An unforeseen problem with 3 way drivers - FR overlap?

[fjhuerta]

Hi all!

I've been working a bit on two current designs: a 3 way and a 2.5 way floorstanders. I haven't yet decided on which one to start first (and I have to finish another pair of MTM floorstanders, too).

Anyway, I had some extra free time this weekend, so I began to simulate the 3-ways with the stock data from PE. The 3-way uses the Dayton RS225 (2), and the RS125 & RS28 in a D'Appolito configuration.

I first modeled the woofers (4th order acoustical, 500 Hz) and the tweeter (4th order acoustical, 1800 Hz), since I have never designed a bandpass crossover.

I found out that I had quite a lot of overlap using 4th order slopes, or by using the FR ranges I chose. Bottom line is - when designing the bandpass filters, the final curve that gave me the flattest response looked terrible all by itself. I assume the tweeter / woofer / mid overlapped at certain frequencies, so LSPCad compensated by using destructive interference and / or strange transfer functions.

I already know designing a 3 way is a lot harder than a good 2 way. But my question is - should I try aiming for steeper slopes? Maybe using 6th order acoustical filters will help with the overlap thing. I could also try going for a wider bandpass filter - for example, 300 Hz / 2000 Hz. But I'd be really pushing the RS125, especially since I think it's upper FR limit is very close to 2KHz, and I don't think it has good bass response, either.

Any ideas / suggestions?

[augerpro]

Can you post some screenies? The close XO points are very difficult to work with in my limited experience. And 500/1800 is VERY close.

[JonMarsh]

As Augerpro points out, you're using rather close crossover frequencies.

In general, if a driver can't manage three octaves as a midrange, it may not work well in a three way once you allow for the transistion bands. You're suggesting something less than a three octave range for the mids. Why use two drivers with a relatively poor Xmax and small Sd, when they don't go significantly higher than the larger RS180?

There's reasons the Wilson three ways look the way they do.

For an MTM midrange, think something more like the Tangband W4-1337S.

Also, look for problems (possibly), in your LP and HP networks and their phase characterisitcs. The optimization process for a three way is more tricky- it seems to work best if you have really clean data all around, optimize the target transfer functions on low and high pass, do the same for the midrange independently (bandpass), and only then try some optimization for flattening the bandpass area for by optimizing components in the midrange crossover- such as ET did in the Duelund synthesis recently in the RS52 CC post. If you check off penalize destrutive interference, you should also have fewer issues.

Destructive interference on axis ALWAYS results in problems in the room power response, because somewhere off axis it will probably be adding in phase...

Also, for three way designs, I really recommend sticking with all-pass type crossovers which have the drivers in relative phase with each other through the transition. You may be doing that anyway, but depending on how you manage driver positioning and alignment of acoustical centers (or compensation in the crossover network) you may have some additional problems.

In general, a three octave span on midrange is advised; two octave is more of a filler driver, in which case you probably should be considering a Duelund alignment with low coeffficient (again, the RS52 CC).

[fjhuerta]

Brandon, Jon, thanks a lot for your thoughts.

I'm using the RS125 because that's what I had around. Fortunately, I haven't committed to a design yet - strangely enough, I decided that simulating things a bit would be better for my mental health afterwards (I guess I'm learning from you guys!).

From the looks of things, I guess using the RS52 as a true mid would be a much better idea. Maybe crossing it over at 600 Hz / 5000 Hz. Would that be a better choice (I just let out an involuntary "Duh!" sound, so I guess it is)?

Just my luck. The RS52 is out of stock again. You guys bought them all out, didn't you?

[JonMarsh]

Nah, only two this last time around. :W

Sometimes I wish they had an RS75 dome. Go a little lower and louder, you know.

The popularity of the RS52 is probably revealed by how fast it sells out.

There should be a lot of potential in the W4-1337 as a midrange



Distortion spectra is pretty clean up to 7-8 kHz

[Paul Ebert]

Sometimes I wish they had an RS75 dome. Go a little lower and louder, you know.

Or the much anticipated RS90. With my luck, they'll probably release it right after I order my RS52s. :roll:

[cjd]

RS90 sensitivity sucked

[fjhuerta]

Nah, only two this last time around. :W

Sometimes I wish they had an RS75 dome. Go a little lower and louder, you know.

The popularity of the RS52 is probably revealed by how fast it sells out.

There should be a lot of potential in the W4-1337 as a midrange



Distortion spectra is pretty clean up to 7-8 kHz

You know... these are the times when I remember I already used 4 of those W4-1337's... and that project could have used my 4 RS-125's instead

At this point, I'm actually tempted to try and simulate the RS125's at 250 Hz / 2KHz, or 200 Hz / 1600 Hz. Reason being, I can't justify spending much more money on this project, and the 4 W4's would cost me around $300 after shipping and taxes.

My other option would be the RS52s.

Would my suggested crossover points work out? On one hand, I'd try to run the RS125's as high as I can because of their 2.5mm Xmax. On the other hand, I can't go much higher than 2 KHz because of their break-up.

Would I be better off just waiting for a while and buying the RS52's? From a design standpoint, I'm sure it would simplify the crossover and measurements a bit, and I'd probably have better lateral dispersion. On the other hand, an MTM would help with a specific requirement I have regarding placement (there's a sofa nearby, and I need to place speakers as high up as I can), and I suspect it'd be a bit more dynamic.

