Please explain the "two and 1/2 way" crossover to me.

I understand that if you have a speaker with a tweeter driver on top and two same sized woofers below it that many crossover designers will roll off the mids from the bottom woofer. Apparently there are some lobing issues or some sort of destructive interaction from leaving both woofers with significant program material in the mids. I currently have some speakers I'm trying to decide whether to modify (Klipsch KG 5.5) that have this configuration but the woofers are both fed the same signal with a crossover to them below 1.5 KHz. Is this issue something that some designs don't have much problem with for some reason? I've been considering upgrading to one of the Crites crossovers, but it seems silly to invest more in a possibly flawed design. I don't have the resources to design a better crossover - is there anyone out there who designs new crossovers for vintage speakers at reasonable prices?
 
With 2 woofers playing the same freq range, you will get a blurred image. I tried that with one of my TMM (2.5 way) Scanspeak builds and decided to scrap that design. Instead, roll off the lower woofer according to the freq transition of the baffle step. If for example your baffle width is 8'', I would crossover the lower woofer at around 570Hz by preferably a 2nd order filter. The great thing about 2.5 ways is the lower midrange has some serious authority and can sound very full and rich.
 
A 2.5 way can be as simple as hanging a large coil on the lower woofer, assuming the woofers are wired in parallel.
I had proposed doing just that, but had been told that the rolloff would be too gradual to have much effect and that the rest of the crossover would have to be redesigned to compensate for the addition for some reason. I also have a pair of Pioneer HPM-900s where the woofer is allowed to run "wide open" and had considered a crossover upgrade for them as well - my discussion of adding an inductor prompted that response.
 
With 2 woofers playing the same freq range, you will get a blurred image. I tried that with one of my TMM (2.5 way) Scanspeak builds and decided to scrap that design. Instead, roll off the lower woofer according to the freq transition of the baffle step. If for example your baffle width is 8'', I would crossover the lower woofer at around 570Hz by preferably a 2nd order filter. The great thing about 2.5 ways is the lower midrange has some serious authority and can sound very full and rich.
The things that bug me about the speakers currently are some "horn coloration" and stridency to the treble and a lack of integration with the bass - almost the feeling that the mids have been EQ'd out. I'm hoping that with a Ti diaphragm upgrade, some treatment of the edges of the horn, and a crossover upgrade I can end up with a much nicer speaker. Do you think the above mods plus adding an inductor to roll off the lower woofer would address the issues I have?
Also, the speaker is 12.25" across on the baffle (with 2 X 10" woofers). What frequency should I shoot for to roll off at?
 
Hi,

The fundamental aspect of a 2.5 way, or any 0.5 way is the arrangement
of the two drivers, one being gradually rolled off early. This automatically
on a large flat baffle would cause the response (compared to one driver)
to rise by 6dB towards the bass. However in a cabinet the box exhibits
6dB of baffle loss, and requires BSC, "baffle step compensation", one
way of do it is via two bass/mid drivers in 0.5 way configuration.

Converting your Kipsch to 2.5 way will leave the bass level untouched
but reduce mid and treble levels by 6dB. Many high efficiency designs
simply ignore BSC and try and use near wall / corner placement.
(With the stated senstivity, BSC is certainly being ignored.)

BSC's speakers need to be away from walls and corners, and
consequently give good imaging, they also interact well the
room nodes very low down with a correct bass alignment.

Its get complicated with larger speakers with drivers near the floor,
and would very much depend on the drivers themselves having a
rising response to work well for this case IM0, as this would allow
some BSC adjustment, or make BSC less than 6dB effectively.

Some details here :

http://www.audiokarma.org/forums/showthread.php?p=4218745

The c/o is relatively simple, but not easy to change without measurements.
Bass : 1mH, 24uF+2R : to twin bass/mid units
Treble : variable resistor/Lpad, 2.5uF, 0.6mH, 8uF : to tweeter

rgds, sreten.

undefinition
Zaph|Audio
FRD Consortium tools guide
RJB Audio Projects
Speaker Design Works
HTGuide Forum - A Guide to HTguide.com Completed Speaker Designs.
Humble Homemade Hifi
Click below to go to
 
Last edited:
More than you probably want to know...

