Help me choose crossover points for my 3-way project...

[fjhuerta]

Hi,

I think I'll start my new project around December. I've been on the planning stages for this one since February or March (I actually got some of the drivers back then), but only now I've begun to set my goals.

I've decided on the following.

1) WMTMW, at least 100 liters.
2) W -> RS225's
3) T -> RS28's

As you can see, "M" is still under consideration.

The thing is - I've madly fallen in love with my Seas CA21 / Neo8 / Neo3 speaker. This is the first time I've really liked a design of mine... and I think the reason for it is the midrange. The crossover frequencies I used were 800 Hz / 7 KHz, 4th order acoustical. The midrange is magical - unlike the speakers I own. For the lack of a better word, I'd say it's cohesive and smooth. Maybe not having a crossover near the 1KHz - 4KHz region is worthwhile; at least this is what I understand I've learned via trial and error.

So, I'd like to take what I learned with this speaker, and design something bigger, more dynamic, but with similar midrange characteristics.

So... the original "M" was supposed to be the RS125 (I already have 4 of them). But I don't see a lot of potential with this design - crossover points would have to be (in order to keep a 3 octave spread) 300 Hz / 2,400 Hz, 4th order. And I'd place the crossover right in the middle of the region I'm trying to avoid.

Second option: the RS52 (I have a pair of them). I could get away (I believe) with a 700 Hz / 7 KHz (or more) crossover - basically using the RS28 as a supertweeter. This sounds tempting enough, and dual RS52's would drop distortion figures quite a bit.

But there may be a better way...

Third option: the Tang Band W4-1337S. I've used them on an MTM speaker, and I can't say enough good things about them. I love everything about them (except, of course, their price ). I'd need to buy 4 of them. But I'd use something like a 300~400 Hz / 6~7 KHz 4th order acoustic filters - basically, the best of both worlds. Knowing already how good these drivers sound, and thinking about using the woofers and tweeters just to "fill in" the rest of the spectrum makes me believe this could be a very "coherent" design.

There's only one problem I foresee - off-axis response may not be good enough for me to use such a high crossover point. But even using a 4KHz crossover point would be good enough for my goals, I think. Maybe 5KHz...

Also, this would add at least $250 to the project....

I believe my best option would be to use the W4's, followed by the RS52s, and the RS125's. Price-wise, the RS52s may be the best idea.

But hey - if going for the W4's would give me the best sound, I'll start saving now....

What do you think?

[Jed]

If you are going MTM for the midrange/tweeter section, I wouldn't cross at 7K, more like 2-3k to avoid vertical lobing/comb filtering issues, and why use an RS28a if crossing that high?

The RS52 and W4 are both exceptional-- if using the W4 with RS225, that is starting to sound like another Statement speaker. I'd do something different, like the RS52 in a Duntech style cabinet, but again, cross much lower, say 3k to a D26 so CTC issues can be under control.

[JonMarsh]

Well, you know I'm in a new rut.... :W

Ola Javier,


I think I have an idea about what you're excited about- I feel the same way as you seem to with your new project about the Modula NeoD using the RS52. Now, we all know I'm kind of opinionated, and sometimes like to follow some sort of off beat or wacky ideas- but bear with me...

Think about the overall design and how you'd manage the crossover- what filter types you might want to deal with the woofer and/or tweeter you select. Now, before Christmas I hope to be modifying one of my existing designs using the RS225, and turning it into a three way. Given the particular cabinet situation, I'll probably use the RS52, though I have some of the W4-1337S also. The 1337S can be run a bit lower, and working with the RS225 that may be an advantage. Driver spacing is important to manage the CTC spacing in the crossover region. If you prefer generally low order slopes, with low group delay, but want to keep reasonable control of the woofer and tweeter, I would heartily recommend considering a Duelund style network like I used on the Modula NeoD.

All the crossover acoustic targets are determined from one equation, and while the mid is a 2nd order bandpass function, the woofer and tweeter functions are in the vicinity of flat delay 4th order. If you use the 1337S and the RS28a, I'd shift the center frequency down a little. OTOH, look at the HD plots, frequency response, and CSD of the D26, as well as potential CTC benefits- that's a heck of a $25 tweeter!!!

Here's a target response. Because of the contribution the midrange makes, it would need good response from 300 Hz to 6 kHz- that's within the 1337S range. Properly done, the drivers are in the same relative phase at all frequencies over that range.



This is using a center frequency of 1300 Hz and an aleph coefficient of 2.8. The peak midrange level is about 2.5 dB below the overall combined output level, due to contributions from the tweeter and woofer. By classical terms you'd call the crossover points 550 Hz and 3 kHz. In reality, the midrange must work smoothly over a wider range. Of course, that's largely true, if ignored, in more conventional designs.

Component count is moderate- midrange is only 2nd order electrical bandpass, with some components for HF breakup supresssion. Woofer and tweeter are 4th order, but being near a flat delay (bessel) alignment, the value ranges are different (wider) than for LR4.