[JonMarsh]

I've done WMTMW speakers with dome mids before- in the 90s. Pretty dynamic. 750Hz and 3.2 kHz crossover points, LR4. Tweeter in the middle, SS Kevlar woofers.

250Hz/2 kHz might be a better bet with the RS125's. Just depends on your crossover- how expensive are you willing to go there? As Steen Duelund says, the crossover usually should cost more than the drivers.

[fjhuerta]

250Hz/2 kHz might be a better bet with the RS125's. Just depends on your crossover- how expensive are you willing to go there? As Steen Duelund says, the crossover usually should cost more than the drivers.

As expensive as it needs to be, but no more. I'm already planning on building a nice cabinet in black / white polyester and maple inserts. In a way, I want to build the best speaker I can. I suspect the project will take at least what's left of the year.

I already suspect the inductors will cost a fortune...

[Mark Seaton]

Just depends on your crossover- how expensive are you willing to go there? As Steen Duelund says, the crossover usually should cost more than the drivers.

While I'm not saying anything new, I'd like to emphasize the point Jon is making here. Most here are hardware junkies and like to drool over the latest and greatest drivers, but it's easy to forget that the crossover execution, not the drivers, are often the biggest point of deliniation between very good and great sounding speakers... (IMNSHO, YMMV, etc, etc...) :roll:

[WillyD]

I almost forgot about the RS90. Should be a fun driver, but as Zaph said, the RS52 is more ideal for a midrange.

[JonMarsh]

While I'm not saying anything new, I'd like to emphasize the point Jon is making here. Most here are hardware junkies and like to drool over the latest and greatest drivers, but it's easy to forget that the crossover execution, not the drivers, are often the biggest point of deliniation between very good and great sounding speakers... (IMNSHO, YMMV, etc, etc...) :roll:

So, SO true Mark, thanks for emphasizing that, too. So often I fear I come across as a bit of a broken record on that topic, though it can't be overstated enough... (except for the folks trying to use full range drivers, and even then you usually need precise BSC, which is, after all, part of a crossover! In the 70's while I was a partner in a Hi-Fi shop in Boulder Colorado, to make a point to my partner and some friends, using the drivers from the B&W DM6 (we were B&W dealers and authorized repair- got on hand a full set of drivers), and built an "optimized" design from the viewpoint of what I knew than (cabinet size, system Q of 0.5, zoble networks, etc) which was dead flat on axis, and it wasn't really fair- because I didn't use the autoformer bass control in the DM6, or electrolytic caps, or ferrite core inductors, it would absolutely clean the clock of the original design...)

Drivers are key components, but the crossover is one ring to rule them all...

[TacoD]

I also using high grade xo parts in my designs, even for the temporal test filters. IMHO it's important to invest in parts quality and don't skimp on the part quality of the zobel/lcr notches either.

[fjhuerta]

So, does that mean going for polyester film caps on the shunt legs of the woofer crossover, too? I can already see how much a 200uF bank of Solen caps will cost me...

I'd better start saving my pennies right now!

[JonMarsh]

Well, yeah, you see, three ways ARE expensive- I don't use electrolytics anywhere. And on my current projects I'm planning on improving the quality of shunt caps in the mid and treble, too, as well as the series.

Everyone has to make those decisions for themselves, of course, but particularly in the treble crossover, I think it's important if you have decent gear and source material.

[fjhuerta]

Well, yeah, you see, three ways ARE expensive- I don't use electrolytics anywhere. And on my current projects I'm planning on improving the quality of shunt caps in the mid and treble, too, as well as the series.

Everyone has to make those decisions for themselves, of course, but particularly in the treble crossover, I think it's important if you have decent gear and source material.

That's my intention - if I'm already spending 600~700 USD on the cabinets alone, I'm not going to spare any expense on the crossover. Or the drivers themselves.

I'm looking into buying 4 of the W4-1337's again. I loved them, and if that's what it's needed to make this project work, so be it...

I'm still puzzling over how to choose between dome mids or an MTM array with 4" woofers. I haven't found much info on relative advantages or disadvantages on them.

[Evil Twin]

I'm still puzzling over how to choose between dome mids or an MTM array with 4" woofers. I haven't found much info on relative advantages or disadvantages on them.

This is where your own experience and ability to connect the dots must come in...

it is largely a matter of physics- Sd, Xmax, onset of linear distortion and increased non-linear distortion in the top end and low end, CTC distance vs crossover frequency- the rest is your personal desires, perspective, and creativity.

Trust your feelings, Javier- you will sense the right direction and find your answer in the Force...

[Mark Seaton]

Well, yeah, you see, three ways ARE expensive- I don't use electrolytics anywhere. And on my current projects I'm planning on improving the quality of shunt caps in the mid and treble, too, as well as the series.

Everyone has to make those decisions for themselves, of course, but particularly in the treble crossover, I think it's important if you have decent gear and source material.

I agree, and while I agree that quality, good measuring & performing parts are very important, let's not forget the concept that the component *value* (actual Ohms, mH, uF, etc) will often have a bigger impact on the sound than component quality. I keep a handful of electrolytics around for testing and quick value confirmations. Often times the choices we inevitably make between incremental component values can add up to subjectively identifiable differences.

I guess it gets back to one of those problems with most traditional hi-fi companies who tend to get caught up in making sure they have the best components and have exhausted all efforts in minimizing the problems inherent to the design choices they made early on. When you start your design, and before you start ordering exocit parts, make sure you've minimized the number of problems you are left to deal with!