A 2.5 way crossover means a system with 2 woofers but different crossovers for each. One (the one nearer the tweeter) runs up to the tweeter crossover while the second one has a rolloff some octaves away. What you get is better vertical dispersion because you don't have two woofers running in parallel at mid frequencies. At the same time you get the full bass power handling of having two woofers for low frequencies.

Rolloff rate is key because if you rollof the lower woofer too slow, then it won't be at a level enough below the uppper woofer to improve the vertical directivity. If you roll it off too fast then the extra phase shift will mess up the on axis response. It only works well with a steady first order rolloff (for 45 degrees phase shift).

A first order rolloff is just an inductor, right? The problem here is that you woofer itself looks pretty much like an inductor, so adding an inductor in series will shelve down the midrange but not roll off at the right rate. This is what the Pioneer guys were telling you.

Doing a proper first order rolloff either requires a conjugate on the woofer to make it look like a resistor, or a heavily damped 2nd order network, essentially the same thing.

David S.
 
A 2.5 way crossover means a system with 2 woofers but different crossovers for each. One (the one nearer the tweeter) runs up to the tweeter crossover while the second one has a rolloff some octaves away. What you get is better vertical dispersion because you don't have two woofers running in parallel at mid frequencies. At the same time you get the full bass power handling of having two woofers for low frequencies.

Rolloff rate is key because if you rollof the lower woofer too slow, then it won't be at a level enough below the uppper woofer to improve the vertical directivity. If you roll it off too fast then the extra phase shift will mess up the on axis response. It only works well with a steady first order rolloff (for 45 degrees phase shift).

A first order rolloff is just an inductor, right? The problem here is that you woofer itself looks pretty much like an inductor, so adding an inductor in series will shelve down the midrange but not roll off at the right rate. This is what the Pioneer guys were telling you.

Doing a proper first order rolloff either requires a conjugate on the woofer to make it look like a resistor, or a heavily damped 2nd order network, essentially the same thing.

David S.

Dave, sorry to ask, but what is a "conjugate" on the woofer? Thanks.
 
sreten; hopefully the changes will reduce some of the stridency I hear in the upper mids and low treble. I worry a little about losing too much from the woofers, as to me the horn currently overpowers the woofers a bit. I'm hoping some of that is simply the ragged high range output from the woofers and with correction it will improve without losing significant "good" mids or bass, and that the change to a Ti diaphragm will both smooth the low range output from the horn as well as extend the highs a bit.

Dave; sounds like the conversion to the 2.5 scheme may be a bit more invoved than a simple inductor, but that if it is done thoughtfully I won't have to change the rest of the crossover. Correct?
 
Dave, sorry to ask, but what is a "conjugate" on the woofer? Thanks.

A conjugate circuit is a couple of components to put in parallel with the woofer to flatten out its impedance. Where the woofer's inductance would make its impedance rise, the conjugate's value drops to flatten it. For a woofer that usually means a resistor in series with a capacitor (both in parallel with the woofer). Sometimese they are called Zobel networks.

The term comes from "complex conjugate" a math term for something with the same real part but the opposite imaginary part.

David S.
 
Dave; sounds like the conversion to the 2.5 scheme may be a bit more invoved than a simple inductor, but that if it is done thoughtfully I won't have to change the rest of the crossover. Correct?

It isn't that simple. Unless you have midrange level to give away then you won't be able to get the response flat. Even if the tweeter network doesn't change then the woofer network would have to be rescaled to work for one woofer rather than two (twice the impedance). Then you would be designing the lower woofer network from scratch.

None of this is practical unless you can measure frequency response of the system.

David
 
I was afraid you'd say something like that. I guess I need to start getting more serious about this hobby and start looking into acquiring some tools to measure what I'm doing. I'll plan to start poking around for the appropriate threads to begin the learning process. My dad was an electrical engineer and physics major, but I'm in health care, so my science is on the soft side.

