Cone materials and 2/3-way designs

[Thooms]

Hi guys,

Since building my sub a couple of months back the bug has really bitten me (seems to be a common story lol) - I'm now looking to build a pair of speakers.

I'm still firmly in the throwing ideas around stage - there are currently a couple of issues I'm trying to figure out.

The first is that of cone materials - looking at sources such as Zaph's (excellent!) site I can see connections between materials and CSD / HD distortion - but how do these materials affect the sound? I don't want to cause bloodshed or open a can of worms by doing this, but do these two measurements tell the whole story?

Assuming effects such as breakup and response hiccups are either dealt with sufficiently in the crossover, would a metal cone sound the same as, say a paper cone?

Can the different constructions be mixed in the same design?

The other thing I was wondering about is 2-way vs 3 way designs. I keep hearing about how good a 3 way system is, that's always a tempting thing to read .

I'm torn between something like a Scan Speak or Seas Excel 2 way, or a Dayton/Seas/Peerless/SB Acoustics 3 way at a similar price. Instinct tells me that the 3 way would offer better performance (read: low distortion, better dynamics, positioning etc) than the 2 way. While it is a bit of a reductive argument, am I thinking about this in the right way? Can anyone recommend any good reads on the subject?

At the moment I haven't really got any firm ideas on component selection yet. As I'm a bit clueless with all this, I'm hoping you gurus out there could give me some pointers

I'm neglecting crossover considerations at this point. Currently, I'm hoping to build an active LR4 crossover network. I have enough bits to build 6 amps so an active 3 way is on the cards if that would be better.

For reference, whatever I build will be crossed over actively to my sub at some point, and will be used for probably 90% music / 10% film.

Hopefully this isn't one of those TL;DR posts - any answers/insight much appreciated. Cheers!

Edit: Just to say, I'm not deciding on actual designs/drivers at this point. Just trying to get the concepts & design methodology straight in my head first

[---k---]

Used correctly, Paper and Metal can both sound good. With distortion levels kept low enough to not be heard, with the exact same frequency response, they should theoretically sound the same. But, it is hard make things the exact same and usually you can hear some differences. People prefer both. There isn't a single "best". Only your ear listening to both can allow you to decide which you prefer.

3-way designs can offer better performance when budget is no limit. 3-ways can get expensive real fast, especially with metal cones that require more crossover parts. With lower budgets, a 2-way can be a better performer. It depends on your goals. Some would prefer to have one high quality driver than pairs of cheap drivers. Others prefer to have lots of dynamic power from several cheap drivers over a single high quality driver. Trade-offs.

3-ways designs are exponentially more difficult to design than a simple 2-way. It is generally not recommended to dive into the deep end of the pool for your first attempt.

[Face]

Be careful in your selection as some metal or mag cones have such a nasty breakup that you'll need either notches or steeper slopes.

[Thooms]

Thanks guys.

I'm not sure I really have enough intuition with these things to be able to pick crossover frequencies and where to put notches and stuff without taking measurements, so a measurement mic and matching preamp are probably first on the list.

That's a good point about the 3 way being harder to set up than the 2 way too. Do the 3 drivers have to be set up simultaneously, or can you set the level of the tweeter/mid drivers and then get those to gel with the woofer afterwards?

[JonMarsh]

Thanks guys.

I'm not sure I really have enough intuition with these things to be able to pick crossover frequencies and where to put notches and stuff without taking measurements, so a measurement mic and matching preamp are probably first on the list.

That's a good point about the 3 way being harder to set up than the 2 way too. Do the 3 drivers have to be set up simultaneously, or can you set the level of the tweeter/mid drivers and then get those to gel with the woofer afterwards?

Wouldn't advise taking that approach. This is part of what makes three ways harder- more things to simultaneously juggle.

Also, many folks take the outlook of trying to graph together what drivers they pick with some combination of crossover approaches with regards to slopes and level- sometimes you get lucky that way, but in my experience the best results have come when using a unified transfer function concept, like Steen Duelunds three way modeling froa single equation. My NeoD CC was the first project I used this approach and it turned out very well. search the site for discussion, and Google Steen Duelund. Having the drivers in the same relative phase over their working range does seem to have some benefits. But then, I'm the Duelund poster child on this site, I think.

There are many ways to approach this, we're not the speaker nazi's here, but if you look for advise, we'll try to point you towards the good stuff. Of course, there's more than one way to skin a cat.

[JonMarsh]

But if building a good two way is skinning a cat, doing a good three wY is more akin to teaching a grizzly how to dance- and something complex at that, like West Coast Swing!

