4-Way Active Cardioid Speaker Build

[montyliam]

Hi guys, I’d like to share my outline for a 4-way active full-range loudspeaker and get your feedback (criticisms, concerns, improvements etc). This is my first attempt at a cardioid design, so I’m starting with more affordable drivers to try and learn the process.

Drivers
KEF Q150 coaxial → midrange / high-frequency (MF/HF).
Peerless FSM-0512 → mid-low (100–300 Hz).
Peerless GBS-200F35CP02-04 → side woofers (20–300 Hz, dual opposing).

Cabinet
Dimensions: 220 mm W × 400 mm H × 400 mm D.
Front baffle: KEF coax + FSM-0512, aligned vertically.
Side panels: GBS drivers and placed as close to the front baffle as possible.
Interior: The KEF coax and FSM driver will each be housed in their own circular enclosure using heavy duty postal packaging tubes. The closest distance the GBS drivers can be to the front panel therefore is around 165mm. I am not 100% confident how to calculate the exact internal volume, but I estimate around 15-20L after the internal parts are taken into consideration.
A generous 35mm round over on the front side edges to control diffraction effects.

Crossover Plan

• KEF HF ↔ KEF MF: ~2 kHz.
• KEF MF ↔ FSM-0512: ~300 Hz.
• FSM-0512 runs 100–300 Hz.
• Side GBS woofers: 20–300 Hz (to create cardioid in the 100–300 Hz band).

I aim here to achieve cardioid directivity in the 100–300 Hz region. Below ~150–200 Hz is where the cardioid effect will naturally weaken due to cabinet depth.

Amplification

• 2 × TPA3250 stereo boards at 100wrms 4ohm per channel
• One for KEF HF + KEF MF.
• One for FSM-0512 (1 channel unused).
• 1 × TPA3255 stereo board at 160wrms 4ohm per channel for side GBS woofers.

DSP & Measurements

• DSP/DAC: Okto DAC8 initially (later possibly a dedicated FIR DSP board).
• Measurement setup: Dayton EMM-6 mic + Steinberg interface.
• Spinorama jig (5° increments) for precise off-axis measurements.
• Initial XO work in VituixCAD; may explore FIR filters if IIR isn’t enough.

Questions

• Drivers – Should the side GBS woofers really run 20–300 Hz, or would a dedicated sub <100 Hz be better to preserve cardioid performance in 100–300 Hz? My worry is that due to high excursion during high SPL listening levels, the cardioid effect will weaken.
• Cabinet volume/tuning – With ~25 L internal volume, is sealed the best starting point for the FSM and GBS, or should I consider vented?
• Amp headroom – Will the TPA3250 modules be enough for the FSM and KEF sections, or should I consider more powerful boards for future scaling?
• DSP approach – For cardioid null tuning, would you recommend starting with simple delay + gain EQ in IIR, or going FIR early?
• Power supply: I am unsure what will be suitable. Is it acceptable to have 1 supply per speaker running all 3 amp boards, or is it better to have 1 per board or 1 per 2 boards etc?
• FIR DSP chip. If I find the Okto insufficient, are there any FIR capable DSP chips which could be integrated into this design?

 

[staticV3]

For a low cost system consider Wondom DSP Amp boards: https://store.sure-electronics.com/products (type "DSP" into the search bar).

There's a 4x100W board, or you can chain multiple boards together using i2s.

Each one has an ADAU1701 chip and can be configured in Sigma Studio.

For programming I'd advise against Wondom's own programmers which have been flaky for me.

This one is supposedly much better: https://www.tinyosshop.com/usbi-dsp-programmer

 

[Ktacos]

You'd probably have better luck not running bass to the side woofers and just using them to get cardoid dispersion for everything above the bass. I'd argue at that point you could consider cardoid with resistive ports on the side ala D&D 8c. Saves on driver and amp costs for effectively the same result.

 

[montyliam]

You'd probably have better luck not running bass to the side woofers and just using them to get cardoid dispersion for everything above the bass. I'd argue at that point you could consider cardoid with resistive ports on the side ala D&D 8c. Saves on driver and amp costs for effectively the same result.

I really don't want to go down the passive cardioid route due to the amount of testing and re-testing required. I was potentially thinking about having 3 of the Peerless FSM drivers and one subwoofer on the rear of the cabinet. So 1 Peerless on the front running 100-300hz, two on the sides running 100-300hz out of phase cardioid, and a subwoofer on the rear running 20-100hz.

I'm not 100% sure how KII operate their cardioid setup, but it appears the side and rear woofers are running both bass frequencies and supporting the cardioid directivity. This certainly seems to be the case with the new Kii Seven, where only side woofers are seen but the product performs both full range bass and cardioid bass down to a certain frequency. This will also be similar to the Mesanovic CDM65 where the side woofers operate both as subwoofers and help the cardioid directivity pattern.

