Here's one for the speaker wizards, once again I seek your wisdom. I'm running a Kappalite 3015 in a 3 cu ft cabinet, tuned to 55 Hz. Sounds absolutely killer but it does of course have some beaming issues. I decided on sticking a 6" midrange in the current bass port, which is the perfect size for a 6" (5.75" cutout), and re-tune the cabinet to 40Hz to make it a little less boomy. I've been looking at the Eminence ALPHA-6CBMRA and their PXB2 crossovers, the PXB2:800 seems to work particularly well when I model this setup in WinISD, I've modelled for 0 degrees at 6 ft and 225 degrees at 6 ft (this is where my position on the stage relative to the cabinet tends to be most of the times). However... I've never done anything with crossovers before and I might be missing something very obvious here. Selecting a crossover frequency solely on a product on offer and a response curve that looks pretty might not be the best way to go about this. Also, there's other speakers available which may suit my needs better (I'm set on the concept of a sealed basket driver because it keeps this mod nice and simple though). Maybe I need the better thermal power handling of the more expensive LA6-CBMR, or maybe the cheaper Celestion TF0615MR which is only rated for 50 watts would work just fine. Perhaps there's a far superior driver I've completely missed. Should I really want to rate this cabinet at 300 watts when I'm running a Kappalite 3015 in it, do I ever even use this much? I'm fairly sure feeding over 200 watts to this speaker peels paint of walls and melts faces a mile away. So far I've managed to find in other threads on this forum that for a crossover at 500Hz I'd need a midrange speaker rated for 25% of the woofer, what percentage would apply for a crossover set at 800Hz? Does the '50% drop in power per octave up' rule apply to this as well? So many questions, help a brother out, you may enlighten me

I'd run the 3015 fullrange and just roll off the bottom end of the Alpha 6-CBMRA. I don't have any first-hand experience with off-the-shelf crossovers, but imo what you want is a ballpark 2.5 to 3 kHz highpass filter for the mid if you want a smooth-ish voicing for the cab, and a 1.5 kHz ballpark highpass filter if you want a growly/grindy voicing for the cab. It's okay for the mid to overlap the top end of the woofer. Try both "normal polarity" and "reverse polarity" connection for the midrange driver, as there are enough unknowns that I can't say with confidence which polarity will work best.

That’s an interesting concept for sure, I think I’d be opting for a more growly voicing. Will a 3rd order HPF be the right choice for this? Just so I’m getting this right, the 3015 would be wired in parallel with the HPF and make for a 4 ohm system in this configuration?

Imo second order is plenty steep enough for this application if you use the Alpha 6-CBMRA or something similar. Third order would work too, but imo it's not necessary. I use a ballpark 1.6 kHz "damped third order" (which approximates a second order) highpass filter on a 20-watt mid, connected in parallel with a 450-watt woofer (one of the 3015's little brothers), and I've never had anyone blow one of those little 20-watt mids. It would still be an 8-ohm cab. The 3015's impedance is in the 16 ohm ballpark at 1.5 kHz, and when you connect the mid + highpass filter in parallel with the woofer, the net result will be well within the ballpark for an "8 ohm" load. It is theoretically possible to design a 1.5 kHz highpass filter for which this would not be the case, but I'm confident that an off-the-shelf highpass filter from Eminence or Parts Express will have been competently designed.

The best crossover you can buy for this is from SpeakerHardware.com It is designed for these drivers and gives a seamless transition, due to being in two parts. One rolls off gradually from the bass driver, and the other gives a gradual introduction to the 6". Do NOT wire them in parallel to give a 4 ohm load, this will possibly destroy the 6", but in any case will not give good results.

So if I got this right, the amp won't actually see the 6" as a load until the HPF starts passing along a signal? I've always been under the impression the filter itself induced a load on the amp, but now that I'm actually thinking about it that makes no sense whatsoever Eminence have an off the shelf 3rd order 1.6kHz HPF that looks good, I like what WinISD shows for this configuration. On-axis the 6" would bump the sensitivity by about 3dB from 2kHz and upwards, but on a 45 degree angle the improvement is dramatic and actually extends the usable frequency range to about 3kHz instead of just 1.2kHz with the 3015 by itself.

That Eminence 1.6 kHz filter will work very well. Remember to try both polarities. And you are absolutely right, the 6" mid will help the dispersion greatly. The Eminence filter will cause the impedance of the highpass section (mid + filter) to rise rapidly below the crossover region, so that its effect on the cab's impedance down in the woofer's region is negligible. You can think of the highpass filter as a resistor-whose-value-varies-with-frequency, in series with the 6" mid. So at low frequencies, it's a very large value resistor (maybe a hundred ohms), and such a filter in parallel with an 8 ohm woofer = an 8 ohm load. ("Impedance" is, in effect, "resistance that changes with frequency"). In the crossover region, our "resistor-whose-value-varies-with-frequency" will approximate an 8-ohm resistor. So the filter's 8 ohms + the mid's 8 ohms = 16 ohms. Recall that in that region the woofer's impedance is also about 16 ohms. The two 16 ohm loads in parallel = 8 ohms! Then higher up in frequency, where the woofer's own impedance is rising rapidly, our "resistor-whose-value-varies-with-frequency" has come down in value so much that it's approaching zero ohms, so the impedance of filter + mid = essentially "just the mid". So up around 6 kHz, for example, the woofer's impedance is about 30 ohms, and the mid's impedance is about 8 ohms, so the net impedance of the two in parallel is about 6 ohms. This is well within the ballpark for what constitutes a "nominal 8 ohm" load, and besides, electric bass doesn't have much energy that high up, so a bit of impedance dippage up there is inconsequential. I made a few simplifying assumptions in the above explanation, to convey the basic concepts with less clutter.

If you use a "standard" crossover then that EMi 1.6khz xover would work well. With that closed back 6"mid you can crossover lower but then you really must design a crossover to make it work with the rather highish fs impedance peak of the 6"mid. So 1.6khz will do the trick pretty good with an off the shelf crossover set at 1.6khz.

I was wondering the same. This makes me think it's the LF. Gotta do what the Duke says, cuz he's Da Man. But hes got to know what you have. LF or not? It makes all the difference in the world. LF might want a maximum crossover of about 800 to 1000 KHz IMO. A 2 way crossover might also be best, but let Duke tell you for sure...

If I have misunderstood and it is in fact the 3015LF, then please disregard all my suggestions above, because they don't apply. If it's the 3015LF, then imo Pickles' advice is right on the money.

Thanks to all for your input above. To get the confusion out of the way, it is indeed the 3015 and NOT the LF. The roll-off at 1.2k is due to the beaming effect WinISD predicts at a 45 degree angle, not because I was modeling the LF, at that angle the 3015 and the 3015LF actually show rather similar response characteristics. I think I've pretty much decided on running the 6" behind an HPF instead of crossing over both drivers, like DLJ suggested. Shipping and handling from SpeakerHardware.com to Europe make them prohibitively expensive, so my off the shelf options appear limited to the 3rd order 1.6k from Eminence or a 2nd order 1.2k from Blue Aran. The latter shows a smoother transfer function in WinISD but I'm not entirely sure if I can pull this off with regards to the impedance, in reference to Arjank's post quoted below: Edit: looking at the impedance curve for the 6" there only seems to be a one ohm difference between 1.2k and 1.6k, I'm fairly sure I can get away with this

IMHO and IME, that's at least partly because WinISD isn't really meant to model performance beyond 500Hz or so, and that's probably being generous. It's treating beaming as strictly pistonic behavior and doesn't have a mechanism for understanding cone breakup and the dispersion effects of the dome.