DIY Cardioid Designs?

[617]

The recent review of the Genelec 8341 SAM has really affected my thoughts on the design of smaller speakers. This, along with a few other cardioid designs, shows performance which I never dreamed possible in a domestic speaker, especially one of reasonable size. The Genelec is a tiny speaker, under 10x10x13, offering directivity control down to 500hz. Other stand mount cardioids have even lower directivity control at the expense of slightly larger size, although they are presumably a bit better with regard to output levels.

One might say that ideal speaker controls directivity down to the Schroeder frequency of the room they are in, and these cardioid designs basically fulfill this ideal. This is a big deal, as previously the only designs which could even approach this performance were really big, typically involving big horns, big center to center spacing between drivers and large expense. These cardioid speakers on the other hand manage incredibly smooth directivity with only shallow waveguides and conventional audiophile class drivers.

I used to be better informed on the DIY speaker scene, and right now I aware of only very few cardioid DIY efforts. I would be curious if anyone here has any information on DIY projects which try to capture some of this performance. I am not familiar with the design tradeoffs involved with cardioid midbass loading, so any information in regards to this issue would be helpful.

One resource is Kimmo Saunisto's page. Kimmo is the developer of VituixCAD and has done some work on cardiod loading of bass:
https://kimmosaunisto.net/CardSub/CARDSUB.html

 

[thewas]

If you go the active way making a cardioid in the lower bass isn't that difficult, you can put two woofers looking to the front (as you need them to be louder than the cancelling rear one), one to woofer looking to the back (in separate housings) and give the rear one the same signal as the front ones, just with opposite phase and delayed by the sound propagation time that corresponds to the depth distance to the front ones. That's actually a simple way that is often used in the PA world and called cardioid subwoofer array (CSA) https://en.wikipedia.org/wiki/Subwoofer#Rear_delay_array

With a passive crossover its more difficult because you don't have such a constant time delay element, but you can change the phase behaviour with a special topology and size enclosure, here is a German example which you can also simulate in the free loudspeaker simulation software boxsim
http://www.roomaudio-bassgun.de/index.php?r=site/diy
and here is a general method how to do so (sorry, also in German language) http://www.hifi-forum.de/viewthread-104-18886.html

 

[Biblob]

There is a thread on here which goes into Constant Directivity Speakers. The user @Jag768 posted his work on a D&D 8C version. He also has a thread on diyaudio.com and the dutch zelfbouwaudio.nl

 

[617]

If you go the active way making a cardioid in the lower bass isn't that difficult, you can put two woofers looking to the front (as you need them to be louder than the cancelling rear one), one to woofer looking to the back (in separate housings) and give the rear one the same signal as the front ones, just with opposite phase and delayed by the sound propagation time that corresponds to the depth distance to the front ones. That's actually a simple way that is often used in the PA world and called cardioid subwoofer array (CSA) https://en.wikipedia.org/wiki/Subwoofer#Rear_delay_array

With a passive crossover its more difficult because you don't have such a constant time delay element, but you can change the phase behaviour with a special topology and size enclosure, here is a German example which you can also simulate in the free loudspeaker simulation software boxsim
http://www.roomaudio-bassgun.de/index.php?r=site/diy
and here is a general method how to do so (sorry, also in German language) http://www.hifi-forum.de/viewthread-104-18886.html

It seems like the active approach is the only viable one for a compact-ish speaker. It's a shame there aren't more DSP/amp products available for DIYers, although the hypex ones are probably good enough.

 

[amadeuswus]

Hi 617,

Your post made me think of forum member Juhazi, who has made his own version of the Gradient 1.4, which I believe is cardioid down to 200hz or so (if I recall correctly). The Gradient speaker I am familiar with, the Revolution, has a compact coaxial head unit with flow vents on each side to create the cardioid pattern.

 

[KLang1]

It seems like the active approach is the only viable one for a compact-ish speaker. It's a shame there aren't more DSP/amp products available for DIYers, although the hypex ones are probably good enough.

What's wrong with using a DSP meant for PA? All we need is a delay, or am I misunderstanding this whole thing?
Does it have to be a FIR DSP to delay the signal in equal phase for every frequency?

 

[617]

What's wrong with using a DSP meant for PA? All we need is a delay, or am I misunderstanding this whole thing?
Does it have to be a FIR DSP to delay the signal in equal phase for every frequency?

I don't believe FIR would be necessary for a cardioid design; just EQ and delay.

 

[Rick Sykora]

Am not all that familiar with Genelec, but the closest thing to a DIY cardioid (that
I know) is what has been done by linkwitzlab.com. See: http://www.linkwitzlab.com/models.htm

 

[HammerSandwich]

Read the OP's link, then click "home" at the bottom and be amazed. Not all of Kimmo's speaker projects are cardioid, but he's built a bunch.

 

[Juhazi]

Search "diyaudio.com cardioid keysir" and you will find early stages of D&D 8c

The challenge is how to match that cardioid 2,5 octaves to low bass and treble. Normal boxed speakers are almost cardioid above 1kHz naturally. DSP helps but physical dimensions vs. wavelength are limiting factors.

 

[617]

Search "diyaudio.com cardioid keysir" and you will find early stages of D&D 8c

The challenge is how to match that cardioid 2,5 octaves to low bass and treble. Normal boxed speakers are almost cardioid above 1kHz naturally. DSP helps but physical dimensions vs. wavelength are limiting factors.

Amazing how much innovation comes from DiyAudio, it's a shame there's so much garbage there.

 

[Biblob]

Amazing how much innovation comes from DiyAudio, it's a shame there's so much garbage there.

