Dynamic headroom questions

[Saurav]

I realize that hi-eff vs. non-hi-eff is almost a religious divide among speaker designers/hobbyists. My intention here isn't to start an argument on the subject, but to get the insights of the people who post here (who are among the most experienced and approachable designers I have come across anywhere).

So... the basic premise is that watts are cheap, so there is no need to sacrifice other attributes to increase the driver's efficiency. Looking at it another way, if you can lose a couple of dB efficiency and gain better frequency response/lower distortion/whatever, that's a no-brainer.

I've seen numbers which talk about a 24dB peak-average ratio for well recorded orchestral music. Assuming a medium/loud average listening level of around 80dB, that requires 104dB peaks. Assuming one is sitting about 2m from the speakers and isn't listening to line sources, that requires 110dB output from the speakers. Assuming an 85dB sensitive smallish midrange/woofer (to use a random mid/lowish number), that requires 25dB from the amplifier, which equates to about 256W.

Here's my question. It's fairly easy to get an amp that will produce 256W. But out of all the midranges/woofers available that produce 85dB when fed 1W power, how many will actually produce 110dB when fed 256W power? And scale linearly along the 'dynamic axis', i.e. produce 107dB with 128W, 104dB with 64W, etc. Also, what does a transient like this do to the distortion performance of these drivers?

Obviously, I've picked slightly exaggerated/worst-case numbers to illustrate my question, and no one with an 85dB woofer hopes to hit 104dB peaks. I'm just trying to understand how this is addressed in speaker designs. I've seen IB designs where on the surface the number of drivers/amplifier power seems outrageous, but it comes down to the same thing, right, having the necessary dynamic headroom. So shouldn't the same principles apply for the higher frequencies as well? Or is it not as important for non-subwoofer frequencies? Are there other considerations which make this irrelevant?

Thanks,
Saurav

[ThomasW]

Obviously, I've picked slightly exaggerated/worst-case numbers to illustrate my question, and no one with an 85dB woofer hopes to hit 104dB peaks

Actually doing that's not a problem if the driver is well designed, is operating in an appropriate passband and has enough Xmax.

You can model the situations you've described and see what the driver does with varying power levels. The catch 22 is that most of those are anechoic models. Adding in a baffle of course increases the output at specific frequencies.

Actually with IB's the recommendation is for much more displacement, given that the drivers are excursion limited not thermal limited as are drivers mounted in an enclosure

[Dennis H]

I'll let the poohbahs give you the right answer (edit, I see Thomas already did ) but here's my take. Those loud transients are usually so brief that the driver can handle them but, if you tried to play a continuous tone that loud, the driver would overheat and compress badly. About distortion, that's certainly a concern so 5" drivers don't cut it unless you cross them really high. It's not so much a matter of efficiency as size and how much air they can move without straining. That's why Jon likes those 8" midranges so much.

[Saurav]

Thanks for the response.

Actually with IB's the recommendation is for much more displacement, given that the drivers are excursion limited not thermal limited as are drivers mounted in an enclosure

Understood. You can hit thermal or excursion limits on the driver, or run out of power on the amp. I think those are the only (or main) 3 limits on the SPL capability of a system.

So what does the thermal compression behavior of real-world drivers look like? I'm not asking about the point where the driver melts, I'm asking about the point where the driver's performance stops being linear, so a 2x increase in power produces a < 3dB increase in output. And thermal-limited behavior would be more complex than excursion-limited behavior, right? For instance, reproducing a peak after the driver's been sitting playing nothing, vs. reproducing the same peak when the driver's been running fairly hard for the last several minutes?

Actually doing that's not a problem if the driver is well designed, is operating in an appropriate passband and has enough Xmax.

How does the math for this work? I think the Linkwitz website has a calculator for this, right? Xmax needed for a given SPL/frequency/driver size? If I can find it I'll try playing with that. Does his calculator only handle dipoles, or is it the same for a driver in a box?

[Saurav]

It's not so much a matter of efficiency as size and how much air they can move without straining.

