Building the best loudspeakers

[ShinOBIWAN]

Originally posted by ScottG
While I could be wrong, I don't think you would find that problem with the neo 3 pdr and the Ushers and a well "sorted out" crossover. (Note that it doesn't have to be 2 kHz - THAT was just what Dennis had applied for his "plop-in-the-box" part deux.) In particular the Ushers (excepting that resonance at 6+k that needs to be notched out), are not terribly extended. As such midbass combing should be low with an LR - 6db base lowpass. Moreover the neo drivers are strange beasties, lots of emmisive area and a small enclosure volumes allow you the ability to cross at *oddly* low freq.s - so long as you can sum flat with the midbass drivers. Even 1.5 should be OK with these drivers - its just a matter of summing with the midbass then (..and excepting a certain loss in ultimate spl). (..the things the Neo 8 will do with a small chamber are just plain WEIRD.)

That's the problem, I don't have direct experience with that driver so its tough to know the quirks. On paper I wouldn't recommend that 1.5K cross using the Neo 3 but from what you say it might just work. The Seas seems an easy choice because its clear that its got output at that level and I'd imagine distortion to be good since its based on the well regarded 27/TBGFG. Probably will be more excursion limited at that crossover point rather than distortion limited. The loading from the wave guide will help slightly as well.

I'd strongly disagree about avoiding issues with going digital on the crossover. Number one: complexity and learning. Being able to make all sorts of adjustments "on the fly" allows for so much more flexibility, speed, and ease of learning with respect to prototyping a given design. Its also a LOT less expensive (unless you hit upon something outstanding to begin with). Number two: a passive crossover is another resonant load on the amplifier - it will play a substantial role in the overall sound of the finished design without respect to its intended purpose.

Yes you could argue that from an overall standpoint passive crossover require more knowledge. You need to understand what components values and circuit topology will get you the intended results. But that's still not enough because the intended results also need to be founded on good knowledge of accurately affecting the minimum phase behaviour of the individual drivers to form a coherent loudspeaker system - this is universal for either passive or digital crossovers. For digital systems you need not concern yourself with how the crossovers are made because they're sat right there as a programmed transfer function but you do need that other layer of design knowledge that's applicable to any crossover implementation. In other words its misleading to word it so that it appears to avoid crossover issue, I do now understand you meant its just an easier and quicker path rather than my mistake of thinking you were hinting that it allows to disregard established crossover pitfalls.

I'd say that digital active or any active system can cost more. The real cost is the extra number of amps required. The better digital crossovers also need investment in DACs and processing, this can amount to large sums of money - for example the DEQX at $3000 or a PC with high quality sound interface and suitable software for around the same amount.

[vaughn]

I am wondering what you guys think of this design;


http://www.lonesaguaro.com/speakers/...7/Cryolite.htm

I built a pair about a year ago and have been listening to them since. I like them quite a bit
but do feel the crossover probably needs more work.

The reason I am suggesting them is it seems the direction the thread was going and would
possibly satisfy the op's needs.

[Ceibal]

Originally posted by ScottG
Because I thought it was a bit hinky with respect to volume - I went back and actually *looked* at the internal volume of that cabinet..

Net internal = 1/2 cubic foot.

At *that* point you might as well have a sealed system. Because with 2 drivers around 7 inches in diameter you are simply not going to get much bass near the reference level regardless of the design.

*IF* you actually had about a cubic foot internal, well then there would be something more to "play" with.

To get more bass look at the offerings in the 5-5.5 inch driver range, OR go back to the *non*-curved side panel box.

i.e THIS is the box that has 1 cubic foot internal:

http://www.partsexpress.com/pe/showd...number=302-750

I am glad you checked that. I liked the design but didn't bother to check and make sure the internal volume was identical to the non-curved version.

[Jay_WJ]

The info on the PE page is incorrect. Its internal volume is 1.0 cu ft. And the depth of the box dimensions is also incorrect. It's actually a bit deeper than the non-curved one. PE should've done better for these product pages.

[Jay_WJ]

I've designed MTM with the Usher 8945P and 8945A:

http://www.geocities.com/woove99/Spk.../Usher_MTM.htm
http://www.geocities.com/woove99/Spk...A_versions.htm

The T/S parameters of the P and A versions are virtually identical. For dual woofers, 1.0 cu ft is a minimum volume for a vented design. Ideally, if you want to have low tuning with a good amount of bass, you'll need 1.5 to 1.7 cu ft.