In a paraphrase of Bones McCoy (Star Trek) - "Dammit Jim, I'm a doctor, not a speaker system designer!"
 
Hi,

A "conjugate" is the same as a "zobel", it compensates for driver inductance.

Its not needed on 0.5 way driver, you exploit the drivers inductance.

Simply put :

The first network should be the mid/treble c/o, this is fed to both drivers,
though the values are essentially designed for one driver, not two drivers.
For the 0.5 way driver additional series inductance is added to roll it off,
This is best placed after the above network, not always done like this.

rgds, sreten.
 
Hi,

Regarding changing the crossover :

1) the tweeter output needs to be reduced by ~ 6dB.
If its not an L-pad in the c/o fit an L-pad to the driver.

2) For the bass crossover :
1mH -> 2mH, 24uF -> 12uF and 2R -> 4R.
I cannot be specific about the additional 0.5 way inductor,
but the value needs to be fairly substantial, around 5mH.

3) No guarantees it will work well, your adding a lot of BSC.
It will sound like it has lots more bass, because it will have.

4) Without the actual driver responses its a bit of a crapshoot.
A full redesign and modelling for a 2.5 way would be a lot better.
That might allow the amount of BSC to be effectively varied.

5) Detuning the port by lining with ~ 10mm foam would be a good idea.

6) Its always possible the lower bass driver has mechanical modifications
that effectively make it a 0.5 way driver, basically adding mass to it at
the voice coil, but the specs suggest the drivers are identical.

7) If you do it, listen to all four drivers in the raw, and if you can tell,
use the two smoothest sounding as the upper bass / mid driver.

rgds, sreten.
 
Last edited:
Its not needed on 0.5 way driver, you exploit the drivers inductance.

Simply put :

The first network should be the mid/treble c/o, this is fed to both drivers,
though the values are essentially designed for one driver, not two drivers.
For the 0.5 way driver additional series inductance is added to roll it off,
This is best placed after the above network, not always done like this.

I've tried this and it doesn't work as well. The issue really is about getting enough roll off of the lower woofer without adding any more phase shift than you have too. Having the second rolloff (related to the shared tweeter crossover point) gives extra phase shift that fights you. I have had good luck using modified shelving circuits because I could pull down the midrange substantially and then, due to the shelf, have the phase flatten out to minimize the phase shift between the two woofers.

Without attention to these details you get considerable ripple in the on axis response.

David
 
Anybody here ever measured the KG 5.5 stock or with the Crites mods? Any speaker design is of course an exercise in tradeoffs, and apparently the original designers went for sensitivity as a major design goal and decided to keep the crossover very simple. I wonder how much of that was due to pressure from the bean counters and how much due to lack of some of the progress in measurement and design technology we have now.

Was the 2.5 crossover in common use at the time this speaker was being designed?
 
I've tried this and it doesn't work as well. The issue really is about getting
enough roll off of the lower woofer without adding any more phase shift
than you have too. Having the second rolloff (related to the shared
tweeter crossover point) gives extra phase shift that fights you.
I have had good luck using modified shelving circuits because I could pull
down the midrange substantially and then, due to the shelf, have the
phase flatten out to minimize the phase shift between the two woofers.

Without attention to these details you get considerable ripple in the on axis response.

David

Hi,

Its works better than the alternative arrangement and your point is irrelevant.

What your saying in nonsense, I'm sorry, but its true, at the c/o point the
0.5 drivers output phase is irrelevant, its totally dominated by the 1 way
drivers output and can be safely be near ignored, (< 90 degrees anyway).

The point of the arrangement is to make sure the 0.5 way driver does not
overlap the 1.0 driver in its roll off region. A simple series inductor to the
0.5 driver forms a shelving circuit due to driver inductance, and this can
overlap the 1.0 driver in its cutoff region due to its higher order filter.

It is the way to do it, but inductors need to be rated for twice the current,
1st filter, so its not the cheapest way way of doing it, but it is the best way.

It may be cheaper, if your messing around with more 0.5 way components.

rgds, sreten.
 
Last edited: