Horn vs direct bass, how is it different?

[DrWho]

Oh, one other effect to consider...

Klippel has shown that thermal performance of a driver can be a significant limit to system linearity. In the horn, thermal management becomes a bigger issue. The small trapped air spaces around the driver make it difficult to expel heat. A classic example where this is a big deal is the Lab12 that benefits greatly from a heat exchanger.

I dunno, I've read and digested those IMD articles and run the numbers and I'm not convinced that's the end of the story. I have yet to hear a bass horn "destroy" a direct radiating subwoofer. It's only two octaves to cover for most subwoofer applications (20Hz to 80Hz), which is a lot different than trying to run 500Hz to 20kHz on a tweeter...

I think it makes way more sense for the MF unit to cover 80Hz to 1kHz, relieving an octave for the tweeter, and then going direct radiator from 80Hz and lower - trying to maximize the loading offerred by the room. This also lets you get away from a 2" compression driver, which all of them have issues when you get above 8kHz. Going to 1" moves all that up an octave getting you clean to 16kHz for the same cone stiffness / motor design. Then if you want beamy highs, you could do that in the horn and get better LF impedance matching - or you could keep the wider HF polars and stress the bottom end of the tweeter a bit more...

[Edgar]

Do you remember what the Jubilee compression is? The Sd of that 12" is probably around 100 sq in, and the throat is what, 2x6? That'd put you at around 8'ish.

I have, from the T/S parameters, the K31 with an Sd of 530 cm² or 82 in² each; from my Hornresp model the throats are a combined 580 cm² or 90 in². So the CR = 2*530/580 = 1.83!

Four 15's would be around 600 sq in total, versus the excursion equivalent of 1600 from the Jubilee.

It works out to 2*82*1.83 = 300 excurson equivalent for the Jubilee.

Greg

Edit: And the KHorn works out to about 450 excurson equivalent.

[PrestonTom]

Do you remember what the Jubilee compression is? The Sd of that 12" is probably around 100 sq in, and the throat is what, 2x6? That'd put you at around 8'ish.

I have, from the T/S parameters, the K31 with an Sd of 530 cm² or 82 in² each; from my Hornresp model the throats are a combined 580 cm² or 90 in². So the CR = 2*530/580 = 1.83!

Four 15's would be around 600 sq in total, versus the excursion equivalent of 1600 from the Jubilee.

It works out to 2*82*1.83 = 300 excurson equivalent for the Jubilee.

Greg

Greg is correct, but he beat me to it. I am one of these odd folks that have memorized the Jubilee specs.

-Tom

[DrWho]

Thanks for running the numbers...I'm traveling on business at the
moment and didn't have my notes in front of me - and my memory is
horrendous. I was also trying to be generous in favor of the Jub LF by giving the driver more surface area and higher compression...

Edit: And the KHorn works out to about 450 excurson equivalent.

So the 36" driver quoted from PWK will be pushing over 900 sq in....

Maybe
PWK was trying to compare the Khorn in 1/8 space to a 36" driver
equivalent in 1/2 space? I do know that PWK was (rightly) quite adamant
about the benefits of 1/8 space and coupling to it. I think it's
interesting to note that horn modelling takes the 1/8 space into
account, whereas T/S for direct radiators is a 1/2 space thing. Maybe that's where the extra 6dB comes in? Or maybe its the lack of DI modelling in the T/S model (which is total power response)?

I
need to find my notes about the excursion relationship to compression
ratio....I can't for the life of me remember the author, or I would know
exactly where to find it online. It was the same author that first
wrote about horn throat distortion. I'm curious if I'm remembering
something wrong, but I'm pretty sure the compression ratio = the
excursion reduction for a perfectly matched horn. Maybe one of you guys can find that somewhere.

One last comment...

I
just wanted to mention that as an engineer, I want the horn to come out
on top. To me it makes sense to control as well as possible the
impedance match of the driver to the air. That is such a fundamental
concept across all things engineering and is such a core essence of a
good approach to operate devices in their sweet spots.