Yes, I find the threads so hard to follow as well. The appearance of ASR makes it a lot more pleasurable to read and post than on other forums. It's my favorite so far. I don't really understand why diyaudio.com won't/hasn't be(en) updated to something more recent.

 

[q3cpma]

You could try to dissect an ME Geithain and see what magic they use to do it passively down to 30 Hz. There's a technical interview about it in https://www.me-geithain.de/en/assets/media/products/studio/RL901K/en/PPThomsen-RL901K-02-05-eng.pdf

Production Partner: Mr. Kiesler, how do you achieve cancellation of waves with
wavelength in meter range, and this not only for a few discreet Hz but within a
1:10 bandwidth? After all, we talk about wavelength, which extend between 1 and
10 meters.

Jochen Kiesler: Well, for sure not by deceiving physics. That won’t do. To sum
it up, the trick was to attach an acoustic phase-rotation line at the back of a
loudspeaker which produces a phase-shift of 160 degrees. Thereby both, the
phase and the amplitude must be kept constant within 5 degrees and – this is
the biggest problem – irrespective of pressure conditions. After all, the range
of pressure inside the box differs from 25 to 125 dB and for all that, the
specific flow resistance has to be the same.

PP: In other words, the flow resistance causes a wideband delay?

J. K.: You could say so. We use a combination of several materials which I
would not like to discuss in detail now. Inside the loudspeaker box RL 901K
there are two separated cabinets which close up with these flow resistances in
their back. The dimensions of the flow resistance and of the membrane
correspond to each other and if you choose the correct combination, it works
for frequencies down to 30 Hz.

PP: Why hasn’t anybody come up with this idea before?

J.K.: Well, there has not been any lack of attempts. Some approaches use two
subwoofer set up against each other – actually a plausible idea. In practice
however, you would need two high-quality low frequency drivers with an
amplifier and the matching enclosure for each. Therefore it never became a
commercial product. Whereas we have some experience with constructing
loudspeakers which emit a cardioid characteristic down to 100 Hz from working
with PA- systems. One day we asked ourselves if we could extend this principle
to still lower frequencies even if there was no theoretical basis for it.

PP: Delay actually implies that the sound velocity is reduced in the acoustic
line. Is this really true?

J. K.: Yes. It is reduced from the usual 330 m/s down to approximately 310 to
220 m/s. This can be determined indirectly by a phase measurement.

 

[Rick Sykora]

Just remembered this: https://www.hifizine.com/2012/12/subwoofer-origami/

As a learning experiment, I built one of these using some Dayton RS225s I had around.

Note the author calls this an M-frame and looks like the subwoofer module for the LX521. Here is what I learned:

1. It is important to mount the left speaker before the right.
2. As you might guess, the woofers need to have some decent power handling - RS225s bottomed out rather quickly. They were not usable for anything with real bass content.
3. When oriented vertically (like the LX521), the structure is rather flimsy. Was worried someone might lean on it and break it. May also be why the LX521 has another cabinet around it. I have seen it used on its side, but not sure whether you get a dipole response in that case.

Finally, the last design on this website refers to its cardiod nature. Linkwitz appears to use this design for the LXmini, but is attributed to John Kreskovsky. Not sure what happened with John, but his website is no longer online.

 

[mac]

Linkwitz was a true innovator, as should be apparent by a quick glance at his treasure trove of a website.

He also shared his opinions of various, innovative designs. Unlike a lot of folks who have blatantly copied someone else's work, Linkwitz cites countless AES authors where some of his work overlaps or builds upon (or improves) aspects of his own work. Pretty uncommon in this industry, IMO.

BTW, I could easily support my motorcycle atop my LX521 W-frame woofers.

 

[briskly]

A rough model for driver directivity.

J1(k*a*sinθ) / (k*a*sinθ)

J1 is the Bessel function of the first kind, k is the wavenumber (2*π*f/c), a is the driver radius.

For ka<1, interference would be required to obtain directivity beyond nil. The cardioid speaker with fixed side vents are a form of "aperiodic" vent, ports stuffed with porous damping material. Normally it was viewed as a patch method for adjusting drivers stuffed into severely undersized enclosures. There is some loss in the low-frequency range of the system as compared to a conventional box.
Anyone who has a theory of aperodic vents for directivity control hasn't said too much on the subject.

So, our first problem is getting a good model for cabinet diffraction.

 

[Peas]

The thing is that if you want to go cardioid up to 500 Hz, you'll need an allpass rather than a simple digital delay.

If you have a larger baffle, 300 Hz will be enough as then there will be the baffle step.

I did it passively, starting from 50 and going up to 350 Hz.

The design is two woofers at the front in a CB with a capacitor and at the rear side there is a third woofer in the same (i.e. doubled) volume. The rear woofer is linked to the normal lowpass with an inductor and a resistor (use 100 W handling). Initially, I also experimented with extra magnets to raise Q of the rear woofer. Thewas linked Spatz and he puts on additional mass to the cone. But using a higher volume and a resistor will do as well.

http://www.hifi-zirkel.de/forum/viewtopic.php?f=15&t=293

Don't mind the peak coming from the tweeter, the measurement is on-axis whereas the design was made off-axis in order not to toe-in the speakers too much. Off-axis, it's rather flat.

 

[GelbeMusik]

I don't believe FIR would be necessary for a cardioid design; just EQ and delay.

This is a very interesting topic. I experimented with cardioid twice. Both were realized mechanically, with slot and foam. Today I would use an electric delay, simply by a combination of low/high pass and recessed woofers. It can be made, according to the simulation, to work down to about 200Hz with a directivity figure of about "6".