But aren't the two related? A larger diameter driver needs less excursion to move the same amount of air (which will produce the same SPL level... correct?). Does the input power directly control the excursion of the driver? So "all else being equal", a larger driver should be more efficient. I have a feeling I'm way-oversimplifying this

[Dennis H]

Yep, if everything were equal (same motor and moving mass) the bigger cone would be more efficient. But bigger cones often have larger Xmax (less coil in the gap) and that decreases efficiency.

At a given frequency:
Input voltage is proportional to excursion.
Voltage is proportional to squareroot of power.
So it takes 4x the power to move the cone twice as far.

[Evil Twin]

Well, Saurav, that's why folks make big expensive speakers like some of the Wilson Audio models (X1, X2), the big Von Schweikerts, big Magneplanar's, and the varioius DIY line arrays, for exmple.

To reproduce well recorded classical music with that kind of peak to average ratio is challenging- as a classically trained musician, I'm quite familiar with what a grand piano by itself sounds like in a moderate size room, and almost no home speakers are capable of reproducing the way one sounds when you play it or are just a few feet away...

Now, another point to consider about the numbers your playing with up there- your assumptions for SPLs are correct if you're dealing with a point source radiator, but for line arrays and line source drivers, the rules are different and the fall off with distance is considerably less.

The only high efficiency drivers that will do these levels with ease, though, are horns- their acoustical impedance matching and loading of the diagphram makes possible a lot of sound without much diaphragm displacement.

High efficiency direct radiators like PHL's, some of the Audax cone drivers, etc, make other tradeoffs- generally, very little in the way of Xmax, so keeping them in the linear range usually means not pushing them very hard from 250 Hz on down.

Regarding your question, very few drivers will really handle 250 watts on any sustained basis. Fortunately for a multi-way speaker, it's usually the composite output from several driver's that's trying to hit that 104 or 107 dB SPL.

For example, in the case of the Saint-Saens dipole system I'm in the final design phases, the drivers are calculated for 107 dB output at 1 meter, at any single frequency- in the low midbass, due to dipole cancellation, this requires either 8 Extremis 6(7) per side, or 8 Peerless or Dayton 8" midwoffers. In the midrange, 600Hz to 4 kHz, an RD50. In the treble, 8 Fountek JP3's per side, which are about 97 dB/watt efficient, with 10-20 watts power handling each. Let's figure no more than 8 watts per Fountek, that's 8 Fountek ribbons each capable of playing at 106 dB anywhere from 5 kHz to 35 kHz. That ought to be able to blister paint on a Y-Wing Fighter at 10 meters.

This is where Hank and I agree fully, that 6" two way speakers are a waste of time except for nearfield computer use, and a standard 8" two way is OK for polite playback with peaks no higher than the upper 90's. Getting beyond that point is what really starts to incurr some series money and effort. If you look at the power distribution of music, you can cut some corners in the range above 250-400 Hz, but IMO distortion and clairty suffers unduly.

Compare the midband distortion products of a Vifa PLW18 against a Peerless HDS182 or SEAS W18- you'll start to understand why some drivers start to sound strained at moderate levels, and others don't give a clue until they thermally burn out or your ears do.

When I first had the X1 Klones running at ThomasW's, we were listening to a variety of music straight out from a Sony XA7ES CD player with volume control into an Aragon amplifier- after our first listening session we really appreciated how clean they were because of how much our ears were ringing becaase we kept turning it up and the system never complained. That's one way to get lifelike SPL's with fairly low distortion. And it's representative of what you have to do.




These are getting rebuilt into a smaller configuration with cleaner Peerless midwoofers replacing the Eton 7's, and Peerless RS270's on the bottom.

The Arvo's will do a fair amount of the same SPL, and stay quite clean- but they won't gut kick you in the bottom end.
But for anything approaching the SPL's you want, you will need a decent size three way at the least.

I don't mean an 8" woofer with a 5" midrange and dome tweeter...

[Saurav]

Well, horns are impractical for my current (and foreseeable future) living situation, so that's out. And in my limited experience, it's only the large front horns that have that sense of ease that I crave. I've heard some rear-horn-loaded single-driver speakers, and they fall fairly close to my "pretty Barbie doll music" category. Not as bad as a 5 1/4" monitor, but not a whole lot better either, especially with music that has dynamic bass.