My struggle is
that we're not starting with physical models of the space we're trying
to energize. At high frequencies it makes sense to ignore the space
because it is immediately a far-field, large room, multi-path reflection
thing (the energy arriving at the speaker isn't affect the speaker's output). The electrical parallel in my mind is that of wireless
transmission. In the midbass region, we start dealing with transmission
lines where we're looking at the modal response of the room and there
isn't enough multi-path to prevent deep nulls and narrow peaks. If we
can't change the room (which is usually the case), then the only way to
address that is by controlling how the signal gets injected into that
space. This is really the acoustic parallel of dealing with reflections
along a trace on a PCB - or maybe a better analogy is distribution of HF
energy across a ground plane. One thing I'm curious about is whether or not the modal response of the room is affecting the speaker's output. We do see reflections in this region affecting the measured electrical impedance... Then at the lowest frequencies, we're actually
pressurizing the room....which means we're in the realm of voltage
transmission - where a low output impedance is driving a high input
impedance. When doing voltage transmission, it actually hurts you to match the input/output impedance.

So to sum up:

Why do we want to "impedance match" to the pressure region of the room? Or what is the ideal "match"?

In the midbass,how does the horn address the room's own modal response? Is it the best option available? How does the room affect the variuos approaches available?

What
is the best device for providing a seamless transition between each of
the three regions? (pressure mode, modal response, radiation pattern)

I'm
really struggling on how the horn is ideal for all of these. At the
high frequencies, the more sound we keep off the walls, the more it will
sound like the midbass region (less reflections). In the mid-bass,
wider polars from a speaker away from the front walls is better because
it gives us more late arriving sound (behaving more like the higher
frequencies). At the very low frequencies, you're stuck with pressure
gain unless you absorb that pressurizing energy. A low output impedance
device will be better at energizing a pressurization zone - which means
no horns for the bottom octave or two. And if you're really worried
about the IMD products of direct radiators - then just get a bunch more
drivers until you get caught up....and reap the benefits of being able
to better tap the room's modal behavior....

Now if we're
listening outdoors in a big concert....then horns all the way. The only
reason touring sound companies aren't hornloaded all the way down is
because amplifier power is cheap, the distortion is part of the target
sound (it's "music" creation), and most importantly: it's a business and
it's way more cost-effective to run multiple DR boxes, which mostly
comes down to quickly packing everything into the trailer and requiring
less trailer space. Redundancy is also easier to achieve. Losing 1/32 boxes is not a significant SPL hit.

Btw,
I hope you guys jump in on the discussion here as I'm not a math expert
by any means and I would warmly welcome discussion on these topics. I
would love to know where these arguments start to break down. For example, I learned a
lot through the previous IMD discussions that Chris linked to here:
basically concluding that perfect amplitude linearity still has a
minimum IMD behavior due to the doppler effect.

[Audio Android]

From outside, it sounds like distant thunder, just a low rumble.

....I can only imagine what it sounds like from inside...[li]...Those are seriously impressive looking....... Thanks for the pics....When you posted that your sub didn't have a big footprint I was totally confused. I didn't realize they were in your wall using the attic as a it's enclosure.....Thats an insane setup, wish I lived closer!!!!...........Thanks again....CECAA......I will check that link out.... infinitely baffled.....Thank you Edgar!!

[DizRotus]

"Thanks again....CECAA......I will check that link out.... infinitely baffled....."

Puns are said to be the lowest form of humor. In this discussion of bass you can't go low enough. I love it.

[Edgar]

Maybe
PWK was trying to compare the Khorn in 1/8 space to a 36" driver
equivalent in 1/2 space? I do know that PWK was (rightly) quite adamant
about the benefits of 1/8 space and coupling to it.

I suspect that this is the case. PWK at times would expound about the benefits of horn-loading, at other times about the benefits of corner-loading. Without context it's difficult to know to which he referred here.

I
need to find my notes about the excursion relationship to compression
ratio....I can't for the life of me remember the author, or I would know
exactly where to find it online. It was the same author that first
wrote about horn throat distortion. I'm curious if I'm remembering
something wrong, but I'm pretty sure the compression ratio = the
excursion reduction for a perfectly matched horn.

It may be that "perfectly matched horn" that makes all the difference in the world. Evidently it's just not as simple as compression ratio, throat area, or mouth area; it's a combination of them all. For any given driver there is an optimum throat size (optimum from the standpoint of maximum SPL output), but then everything just goes downhill from there if the horn isn't the proper shape, size, or length.