If you want to use a subwoofer, you can use a .75 to 1.0 cu ft sealed box for a -3 dB point of 71 to 74 Hz with a nice 2nd order rolloff. This will be perfect for use with a subwoofer with 60 to 80 Hz crossover setting on an HT receiver.

If you like warm and tuneful midrange, the Scanspeak paper cone or the Usher carbon/fiber cone may fit your taste better. In fact, these are the lowest distortion 7" drivers you can get currently. If you like to feel a touch of detail in the upper midrange, a bit higher odd order harmonic distortions from metal cone drivers can do a trick for you. I don't like this, but you may like distortions!

The Neo3PDR is the best non-dome tweeter according to Zaph's test. Its distortion profile is excellent. No doubt 2 kHz or even 1.5 kHz LR4 may be possible. But according to its frequency response (see below), 1.7 to 1.8 kHz LR4 should be the lowest possible point for this tweeter.



A low xover point is preferred for an MTM design. The lower, the better. I'd go with the Seas or the Peerless 810921 tweeter (as in my design) if you want an MTM. The Neo3PDR should be an excellent choice for a TMM 2.5 way design.

[Ceibal]

If I am doing the math right, the Scanspeak is the right woofer for a 1 cubic foot sealed enclosure. Here are the QTC's.

Scanspeak 18w/8535 = .71

Usher 8945A = .58

My understanding is that you want to be just below a QTC of 1.

Am I correct?

[ScottG]

Originally posted by ShinOBIWAN

In other words its misleading to word it so that it appears to avoid crossover issue, I do now understand you meant its just an easier and quicker path rather than my mistake of thinking you were hinting that it allows to disregard established crossover pitfalls.

I'd say that digital active or any active system can cost more. The real cost is the extra number of amps required. The better digital crossovers also need investment in DACs and processing, this can amount to large sums of money - for example the DEQX at $3000 or a PC with high quality sound interface and suitable software for around the same amount.

Absolutely, it not a means to avoid the crossover issue.. What I had written:

"..you could skip a lot of the complexity surrounding crossovers by using an active digital crossover (and more amplifiers). (..and of course there are compromises there as well.)"

(-perhaps it was over-broad. )

Originally posted by Jay_WJ
The info on the PE page is incorrect. Its internal volume is 1.0 cu ft. And the depth of the box dimensions is also incorrect. It's actually a bit deeper than the non-curved one. PE should've done better for these product pages.

Originally posted by Jay_WJ
I've designed MTM with the Usher 8945P and 8945A:

http://www.geocities.com/woove99/Spk.../Usher_MTM.htm
http://www.geocities.com/woove99/Spk...A_versions.htm

Nice designs!

And likely what Ceibal is looking for.

Also, thank you for the update on the PE cabinets.

[Jay_WJ]

Originally posted by Ceibal
If I am doing the math right, neither driver is ideal for the a 1 cubic foot sealed enclosure. Here are the QTC's.

Scanspeak 18w/8535 = 3.247

Usher 8945A = 1.83

My understanding is that you want to be just below a QTC of 1.

Am I correct?

I don't know what kind of formula you used. Qtc for a maximally flat response is .707.

Haven't caculated it for the Scan Speak, but for a pair of 8945P/A, a .75 to 1.0 cu ft sealed box gives Qtc between .6 and .7.

[Ceibal]

Originally posted by Jay_WJ


I don't know what kind of formula you used. Qtc for a maximally flat response is .707.

Haven't caculated it for the Scan Speak, but for a pair of 8945P/A, a .75 to 1.0 cu ft sealed box gives Qtc between .6 and .7.

I did the math wrong. It would help if I converted the size of the box from cubic feet to liters.

I edited and made the adjustments in the original post.

[ScottG]

Originally posted by Ceibal
If I am doing the math right, neither driver is ideal for the a 1 cubic foot sealed enclosure. Here are the QTC's.

Scanspeak 18w/8535 = 3.247

Usher 8945A = 1.83

My understanding is that you want to be just below a QTC of 1.

Am I correct?

You don't understand.

Luckily you don't have to really understand, for that you have basic low freq. computer modeling.

Search for the basic version of WinISD and input Zaph's measured values for the Usher Driver under the Database maintenance section. Save it, and then do some modeling with it.

That will give you a starting point for just this limited area of loudspeaker design. Its not super accurate, and it doesn't take into account things like reductions in volume to drivers/bracing/etc., floor bounce, room gain, and baffle step loss and the resulting need for compensation.

Note that I concur with Jay's assessment for the Ushers with 1-1.75 cubic feet per dual driver enclosure. Considering room gain and moderate BSC, I rather liked what I was getting with 1.75 cubic ft. tunned to 28 Hz.