2-way with wide dispersion and low distortion. Linear CTA-2034. Mechano26

[XMechanik]

Another take on a small two-way monitor design.

While searching through various datasheets, I came across a driver setup that seemed to allow for directivity matching between the tweeter and woofer without using a waveguide. This should allow for good control of the CTA-2034 characteristics as well as wide dispersion (edit: more details of the idea in post #24 link).

The drivers:
SB15MFC30-4
SB26CDC-C000-4

18 mm birch plywood enclosure, dimensions: 314 x 184 188 x 278 mm HxWxD.

I took a complete set of measurements in both the horizontal and vertical planes. The driver choice turned out to be valuable, filter synthesis yielded the following characteristics:



I built prototypes according to the design. There was a small problem however: the measurements taken on the built speaker didn't quite match the simulations. A quick check of the connections revealed nothing. But it turned out that the characteristics could be restored by modifying some of the crossover component values. And that's where I left it for now, although the issue hasn't been fully investigated: entering modified values into the simulation again resulted in some discrepancies. But the measurements are now quite correct:


I had some scraps from 45-degree cuts. I glued them together and used them to brace the enclosure from the inside.


THD distortion:

Measurements were taken at 1m on axis. The graphs are normalized to SPL. At 86 dB, midrange distortion reaches a maximum of -49 dB relative to SPL, which translates to a percentage value of 0.35%. As the signal level increases to 96 dB, distortion locally reaches -41 dB, or 0.9%, still quite ok.

Filters (component values after the modifications)



Front panel diagram (dimensions in mm)

 

[eddantes]

Nice! What's the F3? Any idea of max SPL? Both Z at 0? I'd say woofer is at +10 to the tweeter.

 

[Ktacos]

I don't need to build another speaker, I don't need to build another speaker.....

Let me see what kind of wood I've got lying around. I have the aluminum version of this tweeter, specs are nearly identical so I think it should be a safe bet? Possibly requiring a minor adjustment of resistance. I didn't see a link for the vcad project on your site for this one, any chance we could get access to that?

 

[silverD]

I haven't even assembled my 23s yet....

 

[kyuu]

Just FYI, your links don't work, at least for me. All they do is result in the website trying to force me to create a blog with no way to get to the linked content as far as I can tell.

 

[Newman]

Just FYI, your links don't work, at least for me. All they do is result in the website trying to force me to create a blog with no way to get to the linked content as far as I can tell.

Ditto, @XMechanik

 

[Ktacos]

That woofer circuit is wild. I"m basically guaranteed to wire it wrong the first time

 

[a4eaudio]

...All they do is result in the website trying to force me to create a blog...

Haha, maybe it is a sign that you should start a blog!

 

[XMechanik]

I didn't see a link for the vcad project on your site for this one, any chance we could get access to that?

I hasitaed to release this as there is some discrepacy between the simulation and the finished speaker measuremet. Faulty measurements of the drivers could be one of the reasons so I was considering taking the meausrments again. But maybe it's not a bad idea to share it "as is" and see if there are comments what could be the issue. All I need is to do some orders in subfolders on my hd.

 

[XMechanik]

Just FYI, your links don't work, at least for me. All they do is result in the website trying to force me to create a blog with no way to get to the linked content as far as I can tell.

Fixed.

 

[XMechanik]

Nice! What's the F3? Any idea of max SPL? Both Z at 0? I'd say woofer is at +10 to the tweeter.

I ususally look at f6 or f10 considerng room gain effect. According to simulations f6 should be 42Hz but it's actually around 55Hz which is pretty disappointing. If the assumed value was reached this could be possibly my 'end game' bookself.

xyz should reflect the coordinates of the centers of the H and V measurement orbits, not the acoustic centers of the driver (at least that's how I get it). In that regard both centers were aligned with the front panel, hence z=0.

 

[eddantes]

I ususally look at f6 or f10 considerng room gain effect. According to simulations f6 should be 42Hz but it's actually around 55Hz which is pretty disappointing. If the assumed value was reached this could be possibly my 'end game' bookself.

xyz should reflect the coordinates of the centers of the H and V measurement orbits, not the acoustic centers of the driver (at least that's how I get it). In that regard both centers were aligned with the front panel, hence z=0.