I
just wanted to mention that as an engineer, I want the horn to come out
on top. To me it makes sense to control as well as possible the
impedance match of the driver to the air. That is such a fundamental
concept across all things engineering and is such a core essence of a
good approach to operate devices in their sweet spots.

Indeed, but the concept of the perfect impedance match is compromised by the large bandwidth that we expect from these devices. If one thinks of antenna design as a similar impedance matching task, then the resulting antenna efficiency can be MUCH higher than our acoustic impedance-matching devices because the antenna's bandwidth can be so much smaller.

My struggle is
that we're not starting with physical models of the space we're trying
to energize. At high frequencies it makes sense to ignore the space
because it is immediately a far-field, large room, multi-path reflection
thing (the energy arriving at the speaker isn't affect the speaker's output).

Yes, and we actually have two strikes against us in that respect. At high frequencies not only is the space in the far-field, but the dimensions of the transducer are a significant fraction of the wavelength so the impedance mismatch is not so bad. At low frequencies the space becomes part of the horn, and the dimensions of the transducer are so small relative to the wavelength that the impedance mismatch is very bad.

Then at the lowest frequencies, we're actually
pressurizing the room....which means we're in the realm of voltage
transmission - where a low output impedance is driving a high input
impedance. When doing voltage transmission, it actually hurts you to match the input/output impedance.

That's an interesting observation, and a viewpoint that I had not considered previously. I'm just not sure how to translate it into a woofer design strategy!

Greg

[Chris A]

Would you not agree that there is often correlation between maximum output and output linearity?

I suppose, although I'd quickly point out that freedom from distortion (of all types) is definitely an overriding subject (at least for me) and that would include driver compression distortion, which is essentially the same issue that you point out. But don't forget modulation distortion (FM and AM) and harmonic distortion, among others. Second, I'd point out that directivity control in small rooms is really critical, for the reasons that we've discussed earlier, including listener tests.

However, what's misleading with those numbers is that the Jubilee isn't perfectly impedance matched....and for me personally, those resulting resonances are not acceptable.

Which resonances are you referring to? If you are referring to the peaks in the KPT-KHJ-LF bass bin frequency response, then I'd point out that there is a strong correlation to the undersized mouth of the Jub bass bin and those peaks at ~95, ~110, and ~170 Hz, which are largely correctable using an active crossover (which I've been doing for some time now). If you've not heard the Jubs in full corners with the low-Q response peaks corrected in a real room, you're in for a treat (IMHO). I don't hear any other "resonances" in the pass band that I use (~40 to 450 Hz). You're going to have to cough up the data on this one, I'm afraid. "In God We Trust--Everyone Else Bring Data" []

The dominant difference in my opinion is the polar lobing - which is "bad" for both the Jub LF and the Quad 15's.

Polars in the bass frequencies, below about 400 Hz, typically aren't pronounced for the Jub bass bin (see http://forums.klipsch.com/forums/p/159501/1688024.aspx#1688024), as seen in the polar plots of the bass bin and also more importantly not as objectionable from a human hearing standpoint vis-a-vis midrange and HF.

I also think the reason that the Jub LF sounded more dynamic is because we were listening to source material that benefited from the peaking/resonant response of the Jub LF bass bin.

I hear the source material that we have heard in my room, with those "resonances" (actually fairly broad response peaks - NOT resonances) tamed - and IMHO, it is even more dynamic than you or I heard in the Hope listening room.

I can find quality source material that sounds horrible on the Jub LF that will sound much better on the Quad 15's. Heck, my single 15"at home can sound better (listening at low enough levels that the IMD is buried beneath the noise floor). You can't avoid the resonances from an undersized horn.

And you can't avoid the 15-25 dB higher modulation distortion of direct radiators (predominantly AM distortion for bass frequencies). That's a factor of 15-300 higher distortion on amplitude. I don't believe that you are going to correct for that much difference in IMD using very-high-priced drivers, but you might be able to close some of the gap.

I would also suggest that the idea of pressure vessel gain will impedance match better to a direct radiator, whereas a horn (especially tapped) will see little to no benefit.

