Passive Radiator CSS SDX10 Design and Build

[mwmkravchenko]

So in order to see what happens a way down low for some enclosures Hornresp can simulate down to 1 hertz.

This is your box as you have it mass loaded for your measurements.

 

[EngineerNate]

I'll likely be home this afternoon, I can screenshot my VitiuxCAD sim. Something still isn't jiving here. I would have expected the measured impedance peak/min from the box resonance to be way out of line with my sim if the above was matching up physically. I'm seeing the impedance min around 18hz as predicted by my sim. The frequency response is dropping off more from 40-20hz than expected, but I had attributed that to an error in my measurement setup or capability.

The measured impedance min being almost an octave above the sim seems like a big difference.

 

[EngineerNate]

This is what Vitiux is predicting for me, and the parameters I'm using.

Small Signal Parameters for the SDX10:

Large Signal:

 

[EngineerNate]

No added mass:

 

[EngineerNate]

My sim lines up with the Hornresp if I set extra mass to 920, which makes me think the horn resp sim is double/half counting something, somewhere:

My understanding of the VitiuxCAD interface is that all the parameters in the passive section are for each radiator, and the duplication is done by the code.

 

[EngineerNate]

Interestingly, with the added mass, I am showing 10hz as the effective resonant frequency of the radiator itself, but the resonant frequency of the box+radiator is ~18hz:

 

[EngineerNate]

Project File:

Dropbox

www.dropbox.com

 

[mwmkravchenko]

All of this is moot until you actually measure your drivers in free air.

Whoever wrote that blurb about XBL^2 drivers requiring so-called large signal Thiele Small parameters is not really thinking it through.

It is true that there is a null in the B Field.

Here is an example of a driver I designed back in 2012:

Notice that this motor is a rather well balanced two gap motor. Getting the forward, and rearward sides of an XBL motor equal is no small feat. That is attested to in the Voicecoil measurements. The rearward side is disproportionately stronger.

This is what you actually get across the two gaps and in practice. If you had a deep null, you would never get your voice coil moving as it is supposed to set at the centre height of the B field. ( magnetic filed in the voicecoil gap between the pole in the centre and the top plate )

So as you may already know the Voicecoil measurements are made in the low voltage method. I use those measurements to simulate.

Until you actually measure your drivers we are simply guessing. Cms is a massively variable specification depending on the spider manufacturer. We already know that the motor itself has a flux in the gap bias. So the motor will draw downwards more easily that outwards. That is a part of the reason for the high distortion figures measured in the Voicecoil article:

Test Bench: The SDx10 Home Audio Woofer from CSS Audio

The driver under test in this Test bench is the SDx10 home audio woofer from Creative Sound Solutions (CSS) Audio, using XBL^2 dual gap motor technology licensed from inventor Dan Wiggins. The CSS SDx10 also has a well-appointed feature set that includes a very stiff coated black paper/carbon...

audioxpress.com

So. I know the feeling of a new kid. It feels like yesterday that I held mine. Take your time. I'll help you through this when ever you have free time to do it.

 

[EngineerNate]

The woofer itself won’t change the resonant properties of the box though right? That’s the box volume + PR tuning and that’s what’s causing our mismatch here, I could put another driver entirely in there and aside from differences in displaced volume in the box, the resonant system of the enclosure would be unchanged.

The resonant peak of the woofer in all of the various sims/measurements has been reasonably consistently in the high 30s/low 40s Hz.

 

[EngineerNate]

@mwmkravchenko in your vituix sim where did these numbers come from for the APR10? They are substantially different from the stated specs for Vas, Fs and Qms.

 

[mwmkravchenko]

@mwmkravchenko in your vituix sim where did these numbers come from for the APR10? They are substantially different from the stated specs for Vas, Fs and Qms.

View attachment 526362

https://www.css-audio.com/online-store/CSS-APR10-CLOSEOUT-10-Passive-Radiator-p110031744

Sd: 320 cm^2


Qms: 7


Cms: 0.48 mm/N


Vap: 67 Liters


Mmp: 66 grams


Fs: 28 Hz


Xmech: 26 mm


Maximum Moving Mass: 1200 grams


Minimum Fs: 6.7 Hz (with full 1200g)

And I see an error. Vas is way off.

I'll check if I made the same error in Hornresp.

Yup same error.

I'll re-run the simulation:

Corrected Vas Simulation:

Much more realistic Impedance peaks too! The slip of my tiny fingers. ( Not tiny )

Well that's a lot better! Thanks for figuring out where I made a mistake!

Mark

 

[mwmkravchenko]

No weights again with the Vas correction.

My advice. As we are looking a lot more like your measurements now, ditch the weights and eq the box.

One way that you could do it.

Mark

 

[mwmkravchenko]

The woofer itself won’t change the resonant properties of the box though right? That’s the box volume + PR tuning and that’s what’s causing our mismatch here, I could put another driver entirely in there and aside from differences in displaced volume in the box, the resonant system of the enclosure would be unchanged.

The resonant peak of the woofer in all of the various sims/measurements has been reasonably consistently in the high 30s/low 40s Hz.

The box is tuned via your PR's. The woofer will not change the properties of this system. Your resonance is most likely a woofer resonance as it keeps pretty consistent.

Taken from the Voicecoil article. It is indeed a woofer resonance.

This is a simulation in one of the better simulation programs available. LEAP. Again so accurate that Vance stopped making physical boxes a long long time ago. There was no point. the simulation was just as accurate.

Mark

 

[EngineerNate]

What’s the advantage to ditching the weights and tuning the box with EQ?

 

[mwmkravchenko]

What’s the advantage to ditching the weights and tuning the box with EQ?

You aren't getting useful output below 40 hertz. So losing the weights gets you efficiency.

Mark

 

[EngineerNate]

You aren't getting useful output below 40 hertz. So losing the weights gets you efficiency.

Mark

Gotcha.

I’ll probably fiddle with the weights and take more measurements in a better space, starting with none and adding one at a time and measuring until I get the most ideal actual response out of my specific drivers.

I’m ordering an EasyZ CV100 from Scott Hinson to make taking better measurements at various power levels easier

 

[EngineerNate]

@mwmkravchenko may have found part of the reason I didn’t have spl down low. Drivers were leaking enough that at low hz and high spl it was making audible noises. I’ve sorted one out.

 

[mwmkravchenko]

@mwmkravchenko may have found part of the reason I didn’t have spl down low. Drivers were leaking enough that at low hz and high spl it was making audible noises. I’ve sorted one out.

The ripples in your impedance test showed leaks. At least the possibility of leaks. As for lacking deep bass, well that's a part of the system design. System starting with the driver. It has strong output to 40 hertz. But not a lot below. You can' raise that output via PR.

 

[EngineerNate]

Much better.

 

[mwmkravchenko]

View attachment 526913View attachment 526914View attachment 526912

Much better.

Very much better!