Thanks for your reply. I just built a small pair - but my Vcad sim and measurements are really off. Speaker is still a good one - but as the sim and reality don't match - I condsider it a fail. I was hoping to learn from this project on how to get Vcad to be closer to reality.

 

[a4eaudio]

I was hoping to learn from this project on how to get Vcad to be closer to reality.

There used to be a guy on diyaudio who would jump into every speaker build thread to claim that "crossover design is easy, measurements are hard." It took me awhile to appreciate what he was saying. I still don't agree that crossover design is easy, but I agree that precise, high-quality measurements are hard. I have had many cases where my simulation was not consistent with the final result - whenever I have gone back and done a Post-mortem, it was always user-error on my part somewhere in the process.

Below is a recent speaker where I took full Spinorama measurements of the completed speaker at 2 meters. Blue lines are VituixCAD simulation, red lines are the actual speaker. Of course, I didn't save screenshots of the cases where I messed up and it didn't match. I would think a good portion of that final variability is that I did not actually measure capacitance, inductance, and DCR of the actual crossover components to put into VituixCAD.

 

[eddantes]

There used to be a guy on diyaudio who would jump into every speaker build thread to claim that "crossover design is easy, measurements are hard." It took me awhile to appreciate what he was saying. I still don't agree that crossover design is easy, but I agree that precise, high-quality measurements are hard. I have had many cases where my simulation was not consistent with the final result - whenever I have gone back and done a Post-mortem, it was always user-error on my part somewhere in the process.

Below is a recent speaker where I took full Spinorama measurements of the completed speaker at 2 meters. Blue lines are VituixCAD simulation, red lines are the actual speaker. Of course, I didn't save screenshots of the cases where I messed up and it didn't match. I would think a good portion of that final variability is that I did not actually measure capacitance, inductance, and DCR of the actual crossover components to put into VituixCAD.

View attachment 510653

Wow that looks great! I'd be chuffed to bits if my result was this close.

Actually - for this past project, I did measure the DCR of all the components, and I adjusted the model to reality as I built the cabinet and baffle - and still - wrong. The worst of it is that Vcad design was for an inverted polarity on the tweeter; but subsequent measurements revealed that normal polarity was required.

Gotta be honest - after finding out that fact - I kinda gave up on them. Yeah they sound ok; nice even - especially as they were purpose built for small appartment use for my daughter's boyfriend - and the elevated midbass makes for good low volume listening... but that wasn't my goal.

 

[AnalogSteph]

According to simulations f6 should be 42Hz but it's actually around 55Hz which is pretty disappointing.

Any clues as to who's to blame? There's not exactly too many candidates, either the woofers substantially deviate from the published TSPs or effective box volume / BR tuning are off.

 

[a4eaudio]

Any clues as to who's to blame? There's not exactly too many candidates, either the woofers substantially deviate from the published TSPs or effective box volume / BR tuning are off.

It could also be the crossover - specifically the baffle step compensation - and/or the merge process of farfield and nearfield. Since Xmechanik mentioned that his final measurements were not matching up with the VituixCAD simulation, these shouldn't be ruled out.

Here is an example of F3 being 60 Hz or 48 Hz but the box modeling would have been correct for everything below 150Hz.

 

[a4eaudio]

That woofer circuit is wild. I"m basically guaranteed to wire it wrong the first time

I think I have the component values correct. I am "schematically challenged" and would really struggle with the original layout.

 

[Ktacos]

That is much easier to read but you underestimate my ability to wire things wrong

 

[TurtlePaul]

If you are going to have an 18 part crossover, why not just buy a better woofer?

 

[XMechanik]

If you are going to have an 18 part crossover, why not just buy a better woofer?

With fs of 35.5 Hz, Vas19l, and Qts 0.28, it seemed like a very good woofer choice for a small bookshelf.

What's even more imortant (as I mentioned in #1), I've noticed from the datasheet that this woofer would likely be paired with a tweeter without a waveguide with overall smooth directivity and wide dispersion.