If you are talking about pressurization field effects, then you are definitely in the realm of "cabin gain" (as discussed elsewhere). I'm pretty sure that horns will still see gain in real rooms-small rooms that are found in homes. I've never heard direct radiators placed in corners of a room perform like exponentially loaded bass horns - such as the Jub bass bin. I'd like to know if someone has tried this and succeeded in fooling the participants into thinking they were listening to horn-loaded bass. I'd also like to hear it for myself (not that I don't trust you, Mike...[]).

Chris

[DrWho]

I suppose, although I'd quickly point out that freedom from distortion (of all types) is definitely an overriding subject (at least for me) and that would include driver compression distortion, which is essentially the same issue that you point out. But don't forget modulation distortion (FM and AM) and harmonic distortion, among others. Second, I'd point out that directivity control in small rooms is really critical, for the reasons that we've discussed earlier, including listener tests.

I think modern drivers are getting to the point where the major difference is going to be Doppler generated IMD products depending on the application.

Using the IMD formula from Beers and Belar, I get the following for the JBL 2226 15" driver, which will be 3dB down at 40Hz:

Excursion for 100dB at 1m in 1/8 space is 0.02 inches peak one-way, so 0.04 inches total. Figure the same 600Hz xover as the "well designed horn" and you get:

IMD% = 0.033 * 0.04 * 600 =0.8%

The horn is measuring 0.7% for 100dB at 2ft.

That's pretty darn close in my opinion. Now if we're talking 1/2 space measurements, then the measurement shouldn't be taken at 2ft. The minimum distance for far-field measurements would be closer to 10ft for the Khorn. In other words, at 2ft you won't see much difference between 1/2 space and 1/8 space (been there done that with other speakers already - and Danley has discussed this as well).

Which resonances are you referring to? If you are referring to the peaks in the KPT-KHJ-LF bass bin frequency response, then I'd point out that there is a strong correlation to the undersized mouth of the Jub bass bin and those peaks at ~95, ~110, and ~170 Hz, which are largely correctable using an active crossover (which I've been doing for some time now). If you've not heard the Jubs in full corners with the low-Q response peaks corrected in a real room, you're in for a treat (IMHO). I don't hear any other "resonances" in the pass band that I use (~40 to 450 Hz). You're going to have to cough up the data on this one, I'm afraid.

Let's pick on the 170Hz peak....let's say you EQ down 170Hz to measure flat. What happens when the driver is asked to play an 85Hz signal and creates a second harmonic at 170Hz? That EQ ain't doing you any good. Same applies for any IMD product, or mechanical vibration, or non-coherent wave triggering those resonances. In other words, the EQ only works when the system is perfectly linear. And this isn't a case of the horn being orders of magnitude more linear than the direct radiators either.

Btw, I spent a long time EQ'ing the Jubilees at Kovats' place (who has a very well behaved model response in his room). I'm convinced those resonances add to the sonic signature of the bass bin. I've heard it everytime I've listened to my source material. Btw, if you're using music that doesn't have energy in those problematic passbands, then of course you won't hear it...

Do you remember comparing a Jub LF to the Lascala LF and how the Lascala had a bit of that cavernous sound to it? (dunno how else to describe it). Roy said it was dominated by the fact that the area expansion wasn't perfect (there's a section of no expansion on that horn). The Jub LF ain't a perfect expansion either. The only time I've heard those resonant / cavern sounds go away is with straight horns....

Would you be willing to capture an impulse response of your system that we can use to convolve with music? You can create a file using your source material and I can create a file using material that I think exhibits the effect. Then you playback the original music and compare it against the convolution while listening on headphones. This is an incredibly effective way to identify the sonic signature of a system (as capture by the measurement microphone).

I can find quality source material that sounds horrible on the Jub LF that will sound much better on the Quad 15's. Heck, my single 15"at home can sound better (listening at low enough levels that the IMD is buried beneath the noise floor). You can't avoid the resonances from an undersized horn.

And you can't avoid the 15-25 dB higher modulation distortion of direct radiators (predominantly AM distortion for bass frequencies). That's a factor of 15-300 higher distortion on amplitude. I don't believe that you are going to correct for that much difference in IMD using very-high-priced drivers, but you might be able to close some of the gap.

I don't agree with the premise here that the higher distortion is the reason for the perceived difference. I also don't think there should be linear weightings for the various types of artifacts. It's also highly source material dependant as to which artifacts play a higher role.