Alternatively, I could use the 3 Peerless FSM and KEF coax in a one box and then build a bass extension stand below to extend the cardioid pattern.

 

[somebodyelse]

Tinysine DSP amps are another option. https://www.tinysineaudio.com/collections/audio-amplifier-with-dsp

 

[jaakkopetteri]

I'd argue at that point you could consider cardoid with resistive ports on the side ala D&D 8c. Saves on driver and amp costs for effectively the same result.

It's nice "to consider" for sure but passive cardioid is not easy. Finding the right combination of stuffing and enclosure dimensions can be difficult, where as drivers are fairly straightforward to tune.

And why exactly would you not run bass to the side woofers? What's luck got to do with it?

 

[Ktacos]

It's nice "to consider" for sure but passive cardioid is not easy. Finding the right combination of stuffing and enclosure dimensions can be difficult, where as drivers are fairly straightforward to tune.

I think this talk of testing is a bit exaggerated. It's really not that hard to get good cardoid performance without testing slot locations. My own testing showed shockingly good cardoid performance on the first try, just some slots behind the woofer. Something like the shivaudyo is literally just a bunch of holes behind the driver and it achieves exceptional performance. Gethain gets good cardoid performance with the most basic of geometry and placement of their slots. As far as dampening it's not that complicated either, fill the chamber with dense material. I used cotton insulation with good results. These are my personal findings, others are welcome to share their findings.

And why exactly would you not run bass to the side woofers?

Active cardoid tends to require sealed enclosures. OP's speaker format appears to be a bookshelf, albeit it a fairly large one, but still a small volume. One can model the peerless woofers in vcad and see that the low end roll off will be steep and early and require some EQ. With EQ you hit xmax really early. D&D use a much more capable woofer in the rear of their speaker and distortion is still quite high. It probably depends who you ask, but in my own testing I found these small volumes with high excursion to not sound nearly as good as a dedicated sub. Considering the bass will be omni, and how bass interacts with rooms, to me it makes much more sense to leave bass duties to some dedicated sub(s).

 

[PGansz]

I think this talk of testing is a bit exaggerated. It's really not that hard to get good cardoid performance without testing slot locations. My own testing showed shockingly good cardoid performance on the first try, just some slots behind the woofer. Something like the shivaudyo is literally just a bunch of holes behind the driver and it achieves exceptional performance. Gethain gets good cardoid performance with the most basic of geometry and placement of their slots. As far as dampening it's not that complicated either, fill the chamber with dense material. I used cotton insulation with good results. These are my personal findings, others are welcome to share their findings.



Active cardoid tends to require sealed enclosures. OP's speaker format appears to be a bookshelf, albeit it a fairly large one, but still a small volume. One can model the peerless woofers in vcad and see that the low end roll off will be steep and early and require some EQ. With EQ you hit xmax really early. D&D use a much more capable woofer in the rear of their speaker and distortion is still quite high. It probably depends who you ask, but in my own testing I found these small volumes with high excursion to not sound nearly as good as a dedicated sub. Considering the bass will be omni, and how bass interacts with rooms, to me it makes much more sense to leave bass duties to some dedicated sub(s).

Agreed. This needs to be a floor-standing cabinet with basic calculations suggesting >65L.

I would triple your volume to 81L.

 

[montyliam]

I should be clear the goal is cardioid directivity to as low as frequency as possible in an enclosure this size. I certainly don’t want to make the box much larger than this.

In terms of passive vs active cardioid, woofers are cheap and amplification is cheap too. If I follow the principle of the side woofer’s acoustic centres being as close as possible the the front woofer acoustic centre, this should allow for some experimentation in VITUIXCAD vs manual experimentation (drilling holes) with passive. I am 100% not interested in passive cardioid setup for the moment.

My main point of contention is around the side woofers and their frequency range. I plan to have the front FSM driver in around a 2l enclosure. It will be high passed around 100hz and low passed around 300hz (whatever works best with the coax). The side woofers will either run:
1. 20-300hz, controlling directivity in the 1-300hz range.
2. 100-300hz, only controlling directivity, and another driver in the rear of the cabinet to handle below 100hz.

If I go for option 1, I will use the GBS drivers on the side.
If I use option 2, I will use the FSM drivers, all high passed at 100hz to handover to the rear subwoofer.

Option 1 is the most attractive to me at the moment (4-way vs 5-way), but my main concern is how well the drivers will work in the 100-300hz ‘directivity controlling’ range if they also have to produce bass below 300hz which increases excursion. I think I may be unnecessarily worrying about this, as it is quite common to have drivers spanning this range anyway. It also should make integration a little easier.

