Best to identify a problem, find its cause, and rectify it.
What sounds are your speakers finding difficult?
Remember, it takes strong medicine to change a gentle filter much.
What sounds are your speakers finding difficult?
Remember, it takes strong medicine to change a gentle filter much.
Okay. I think I get it. If I change the tweeter crossover from 1st order to 2nd order by adding a 0.30mH inductor, that would reduce the power the tweeter receives before the XO point. Then I could lower the XO point (4khz to 3.8 or 3.5hz) by raising the 3.0uF cap to 3.3uF or 3.9uF without damaging the tweeter.
Yes, you change the 'slope' of the 'curve' but the (elettro)mechanical point when the tweeter distorts Is mantained away.
I edited my main post with additional information. Please provide feedback. I want to make sure I'm on the right track before I prototype a new crossover.
Just throwing this out there and I understand if you want to continue the path you're already on... But given the worth of the Allison-One speakers and the great condition yours appear to be in, in my opinion I think you might be able to get better sound if you sell the speakers and build an entirely new set than possibly redoing the crossovers on these. You'll already have an expensive crossover if you rebuild your Allison One speakers.
Theoretically yes, but these tweeters are already the first thing to blow in these Allison Ones, so be careful as you don't have much room here.
That's wise advice BUT I'm a hot rodder. I want to bastardize the speakers so no one can ever hear the original 1970s sound.
I see 2 options:
Modify the existing XO with the Allison speaker.
Roy Allison used a 2nd order XO on the latter CD-9's tweeter. So my 1st modification should be to replace the 1st order XO with a 2nd order. Based on my math If I use a 2.7uF cap + 1.5mH inductor then I should be able to get a similar response. I could then try fine tuning the speaker by changing the resistors in the midwoofer's circuit. This is the cheaper strategy and it would increase the Tweeter's power handling and it would keep the sound response closer to the Allison spec.
Replace Allison Tweeter with SB Acoustics SB26ADC-C000-4
These tweeters are drop-in replacements and they'll allow for a more extensive redesign of the crossover. The Allison Tweeters have an Fs of 1750Hz and a linear region from 3.5kHz. The SB has an Fs of 680Hz and a linear region from 1700Hz. With the SBs, I would be able to drop the mid-tweeter crossover from ~3.5kHz to 2-2.5kHz. This would require a substantially redesign of both the tweeter's and midwoofer's circuits but it'll massively the speaker's linearity, off-axis integration and power handling too. It'll turn the Allison into a modern system.
Currently, I'm learning REW so I can measure the speaker. Then I'll try option 1 and if I'm not satisfied with the result then I'll do option 2 and sell my Allison tweeters.
Last edited:
I get it, just tossing options out there. And the idea of making a 2nd order xover for the tweeter is a good one, but without measuring equipment it'll be tough to know the reality of the (actual) acoustic roll-off you're achieving with your 2nd order electrical filter, and the acoustic roll-off is the one that matters.
In my opinion, this would definitely be a fun project. A good friend of mine had the original Allison One speakers, so I've spent a little time listening to them. They would be tough speakers to (re)do a xover on, given that they're designed to always be listened to off axis. But what makes them more difficult, looking at the existing xover for these, they're a dipole configuration, and I've never designed a dipole speaker (or a bipole for that matter), which is why it would be a fun learning experience for me. But I would not tackle this without at least a microphone to measure the response. In my opinion this is way beyond elementary, this is about as advanced as it gets with regard to xover design. Not saying you can't get better sound, just making the statement 🙂
For example, my first question regarding these speakers is... is wiring those 2 woofers out of phase really the best approach, since bass is pretty omni-directional? I have no idea about the T/S parameters of the woofers they used in these, so I'm not sure how this would impact the cabinet size and port requirements for this woofer. Why did he reverse the polarity? Was it simply to lower distortion? Also, should the cabinet size double if the polarity gets reversed and both woofers are in phase? I'd be interested and probably start there and listen to how flipping polarity on one of the woofers impacted the bass response. From there... I'd move up to attempting a better tweeter network that's safer for the tweeter.
But while playing with the tweeter response, I would also attempt to tame that gigantic 5 db peak from 800 Hz to 6khz, but now we're also playing in the midrange. And this is also where a microphone to measure response would be handy- My understanding is, that peak is there because these were measured out in the room, rather than against the wall where they're intended to go (I could be wrong?). And putting them against the wall will bump the response in the lower region (I would want to measure these against the wall to see if that's actually true). But that still doesn't explain what would then be a dip starting at 5khz, it's almost like these beg for a super-tweeter. But again, I would definitely want a mic for measuring before I tackled this.
But while playing with the tweeter response, I would also attempt to tame that gigantic 5 db peak from 800 Hz to 6khz, but now we're also playing in the midrange. And this is also where a microphone to measure response would be handy- My understanding is, that peak is there because these were measured out in the room, rather than against the wall where they're intended to go (I could be wrong?). And putting them against the wall will bump the response in the lower region (I would want to measure these against the wall to see if that's actually true). But that still doesn't explain what would then be a dip starting at 5khz, it's almost like these beg for a super-tweeter. But again, I would definitely want a mic for measuring before I tackled this.
I agree. The crossover for this system is on the difficult side. It uses a zobal and the system is a dipole design. Also the drivers are separated quite far from each other. Understanding this system requires quite a bit of knowledge.
I think the best way to optimize it is to get the response as flat as possible on each face (so 45°) and allow the on axis measurement (0°) to be what it is. The large peak from 800-6kHz is caused mostly by the Tweeter's response and the midwoofers zobal network. The result is the sound is too heavy or thick, almost like an overly dramatic Hollywood traitor.
The more I think about it, the less I think a 2nd order XO will work. There is 15CM between the center of the tweeter and the midwoofer. This equals the wavelength 2450hz. So If the XO is above this frequency then the driver blending suffers. A 2nd order XO would make the blending worse. The better strategy is to add a notch filter from 2-3.5kHz. This will mostly remove the bump. Then the midwoofer can be adjusted to lower the bump at 1.5kHz. The best method is probably move the 4 ohm padding resistor before the zobal network and change the zobal resistor to 7.5 ohm.
I also a set of EPI 500s and I need to measure and see if I can improve. That'll be a fun project. I'll probably sell whichever set of speakers that I like less.
Last edited:
Obviously up to you, but in my opinion a 2nd order filter would probably be fine here. Speaker building is a game of compromises, and that compromise is extremely minor here. I'd at least try it, this speaker is going to be hard enough without holding to theories that may or may not apply.
Also, if you do have a mic you might be able to fix that tweeter. I'd measure off axis (whatever degree will be facing the listener) and aim for a flat response there. Not sure how much efficiency you have to play with in that tweeter, but a wide notch that pads the tweeter from up to 6khz might help.
But I'm just "shooting from the hip", this is why these would be fun to play with.
Last edited:
Look at the electrical crossover for the midwoofer. The zobal is skewing the midwoofers response to the right.
Also look at the SPL chart. the Zobal is causing a dip around 4kHz and you can see that in this measurement. On axis measures about 76 db at 4kHz.
I'll try adding a 2nd order to the tweeter but I think that won't work as well as a notch filter and a 1st order XO. I think the midwoofer's circuit must be adjusted to flatten out the response.
I measured the speakers today. I measured the speaker 45° off axis, so on axis with the dipole, and with a towel covering the other dipole. My room isn't acoustically the best so below 300hz isn't accurate. But I did use a calibration file with my microphone and my response matches that anechoic measurement above from 600hz to 5kHz. So my measurement probably is accurate.