I'm pretty sure that horns will still see gain in real rooms-small rooms that are found in homes. I've never heard direct radiators placed in corners of a room perform like exponentially loaded bass horns - such as the Jub bass bin. I'd like to know if someone has tried this and succeeded in fooling the participants into thinking they were listening to horn-loaded bass. I'd also like to hear it for myself (not that I don't trust you, Mike...).

The tapped horns I've measured don't exhibit any gain difference between in-room and outdoor measurements. There is some difference to a bass horn when its undersized, but it's usually more of a flattening of the passband than actual increased output at the peaks.

I would publish the data, but I'm usually striving towards an end result when taking measurements so I rarely save the incremental steps. If I remember during my next round, then I'll save a few comparisons.

[ClaudeJ1]

Even Keele's article on horns vs. vented enclosures was not comparing sound quality, but rather relative volumes--a subject that I'm not really not that interested in (...but the "small speaker crowd" IS interested). Quality of sound usually is overlooked in these type of discussions.

Would you not agree that there is often correlation between maximum output and output linearity?

For example, a single driver with X radiating surface area requires a certain excursion to achieve its maximum SPL. If another system is able to exceed or match that maximum SPL, then is not the excursion less? Or rather, is it not easier for that louder system to be linear at the lower SPL requirement?

.

According to Danley (and many more) all moving coil drivers have a non-linear power compression in their output at 10% of their rated wattage input. Since most horns increase output by 16 db or more (a factor of 40) for a given electrical input, one can see there's a huge benefit there.

Although, I must admit that a boatload of DR 18's with kilowatts of power would be very impressive to FEEL.

[DrWho]

According to Danley (and many more) all moving
coil drivers have a non-linear power compression in their output at 10%
of their rated wattage input.

So what is the practical outcome of statements like that without quantified description of real behavior?

For
example, I can say that all speakers regardless of their construction
(so including horns) exhibit non-linear power compression at 0.01% of
their output because you will never find any transducer with 0% THD at
any output level. How does that fact change the behavior of the
designer, or the end user? It's not like we can go buy or build
something with no distortion...

Btw, I also think it
is a bit misleading to think about horns as free gain on top of a
driver. The horn is providing an impedance match to a system that can
transfer at most 50% of the consumed energy. In other words, there is a
cap to this "free" gain. A perfect 50% efficient point source with a
90x90 polar pattern will be 115dB with 1W at 1m.

The Jubilee (a
good horn) is claimed to be 108dB with 1W at 1m, which is 20% of the
ideal. A 96dB driver put into a corner gives you a 90x90 polar pattern
and effectively 102dB with 1w at 1m. That's roughly 5% of the ideal, or
'only' four times less energy transfer as the horn. This means you need
4x more power dissipation in a direct radiator solution to keep the
power compression the same.

A very straightforward overkill way to
achieve that would be to have four times as many motor structures,
which would give you the same thermal performance for the same output
level. However, that actually gives you another 6dB of efficiency and
makes the rear volume four times as large - so really you're looking at
1/4 the power dissipation (in other words, you're 4x better than the
horn in terms of power handling). If we don't need that extra 6dB of
output, then we could lower the sensitivity of each driver and have the
same total box volume for the same power compression.

Btw, these
are all strictly power in / power out numbers - this doesn't address the
difference between doppler distortion of the two approaches.

I
think we could get the same 108dB sensitivity from eight twelves in an
enclosure with a size similar to the Jub LF. I also think the eight
twelves could be arranged in such a way so as to achieve less polar
lobing at 1kHz. The only potential remaining advantage to the horn
loaded LF is a reduction in doppler distortion. My whole supposition
thus far has been that the doppler distortion differences should be
relatively small in the grand scale of things, and that there are other
compromises from the folded horn that outweigh that potential advantage.

Btw,
this only works so closely for the 1/8 space scenario where the room
corner is basically providing a lot of what a horn would be doing. Once
you go to an environment with less/smaller boundaries, then the
difference starts to spread (although Keele still claims the math works
in favor of the direct radiator, but at the expense of polars according
to Roy). At high frequencies, the room corner doesn't meet the 1/8 space
criteria so the horn becomes that much more important. It's also harder
to get multiple transducers to sum coherently when the drivers are much
larger than the wavelengths being created. At lower frequencies, the
wavelengths are so large that coherency is easier to achieve.