[---k---]

I'll answer your question with a question. Take your current speakers and with the subwoofer off turn the up the crossover on your receiver to 120hz, 150hz and 200hz. Listen to how taking away the bass changes your perception of the whole speaker, including the upper end.

[Curt C]

I’ll throw in my $.02. –Well, with inflation, it is around $.15 now…

IME, Cone materials, and how they are implemented, as well as the rest of the driver design, can have an audible effect on the sound. Further, this audible difference will not necessarily show up on typical response and distortion plots. I suggest (this without any facts to back it up however) that some issues not easily measured are quite audible, while some other things we can easily measure, may not be all that important.

Take two midwoofers, say a SS 18W8531 Revelator and a Seas W18Excel: One is a slit paper cone, the other magnesium. Both have very low HD numbers. If one was to use DSP to provide identical transfer functions and passbands in their pistonic regions, I suggest they would almost assuredly still sound different.

Possible reasons might be:
Differences in the dispersion and magnitude of the HD spectra.
Differences in IM distortion
Cone damping and resolution. - Is the paper cone damping some fine detail, or is the metal cone passing along low level distortion guised as fine detail?

Since no driver is perfect and all drivers exhibit distortion to some degree, the amount, type, and dispersion characteristics of distortion of a particular driver can be expected to give it a particular flavor.

-Think of it as spice in the sauce: Some like lots of pepper, some prefer more garlic…

C





First one must assume that the goal of a speaker design is to reproduce the source material as accurately as possible (least amount of distortion) throughout its full pass band, and at some defined SPL and polar response. This might sound like an asinine premise, but all speakers are compromised in some manner or other.

[cjd]

Be careful in your selection as some metal or mag cones have such a nasty breakup that you'll need either notches or steeper slopes.

Softer cones without the clearly distinct break-up issues still have resonance problems, they're just absorbed naturally - along with who knows what other data you would rather hear. So you don't really avoid having to cross properly and manage resonances.

[Thooms]

Thanks Jon, I really enjoyed reading that Duelund article. I'm thinking of going for an LR4 system, but I haven't made my mind up yet. If I did use the Duelund approach, could I set the gain on the buffers in my crossover to a normalised setting for all drivers having measured their relative sensitivity in their operating range?

Would such measurements be best performed in an infinite baffle or in the actual enclosure/on the baffle they will be used?

I've had a play with the crossover on my system on several occasions Ryan. One of my main reasons for building it was in the hope of reducing IMD on my main speakers - hence my thinking a 3 way would have potentially higher performance than a 2 way constructed from equivalent drivers...

Curt - I suspected that that would be the case that there would be something that the usual spread of measurements may not (easily) reveal about the subtleties between different drivers. I'm intrigued to build something using non-metal drivers to compare it to my existing aluminium driver'd speakers.

I doubt I could hear something like a +/- 1db wiggle somewhere in a frequency response, but many people mention a particular 'sound' associated with say a PP cone compared with a metal one, even though it would make sense that if any nasty breakup were filtered out, the measurements suggest they would sound more or less the same...

That's an interesting point Chris - I hadn't thought of it like that before. It's difficult to know for sure about these things, particularly when you have to buy everything without actually hearing any of it...

[JonMarsh]

Go check out MarkK's and Sigried Linkwitz's measurements of drivers for ETC curves with shaped sine bursts at various test frequencies- this plus CSD testing holds a lot of clues for the differences in driver sound that aren't just SPL plots and nonlinear distortion/IM.

I've generally found that systems built with drivers with clean CSD profiles as well as low nonlinear distortion are a lot better at getting to the disappearing act point.

With regards to SPL response differences, 1 dB is bigger than you think for long term listening and early arrival perception. Used to work on phono preamp designs, using headphones, getting them down to the sub 0.25 dB deviation from RIAA, back in the 70's- the ear can't hear 0.25 db at one frequency, but put it over a range of octaves with pink noise, like an RIAA curve or speaker voicing, and you can reliably identify the difference. When you're seeking perfection, everything counts.

[Thooms]

Interesting stuff! I didn't know such a small difference was audible - makes sense to get it as close to perfect as possible though...

I didn't have much luck finding a link to the Linkwitz driver tests, although I did find Mark K's ones.

I'm starting to feel like I'm getting somewhere with figuring out what to go for now which is always a nice position to be in

[neuro]

Oh, don't worry. If you feel like you are getting somewhere, you are sure to be pleasantly lost if you do some more research .