For now, I’m ruling out the bass module idea. I want full range sound from a box this size with also cardioid directivity to as low as a frequency as the box size dictates. I am not expecting high-spl from this design, but the 8” GBS drivers are capable to produce quite a large amount of bass below 100hz for most rooms.

 

[montyliam]

I think this talk of testing is a bit exaggerated. It's really not that hard to get good cardoid performance without testing slot locations. My own testing showed shockingly good cardoid performance on the first try, just some slots behind the woofer. Something like the shivaudyo is literally just a bunch of holes behind the driver and it achieves exceptional performance. Gethain gets good cardoid performance with the most basic of geometry and placement of their slots. As far as dampening it's not that complicated either, fill the chamber with dense material. I used cotton insulation with good results. These are my personal findings, others are welcome to share their findings.



Active cardoid tends to require sealed enclosures. OP's speaker format appears to be a bookshelf, albeit it a fairly large one, but still a small volume. One can model the peerless woofers in vcad and see that the low end roll off will be steep and early and require some EQ. With EQ you hit xmax really early. D&D use a much more capable woofer in the rear of their speaker and distortion is still quite high. It probably depends who you ask, but in my own testing I found these small volumes with high excursion to not sound nearly as good as a dedicated sub. Considering the bass will be omni, and how bass interacts with rooms, to me it makes much more sense to leave bass duties to some dedicated sub(s).

Doesn’t passive cardioid tax the front woofer more than it would in a sealed enclosure? The woofer is basically free air at some frequencies with no back wave to support? Maybe my understanding is wrong here but the reason the D&D has higher distortion around 100hz is due to the front woofer (not the rear) having to use more of it’s excursion due to lost energy through the slots (vs if it was sealed).

The front woofer is not going to reproduce anything below 100hz, so the 5mm or so of XMAX seems sufficient to cover the 100-300hz range where excursion demands are low anyway.

 

[montyliam]

Another question, is it advantageous to go with as large a round over as possible on the front baffle? I can see the Kii 3 has a very large round over, but the D&D does not. My Genelec’s also have large round overs on all sides of the cabinet. For now, an r35mm radius is what is shown in the pictures. I could push this to r45mm round over but not much further without encountering cabinet construction issues (the round over begins to eat into the front and side panel join).

 

[Ktacos]

Another question, is it advantageous to go with as large a round over as possible on the front baffle? I can see the Kii 3 has a very large round over, but the D&D does not.

Roundover size kind of depends on what waves are actually making it to the edges of the speaker. With the D&D the waveguide is quite a bit larger than the DXT and loads the tweeter to a lower frequency, so much that a small roundover is enough to prevent edge issues. Kii's baffle extends out beyond the waveguide and the waveguide controls DI at a higher frequency, so you have to contest with not just the tweeters baffle interaction but the upper range of the mid as well which tends to require quite a large round over.

A speaker with similar tweeter loading to the D&D is stuff from Buchardt, they do hard edges and you can see the diffraction in the dispersion, A500 has a mild edge interaction at ~4.5khz.

Since your speakers design goals state using the q150 driver, you can look at the spins on that in the stock box and see that edge related issues start to occur around 2khz and mostly affect the woofer, so you'd need quite a large round over to deal with it, somewhere along a 2" radius.

With that said, pretty much every speaker seems to benefit from some sort of edge rounding even it's not quite as large as what would be ideal. I have the opinion that edge diffraction is one of the most detrimental issues a speaker can have (assuming the speaker is already neutral and has somewhat controlled DI). It is after all the creation of secondary sound sources that interfere with the initial sound. Getting rid of edge related artifacts tends to greatly improve areas where the input signal exhibits something closer to noise (the music gets really busy) and is more sustained, so things like reverb tend to be presented much more realistically on speakers with no edge issues.

My last experiment was small bookshelf with 3" radius round overs and it gave me the cleanest dispersion I've had the chance to hear, diffraction issues were moved way down to around 700hz. I never quite figured out how to make it look nice and actually live in a home so until I figure that out it will remain a prototype. That's a $14 USD tweeter, pretty amazing.

 

[Ktacos]

Doesn’t passive cardioid tax the front woofer more than it would in a sealed enclosure? The woofer is basically free air at some frequencies with no back wave to support? Maybe my understanding is wrong here but the reason the D&D has higher distortion around 100hz is due to the front woofer (not the rear) having to use more of it’s excursion due to lost energy through the slots (vs if it was sealed).

I would maybe give my original distortion post a quick glance again. I was referring to the issue of small sealed volume the side woofers have and asking them to achieve bass reproduction. It's a lot to ask and you hit xmax issues pretty quickly with the chosen woofers and volume after to apply EQ to get the low end back up. Most of the cardoid speakers we see out there that are all in one box solutions tend to have pretty high distortion figures in the bass, and they also give up the cardoid effect in the bass region. I feel it makes more sense to just let a traditional sub handle bass.