Need help replicating this XO in VituixCad

[bunkbail]

Yeah the JA stuff is beyond what I can afford right now. But to be honest although I can afford them, at its current price a pair of Solstice also stretches my budget quite a bit. Just realized that Erse is no longer stocked anywhere, US coils inductors looks to be similar enough though I'm not too sure if they can be used to replicate the author's intended freq response.

 

[killa25]

The author made a few claims without much concrete evidence. The only reason I was intrigued by the speakers is that the author was the chief engineer of Joseph Audio and the engineer behind the infinite slope topology implemented in Joseph Audio speakers. He also claimed that his modified Solstice sonically rivals his set of the top-of-the-line Joseph Audio Pearls. I've not heard the Pearls, but the cheaper Joseph Audio Perspectives were one of the most beautifully sounding speakers I've ever heard.


So I've looked into this a bit and apparently K means coupling factor of the inductors and K value of 1 means perfect coupling between the inductors. Transformers specifically made for audio use typically has very high K value but this is not desirable in this topology since lower K value resulted from using 2 identical inductors can achieve steeper filtering slope and help mitigate distortion induced by the use of iron core inductors. The calculation of K value is over my head however. Fortunately in the article the author shared on how to mount the inductors (ERSE 18 AWG) with the exact spacing and orientations in order to get the desired K value.

I wouldn't doubt this kit might rival the Josephs but not necessarily because of the reworked xo improving the sound. I would try to find a pair with the original xo and give them a listen. Are there any measurements for the reworked xo that show crossover slopes and phase?
Using two inductors to reduce hysteresis might work but you could just use air cores. I don't see why the slope would be any different other than the fact that placing them close together might change their value a little (not sure on that, would have to measure).

 

[kimmosto]

2.5-way is very okay also as passive IIR. Phase shift at mid-range could look quite strange in vertical plane, but it can be real benefit compared to 2-way MTM or MMT due to smoother power spectrum. Original XO smells like old-fashioned flat on-axis forgetting everything else.

 

[killa25]

Original XO smells like old-fashioned flat on-axis forgetting everything else.

I am sure there was more thought put into it than on axis response. Jeff was a very well versed crossover designer and I am sure he took off axis response into account. If you knew the guy you would know he always tried to take into account all the tradeoffs of crossover design.
To add a little more perspective to the designers knowledge he created his own software for speaker design http://audio.claub.net/software/jbabgy/jbagby.html

 

[Rick Sykora]

2.5-way is very okay also as passive IIR. Phase shift at mid-range could look quite strange in vertical plane, but it can be real benefit compared to 2-way MTM or MMT due to smoother power spectrum. Original XO smells like old-fashioned flat on-axis forgetting everything else.

Thanks Kimmo!

While I get the power response advantage of the 2.5 way, just have trouble with one woofer effectively being passively driven at lower frequencies. My BGs are small midwoofers and do not produce enough low bass to not power one of the woofers for more bass...

On the Bagby crossover, it does look traditional and does not help that he did not post some off-axis measures, but he was mindful of such things. Am not sure how weel his tool might sim, but with due respect to Jeff, your VCAD software leaves his (admittedly useful) spreadsheet effort in the dust.

Hope you have a happy and heathy New Year!

 

[killa25]

Thanks Kimmo!

While I get the power response advantage of the 2.5 way, just have trouble with one woofer effectively being passively driven at lower frequencies.

Hope you have a happy and heathy New Year!

In a 2.5 way the woofers are both driven with power at low frequencies. At higher (mid) frequencies one woofer is making the majority of sound.

 

[Rick Sykora]

In a 2.5 way the woofers are both driven with power at low frequencies. At higher (mid) frequencies one woofer is making the majority of sound.

yes, thanks. my mistake as I see the lower woofer has the additional high value inductor to reduce the higher frequencies to it. Need to go review my other crossover. They may be miswired.

 

[Rick Sykora]

@killa25 do you have any decent Soltice measurements? Notably for individual drivers and/or off-axis?

 

[kimmosto]

On the Bagby crossover, it does look traditional and does not help that he did not post some off-axis measures, but he was mindful of such things. Am not sure how weel his tool might sim,

Traditional box shape without rounding and wave guiding is problematic, but of course balancing could be some compromise between flat on-axis and linear or otherwise smooth power response. It will not cure e.g. sound stage problems to different listening distances and toe-ins due to strong diffraction.
Two 6" mid-woofers is not impossible with conventional tweeter, but steep HP slope indicates that LP slope is steep too and some (basically unnecessary) tradeoff is required. DI probably has quite clear S-curve. That's nothing new with rapid, easy and cheap DIY kits.

My diy and simulator programming hobby was on break when Jeff's tools were most popular in North America. In the beginning of 2k when I rebooted this bobby I skipped Jeff's and other tools by FRD consortium. Mostly because Excel-based UI is terrible imo, performance is low, project handing and version control including new features and possible add-ons are locked to worksheet dependent on third party application. So those tools were absolute no, and I started to rewrite my own simulators. Versions 0-1.1 were much closer to LspCAD 5.25 Pro.
Jeff's tools have some nice mathematical features which were worth to "borrow" to VituixCAD. For example minimum phase exctraction and equations for filtering diffracted response with rounded edges. In the other hand, some features such as simulated off-axis response is not correct due to limited edge model so diffraction simulation is usable to quite narrow on-axis sector, and close to 90 deg is clearly wrong. Fortunately off-axis 0-180 deg is not very usable in VCAD either so not much harm done.