20hz while the F-M curve is only up 10db at 20hz. But even so with the R-D curve at 40db you are then looking at 50db up at 20hz, still not even close to the 60db that you mentioned to sound like rustling leaves.

I do think, now here we go, that the R-D curve is not as useful as the F-M curve. I believe the F-M curve to be the more accurate curve for our purposes. And yes thats good for me as now I do only have to be 10db up at 20hz as opposed to the 30db as suggested by R-D. Wow thats a big difference (20db). That is four times the difference between the two and the implications of power requiremens are huge. I will go with the F-M just because it is at least possibly obtainable by most audiophiles. And no I am not wimping out:)

Bob
Bob,

The general idea is that low frequencies are much less audible compared to midrange frequencies...with the biggest loss as you go down from 60 to 20 Hz. I may have exaggerated it some but it certainly "significant" whichever way you look at it.

The implication is that harmonic distortion in the 3rd and higher harmonic of a 20 Hz note may be detrimental - even in tiny amounts - making the sound louder than intended on the recording. See what Seigfried Linkwitz has to say - scroll to the bottom of the page.
This is a review of an important AES paper on subwoofers that appeared in Stereophile - The Science of Subwoofing.

Subwoofer
distortion guidelines are offered: Harmonic distortion will not be audible if the second harmonic is below 3%, the third around 1%, and higher harmonics no greater than 0.1–0.3%.

and

The authors sadly report that it is evidently believed in the loudspeaker industry that woofer distortion is not particularly important; as a result, there are very few drivers built that possess adequate linearity. All of this bodes ill for our chances of finding the ideal subwoofer from a commercial vendor.

For details see: Louis D. Fielder & Eric M. Benjamin, "Subwoofer performance for accurate reproduction of music", JAES, Vol. 36, Number 6, pp. 443 (1988).

Most of us are listening to distortion on our subwoofers...
Acoustat6, sorry this reply is late; I've been on the road and haven't logged in since, oh, some time last year.

Anyway, it's hard to say that a given frequency response specification is a guarantee of good sound. It really depends on how it was measured and what the actual data looks like. Consider this aspect of human hearing: The ear tends to integrate peaks and dips over intervals of perhaps 1/3 octave of so (it varies with frequency), so if there are offsetting narrow-band deep dips and sharp peaks close to one another, they might look disastrous to the eye but be completely overlooked by the ear. On the other hand a very broad but shallow (1 or 2 dB) bump or dip is easily overlooked by the eye, but may be quite audible.

An actual in-room response that falls within the numbers you mentioned is probably quite rare, and more likely than not sounds quite good, but it's still possible to have poor sounding bass that falls within those same specs.

Duke