Give it some thought - I'd suggest emulating it in LspCAD, but I think that's only possible with the Pro version.

Here's the LspCAD sim and a measured example for the Modula NeoD (crossovers weren't in box, so bass alignment is mistuned).






My experience with this crossover approach is that it provides a very seamless frequency balance- even standing near the speaker you don't hear the individual drivers. Pink noise is the killer test. Really great on vocal music and piano.

Anyway, whatever you decide to do, have fun, I bet it will turn out very musical with that driver complement.

~Jon

[cjd]

I think the RS28A is a mismatch in this lineup, if you continue along your current thought process. It just doesn't serve as a supertweeter, but does cover deep into the upper midrange quite comfortably (and very nicely I might add).

With the W4 or RS52 I'd second the D26 tweeter.

With the RS28A I'd go RS150. (sound familiar? )

I, of course, need to work with a midrange topology along the lines you're describing (I am thinking about doing a TMWW option to go with Jon's latest WTMW for my HT) to really hear if the difference is there. But. I can say that the RS driver line is very cohesive when crossed well, such that I can not pick out crossover frequencies on the 3-ways I've done while listening to solo violin (the key here is that the violin easily plays fundamentals that are covered by each driver: 200-3kHz fundamental range - of course, the other half is that I've been playing for 29 years...)

C

[fjhuerta]

If you are going MTM for the midrange/tweeter section, I wouldn't cross at 7K, more like 2-3k to avoid vertical lobing/comb filtering issues, and why use an RS28a if crossing that high?

Ouch. I had forgotten all about vertical lobing issues. No way I could use the crossovers points I intend.

All the crossover acoustic targets are determined from one equation, and while the mid is a 2nd order bandpass function, the woofer and tweeter functions are in the vicinity of flat delay 4th order. If you use the 1337S and the RS28a, I'd shift the center frequency down a little. OTOH, look at the HD plots, frequency response, and CSD of the D26, as well as potential CTC benefits- that's a heck of a $25 tweeter!!

Jon - is designing a Duelund crossover too difficult? I'd love to try. I don't mind a bit of calculus - hopefully, it isn't that difficult to understand... do you have any links that explain the background? Or, what would I need to understand the theory behind the filter?

I think the RS28A is a mismatch in this lineup, if you continue along your current thought process. It just doesn't serve as a supertweeter, but does cover deep into the upper midrange quite comfortably (and very nicely I might add).

Yeah, I know... it's just the tweeter I have around. Since the original plans were based on the RS125, I'd cross the RS28 at 2.4 KHz. I just want to avoid the 2 - 4 KHz region, but apparently, it'll be impossible if I'm to use an MTM.

Actually - if I use an MTM, I guess a crossover frequency of 2 KHz would be the highest I should try, so that would mean I'd need to cross the midrange over quite low - say, 300 Hz. Which would mean, the RS125's would be a perfect match.

Hmm. I'm back to square one!

Unless I consider a TMWW, using a single W4...

Am I correct in assuming an MTM with the RS52's and the 28's is not a good idea, due to the vertical lobing involved?

[JonMarsh]

Jon - is designing a Duelund crossover too difficult? I'd love to try. I don't mind a bit of calculus - hopefully, it isn't that difficult to understand... do you have any links that explain the background? Or, what would I need to understand the theory behind the filter?


Am I correct in assuming an MTM with the RS52's and the 28's is not a good idea, due to the vertical lobing involved?

Here's a post with the info and links you might need:

Modula NeoD Post #2

A WMTMW could be feasible, but CTC issues would need careful management- I'd suggest the D26 used with the W4-1337S if you want to go with the Tanband midranges- that way, the total CTC in the top end might be around 6". That will probably work OK. Then the woofers need to be pretty close in, too- ideally 10 - 11", but going wider would probably be OK for on axis sound. In the Modula NeoD CC, due to the driver arrangement for a CC configuration, I do come pretty close to that.

All you need to do is calculate the target function, and import to LspCAD- you could use Dennis H's XL sheet. Of course, in my post I show how to approximate that pretty well without even using a target input. I used MathCAD to calculate the net acoustic target- it's less hassle than setting up a spreadsheet like Dennis did (Kudo's to Dennis for taking the time! :T

~Jon

[cjd]

RS125's at 2k should be capable of a very seamless integration if you target ~4th order.

Why not build up a test MTM in that configuration to see how it treats you? Having the 125's handle bass will hinder the full midrange potential but if you have even a cheap receiver with a 100 or 120Hz crossover you could roll in a sub and get an idea...

C

[fjhuerta]

Thanks for your help, Jon! I have the time. I'll surely study the theory behind a Duelund network. Who knows - maybe I'll even understand it!

c, I'm very lazy when it comes to prototyping, but I believe the only way to answer my (tons of) questions is to get my hands dirty. One thing I do know - if I'm using a 2 KHz crossover slope, i'll definitely use the 125's. $250+ dollars is too much money to spend for the W4s!