Hello all,
I would like to gain some insight from our member experts as to using 18db/octave crossovers in a 2-way system, for example at 2,600 Hz. Good, bad, indifferent?
I can supply some driver and cabinet information if needed for a better professional opinion.
Thanks in advance. /Robert
I would like to gain some insight from our member experts as to using 18db/octave crossovers in a 2-way system, for example at 2,600 Hz. Good, bad, indifferent?
I can supply some driver and cabinet information if needed for a better professional opinion.
Thanks in advance. /Robert
Thanks, so what I have are Celestion G15-150 woofers (150W, 98dB 1W/1m, 30-5KHz, fo 35Hz) in 4 cu.ft. ported JBL-cabaret style cabinets. The original tweeters were Celestion RTT50 horns which used to click and buzz. I replaced them with a Viston SC-10N dome tweeter.
My concern is with the power mismatch of the tweeter (90dB, 1KHz-20KHz, fo 1,700Hz, 100W @ 4KHz 12/db octave). I contacted Visaton and asked what the power rating would be crossed at 2,000 Hz 12db/oct and was advised around 20W, and that it was risky operating near the 1,700 Hz resonance and I shouldn't go lower than 3,000Hz.
I then asked about power ratings at 3KHz 12dB/oct and was advised to be around 50W. Fine. So then I did some crossover modeling with Spice and noted using this suggestion the 1,700 Hz point was about -16dB down. I then did some modeling with an 18dB/oct crossover at 2,600 Hz and found the same, about -16dB at 1,700Hz.
However a friend suggested I avoid 18dB crossovers like the plague due to funny 90°/270° phase shifts. How bad could that be?
My concern is with the power mismatch of the tweeter (90dB, 1KHz-20KHz, fo 1,700Hz, 100W @ 4KHz 12/db octave). I contacted Visaton and asked what the power rating would be crossed at 2,000 Hz 12db/oct and was advised around 20W, and that it was risky operating near the 1,700 Hz resonance and I shouldn't go lower than 3,000Hz.
I then asked about power ratings at 3KHz 12dB/oct and was advised to be around 50W. Fine. So then I did some crossover modeling with Spice and noted using this suggestion the 1,700 Hz point was about -16dB down. I then did some modeling with an 18dB/oct crossover at 2,600 Hz and found the same, about -16dB at 1,700Hz.
However a friend suggested I avoid 18dB crossovers like the plague due to funny 90°/270° phase shifts. How bad could that be?
Last edited:
What you have done in comparing (eg -16dB) is some reasonable research and I'd tend to agree. The phase difference can actually be a good thing, but this is not popularly understood. Provided you don't introduce other factors into it, you should have no problem with that crossover.
Thanks. Correction above, actually -12db down, not -16dB. I think I'm going to go ahead and try the 18dB arrangement, maybe only do one cabinet and conduct an A-B test. The proof is in my ears. 😉🙂
Alternatively I could change the existing 12db crossovers to 3KHz, but I feel the tweeter is better suited for the upper mids than the woofer is.
Alternatively I could change the existing 12db crossovers to 3KHz, but I feel the tweeter is better suited for the upper mids than the woofer is.
Not surprising, those RTT50's are super tweeters which means they are designed to add top end sparkle(10khz+) above a larger compression driver. They got no business doing midrange at all. How low were they crossed to the 15's?
This doesn't make much sense either. How are these boxes used? There is no way to effectively use those tweeters in a PA application.. they just do not produce enough output so the potential crossover slope used is irrelevant at the moment. If this isn't for a PA application please say so because that means you don't need all the output of the G15's so other options can come into play.
Here the presentation of a two ways 18 dB/o crossover (of enhanced performances compared to the classic 18 dB/o Butterworh) belonging to a class known as "quasi-optimal" in France.
For a normalisation at 1 kHz :
both ways are made of filters built with a 12 dB/o Butterworth (Q = 0.707) in series with a 6 dB/o, both at 1000 Hz.
A delay of 210 µs (72 m) is introduced in the treble way.