[ClaudeJ1]

Well, Doc,

my point was strictly power compression on a Voice Coil at 10 percent of rated power. So a 200 Watt woofer would show power compression at 20W. So it seems to me that a bass horn would prevent this from happening since it would use way less than 10 percent of VC power rating to make music.

[DrWho]

Well, Doc,

my point was strictly power compression on a Voice Coil at 10 percent of rated power. So a 200 Watt woofer would show power compression at 20W. So it seems to me that a bass horn would prevent this from happening since it would use way less than 10 percent of VC power rating to make music.

First, I disagree with the blanket 10% power compression claim - and there is plenty of Klippel data showing otherwise.

But let's take a driver with compression at 10%...

In a room, putting a horn in front of it is only ~6dB better. The other alternative is to double the number of drivers and you get the same power compression with a smaller cabinet.

[CECAA850]

Fascinating posts Michael, thanks.

[ClaudeJ1]

Well, Doc,

my point was strictly power compression on a Voice Coil at 10 percent of rated power. So a 200 Watt woofer would show power compression at 20W. So it seems to me that a bass horn would prevent this from happening since it would use way less than 10 percent of VC power rating to make music.

First, I disagree with the blanket 10% power compression claim - and there is plenty of Klippel data showing otherwise.

But let's take a driver with compression at 10%...

In a room, putting a horn in front of it is only ~6dB better. The other alternative is to double the number of drivers and you get the same power compression with a smaller cabinet.

Only 6 db? How come a Lab 12 driver specs out at about 89 db/watt efficiency, but when you put two of them in a LabHorn in parallel you get 108 db output? That's 15 db better, not 6, which is probably what you get in addition to the 15 when you fire it into a corner. Please explain.

[CECAA850]

7.5dB per driver?

[CECAA850]

I'd also think you'd get more output from a TH than a FLH. Michael may have been referring to a conventional horn.

[DrWho]

Well, Doc,

my point was strictly power compression on a Voice Coil at 10 percent of
rated power. So a 200 Watt woofer would show power compression at 20W.
So it seems to me that a bass horn would prevent this from happening
since it would use way less than 10 percent of VC power rating to make
music.

First, I disagree with the blanket 10% power compression claim - and there is plenty of Klippel data showing otherwise.

But let's take a driver with compression at 10%...

In a room,
putting a horn in front of it is only ~6dB better. The other
alternative is to double the number of drivers and you get the same
power compression with a smaller cabinet.

Only 6 db? How come a Lab 12 driver specs out at about 89 db/watt
efficiency, but when you put two of them in a LabHorn in parallel you
get 108 db output? That's 15 db better, not 6, which is probably what
you get in addition to the 15 when you fire it into a corner. Please
explain.

What is the sensitivity of a Heresy which uses a
single 12" driver? Or how about the Chorus II which uses a single 15"
driver? (99dB and 102dB claimed with whatever standard Klipsch uses).

Clearly
the Lab12 driver is not a good candidate for comparison here. Or to put
it another way, there are other more efficient 12" drivers available.
And doubling up on drivers gives you 6dB more each time.

What
will probably blow your mind is that going to a higher efficiency 12"
driver in your Lab Horn will not improve its sensitivity any
further....it will also require a much larger rear volume, and probably
have a more jagged frequency response. In other words, the Lab12 was
designed for horn loading, whereas something like the Delta 12A was
designed for a vented enclosure:
http://www.eminence.com/pdf/Delta_Pro_12A.pdf

[ClaudeJ1]

The reason I use the LAB 12 driver as an example is that the LAB horn uses two of them so if we add 6db gain for the extra driver before the 1/4 wave horn loading, there is still a 95 to 108 db difference in output. Wouldn't a 13 db gain in efficiency via horn loading reduce the power input requirement for a given Sound Pressure Level?

If that is true then my point about avoiding thermal compression by keeping the power 10% or less of rated driver wattage would likely occur more frequently with horn loading, espeically in a home environment where you can fire said LAB horn into a corner and get flat response to 30 hz.

[DrWho]

My point is that if you use more drivers instead of a horn, that you also end up with less power dissipated per driver.

There's also the benefit of a smaller solution that avoids the compromises of an undersized horn, and a direct radiator system also benefits more from corner loading and room gain than a horn will (because the horn is already providing the DI).