*Very* loosely speaking, I guess I'm on a similar track as you, just scaled down to about 10% or so I have a single Audax midrange on an open baffle crossed at 500Hz, and a single JP-3.0 crossed at 3500Hz. The 500HZ point was picked based on music power distribution. All of this has 3W of amp power behind it, so right now that's the SPL bottleneck in my system. Below that is an Eminence Beta 12CX with 25 SS watts, which will keep up with the top half. Based on my calculations, this should be good for average SPL levels of about 75dB.

Some day I'll have the space to start thinking about line arrays and horns and really big dipoles I guess a simplistic view of this would be that the alternative to using high-efficiency drivers is to use lots of drivers.

[Evil Twin]

Some day I'll have the space to start thinking about line arrays and horns and really big dipoles I guess a simplistic view of this would be that the alternative to using high-efficiency drivers is to use lots of drivers.

Well, smaller line arrays or dipole speakers can deliver a fair amount of music in a fairly small footprint- think about ThomasW's current Arvos, for example, with udal TC2+ on the bottom end instead of the DPL12's (lower upper bass distortion), dual M8a, and the SS9800 tweeter. I don't think their floor footprint is appreciably greater than what you describe; they will handle 250 watts @ 8 ohms, considering both mids and woofers are wired in series. Sensitivity is about 85 dB/watt. They sound nice driven with just an Ayre V5, though a little more power is proabably better.

Is your current limitation room placement? How do you have your current system placed? Have you tried dipole bass? While I like the dipole effect also in the midrange up to 1 kHz or more, it's most remarkable application I believe is in the range below 300 Hz...

I.E., you may not know what you're missing...

[Saurav]

Ah yes, dipole bass, that's something to try one day too. My system is currently in my apartment living room. The system is set up along the long wall, so the speakers are pretty far apart, and a good 3' or more from any room boundary. Sub is monopole for now NHT 1259 in a sealed box that someone gave me for free. The room's probably 25'x20' and a little old lady lives on the other side of the wall, so this is enough bass for me for now. Though I know that dipole bass would improve quality, not quantity. For now all I have to work with is this sub and a BFD. I think ths results are OK. I can hear most of the notes of most walking bass lines, something I definitely could not do in the past.

I think it'll be a long time before I go back to high powered SS amps, if ever. Right now I'm firmly in the low powered tube amp camp. There's just something about the sound of a 2A3 (created mostly by the distortion spectrum, I would guess) that sounds really beautiful to me. At the same time, I heard Eidolons at CES 2 years ago driven by (I think) Ayre electronics, and that was a whole different kind of magic. So I'm not really totally set in stone in any design/philosophy camp. I'm mostly trying to learn whatever I can about what goes into the different design approaches - what is considered important vs. what is expendable, things like that. Hopefully getting a feel for both approaches can help me decide on what would work best for me, given my tastes/desires as well as my limitations.

Hope that makes sense

[Saurav]

Speaking of the Avros... I started with 5 1/4" monitors with pinpoint imaging that made very nice Barbie doll music (a phrase I read somewhere and fell in love with - 'It sounded like a band of Barbie doll musicians playing on a 6" high stage'. That describes so many speakers perfectly). I then built Adire HE12.1s (That's a 27" TV. Yes, I know I have a very understanding wife):



Then converted them into woofers and added an open baffle on top:



Which now looks like this in its latest incarnation:



The open baffle/dipole sound was an eye (ear?) opener. I'm not going back to boxes for the midrange. I could replace the lower box with a dipole frame and suitable drivers. That would get me dipole operation down to 80Hz (or wherever I decide to cross to the subwoofer at). Right now the woofer's handling 80Hz - 500Hz. So if I understand you right, what you're saying is that I would see a dramatic improvement if I did that.

The Eminence drivers in there now are 97dB, so I have a ~ 30W Gainclone driving them. Do you know of any drivers which can be used as dipoles from 80Hz to 500Hz and can be run off 30W of power?

Thanks a lot for all the help.