Crossover v3
I don't like that 3-4kHz directivity dip and power response hump. So maybe there is still work to do on this one..
What is not helping with things is the very small waveguide I am using. Ideally I should have gone for a 5inch waveguide for this bass/mid driver. The main reason why I went with a smaller size wavecor 4.5insh mid in the first place. But for this application, I won't be able to use it because of its even lower low end extension..
I don't like that 3-4kHz directivity dip and power response hump. So maybe there is still work to do on this one..
What is not helping with things is the very small waveguide I am using. Ideally I should have gone for a 5inch waveguide for this bass/mid driver. The main reason why I went with a smaller size wavecor 4.5insh mid in the first place. But for this application, I won't be able to use it because of its even lower low end extension..
I like this one better 'electrically' at least, I don't have so much experience what can be heard with small directivity changes. Impedance in the treble is not too low, and you potentially lower distortion a little with the impedances the drivers can 'see'.
Most transistor amps are not sensitive to high impedance around XO (but tube amps usually are), but low impedance like the 2ohms you had before can be problematic. It can also affect the FR of class D and tube amps.
Could you try to simulate an impedance flattening circuit on the woofer terminals to flatten the peak at resonance? You usually end up with something like 10-15mH and a couple of hundred uF + a total resistance close to the DCR of the woofer. You can use thin wire air core inductor (coil DCR will usually be close enough to the resistance you need) and electrolytic caps, so it will not be too expensive.
The impedance of the woofer reacts with the impedance of large coils in the XO causing boost and GD, similar to the big cap you had in series before. If you add the impedance correction and then simulate FR with and without it, you will see what happens, and how much in your case. I always liked the bass better subjectively using the correction, even if it was just a small diff in FR.
Most transistor amps are not sensitive to high impedance around XO (but tube amps usually are), but low impedance like the 2ohms you had before can be problematic. It can also affect the FR of class D and tube amps.
Could you try to simulate an impedance flattening circuit on the woofer terminals to flatten the peak at resonance? You usually end up with something like 10-15mH and a couple of hundred uF + a total resistance close to the DCR of the woofer. You can use thin wire air core inductor (coil DCR will usually be close enough to the resistance you need) and electrolytic caps, so it will not be too expensive.
The impedance of the woofer reacts with the impedance of large coils in the XO causing boost and GD, similar to the big cap you had in series before. If you add the impedance correction and then simulate FR with and without it, you will see what happens, and how much in your case. I always liked the bass better subjectively using the correction, even if it was just a small diff in FR.
Phase matching suffers but I am not so sure about vertical dispersion. Having a solid lobe tilting up a little is not a bad thing, speakers with a patten like this often sound quite good all around the room and standing up.
Unfortunately for us, when we have two competing passive crossover designs that are this close, it is very hard to predict which one we will prefer based on graphs and plots. To be sure, we have to build both filters, and evaluate them by ear. I have been in your position, and it is a hard choice.
People who design a lot of passive crossovers usually have a large collection of inductors, caps, and resistors. They have tools and techniques to connect up a prototype filter (i.e. bread board) to evaluate. I was recently made aware of this device which I find very helpful, Wago lever nuts, and the similar product from Ideal.
@wolf_teeth might offer some thoughts on this... I do not have a large collection of passive component parts, so I just make a best guess, and order enough extra parts to "trim" the main reactive components.
Thanks for the detailed explanation. It helps a lot. Took me a while to come back and read it because I was trying (and failing) to measure woofer/tweeter diaphragms to model in ATH. My brainpower was used up on that for a while. I can see how the ATH/BEE simulation is a huge time saver in getting one to a point where they can fine tune the resistance damping. Viewing your simulations against your real life measurements is excellent information. 👍 I don't think anyone else has done a matchup of simulation to real life measurement iterations like that in a post. It develops intuition on how to use the simulation to achieve a real life result.
I had been curious about your termination around the coax frame. I didn't know if it was just cosmetic or if you had also planned it as a cone/waveguide termination. Thanks for explaining that. It's also interesting how you rounded the enclosure into a teardrop shape because I had no idea how that would impact the cardioid radiation pattern. Now we can look at your measurements and see what it does.
Personally, I'm grateful when a DIYer explains details in their design iterations. Especially when followed up with measurements and subjective listening impressions. It's hard to get that from commercial designs and online reviews. Commercial product measurements and reviews are helpful but you only see a final product not the iterations in the design process.
You can simulate the filters as EQ
https://www.diyaudio.com/community/threads/vituixcad.307910/post-6909614
I think that is the best we can do, but it is still a really good approach. If I was starting with a DSP filter simulation for a driver, I would save the response as an *.frd file, and then use that response as a target function in the optimizer tool. Then I would guess at a filter topology and let the optimizer crank out some values. I usually start with 2nd order topology, and then 3rd order, and see which optimized response seems best. Then I try other filter elements like notch filters and zobels, etc. I often copy the topology that others have used successfully.
I don't beleive so.
There is this though
https://xmachina-ai.blogspot.com/p/xmachina-is-passive-crossover-planning.html
These are pretty useful in complicated wiring situations too, I have quite a few in my line arrays. Knock off Chinese versions are available from ebay for considerably cheaper than the official wago ones, and I can honestly not tell the difference with the real ones that I got from element14. The only difference was that a few were mising the actual metal bits inside, so order more than you need.
Attachments
I have not been able to work on speaker projects for a while now.. But when I got a little time today, I completed a first cut version of the SB15CAC-8 + SB26CDC on waveguide TV speaker.. 
I have been contemplating about buying the passive crossover components for a while but never got time to do it yet. I suddenly realized that I had one of these DSP amps lying idle.. https://zoudio.com/product/aio438/ (a perfectly useful product for a low SPL DSP crossover 2 way system)
I just made a temporary enclosure for it using a plastic box, wired it all up, dialled in a quick DSP crossover and here they all are
TV is connected to the amp via Bluetooth. I have had no lip sync issues with Netflix, youtube etc and it is all working perfectly well..
I never thought wireless would be so convenient.. (though it my day job making such products )
Need to clean up the cable clutter all around soon though since it is the living room..
Crossover v1
A quick 4.5ms gated measurement of a single speaker from 1m away
A measurement of the left and right speakers from MLP (2.5+ m away)
I am happy with this entire setup's sound as of now..
I have been contemplating about buying the passive crossover components for a while but never got time to do it yet. I suddenly realized that I had one of these DSP amps lying idle.. https://zoudio.com/product/aio438/ (a perfectly useful product for a low SPL DSP crossover 2 way system)
I just made a temporary enclosure for it using a plastic box, wired it all up, dialled in a quick DSP crossover and here they all are
TV is connected to the amp via Bluetooth. I have had no lip sync issues with Netflix, youtube etc and it is all working perfectly well..
I never thought wireless would be so convenient.. (though it my day job making such products
)Need to clean up the cable clutter all around soon though since it is the living room..
Crossover v1
A quick 4.5ms gated measurement of a single speaker from 1m away
A measurement of the left and right speakers from MLP (2.5+ m away)
I am happy with this entire setup's sound as of now..
You are wise to no incorporate any Linkwitz Transform to extend the bass response of the SB15 driver. In my experience, any bass extension EQ on this driver creates a sloppy "muddy" sound... a noticeable decrease in clarity. The midrange clarity of this driver is so high that it seems like a crime to reduce it...
Nice project !
Nice project !
though it my day job making such products
Need to clean up the cable clutter all around soon though since it is the living room..
Thank you Vineeth!
Once you go wireless, DSP and class D there’s no way back!
Humans we tend to value convenience and low cost…