Performances 20-20000 Hz :
Variations levels for the sum
0.9 dB axis level ripple
171 µs group delay riplle
Levels
-19.2 dB treble way at 500 Hz
-6.0 dB both ways at 1000 Hz
-19.3 dB bass way at 2000 Hz
Phase treble - bass
52° at 500 Hz
14° et 1000 Hz
-50° at 2000 Hz
For a normalisation at 1 kHz :
both ways are made of filters built with a 12 dB/o Butterworth (Q = 0.707) in series with a 6 dB/o, both at 1000 Hz.
A delay of 210 µs (72 m) is introduced in the treble way.
Performances 20-20000 Hz :
Variations levels for the sum
0.9 dB axis level ripple
171 µs group delay riplle
Levels
-19.2 dB treble way at 500 Hz
-6.0 dB both ways at 1000 Hz
-19.3 dB bass way at 2000 Hz
Phase treble - bass
52° at 500 Hz
14° et 1000 Hz
-50° at 2000 Hz
If I have found the woofer, it is a 15" PA guitar speaker. 98dB.
You'd use a compression driver with a horn to match that level and loudness.
A 1" 90dB SC10N hifi tweeter just won't get near that loudness or be tough enough.
The Eminence PXB2 2K5CX is your typical second order 2.5kHz solution used in some of their co-axials which are essentially compression drivers in a horn.
PXB2:2K5CX - Crossovers | Eminence Speaker
More detail here:
https://www.diyaudio.com/forums/multi-way/324695-cupcake-speakers.html#post5482054
You'd use a compression driver with a horn to match that level and loudness.
A 1" 90dB SC10N hifi tweeter just won't get near that loudness or be tough enough.
The Eminence PXB2 2K5CX is your typical second order 2.5kHz solution used in some of their co-axials which are essentially compression drivers in a horn.
PXB2:2K5CX - Crossovers | Eminence Speaker
More detail here:
https://www.diyaudio.com/forums/multi-way/324695-cupcake-speakers.html#post5482054
@conanski, previously I was using the cabinets for PA DJ use, before I moved up to a larger JBL bi-amp system.
I'm now retired from DJ work. I'm presently using the cabinets on my home stereo powered by a Marantz 2238B.
You may be thinking of the Celestion HF50 bullet tweeters. The RTT50 were midrange-tweeter horns, 50W, 1.5K to 15K, 101dB 1w/1m, with recommended crossover 2K 18db/oct. Mine were crossed at 2KHz but at 12dB/oct. However they were 16 ohm drivers padded down and impedance matched with a 20W 16 ohm resistor in parallel for 8 ohms.
I like the smoothness of the SC10 dome tweeters. They are the best match I could find for the woofers and from available part suppliers here in Canada.
Ok thanks for that... looks like I got some bad info out there on the interwebs. Can't imagine how that happened... 😀
FWIW... I just happen to own a pair of G15-150s as well, unfortunately they are in need of a recone after falling victim to my early DJ antics, seems their power rating is accurate and I needed something bigger. I couldn't bring myself to get rid of them so they have been languishing in storage for decades. One of these days.
Do you know about QComponents
Allright. That means you don't need all the output of the Celestion 15. As you are aware it's normal to have to pad down a CD to match the low/mid driver, now you need to do the opposite. You could do the same with an L-pad on the woofer but one of the downsides of a PA drivers high sensitivity is poor low frequency output. Compounding that is something called baffle step, are you aware of that phenomena? It's the drop in forward energy from a speaker at lower frequencies that have wavelengths longer than the baffle width. At frequencies above baffle step most output from the LF driver is unidirectional and travels forward.. the baffle acts as a wall relative to the sound wavelengths. Below baffle step sound wraps around the box and become omnidirectional, the total sound energy is the same but because some is "lost" around the sides and back the portion that moves forward towards the listener drops in level.
Baffle step compensation is a filter that reduces output above the step frequency to get a broader flat response, and the beauty is you can build in as much attenuation as you want. Normally this isn't done with PA speakers as it throws away a ton of usable output for no good reason.. a user typically would just boost the lowend a little with EQ and the subs being used will also fill in the very bottom octave or two as well.
But your situation is unique and should work well, you may even end up with the best sound you have ever achieved from these speakers. Here is link to a baffle step calculator that may come in handy.. Loudspeaker Diffraction Loss and Baffle Step Compensation Circuits