Why is the Active Cardioid unpopular in DIY projects?

[voodooless]

Danny Richie

industry leader

My point is that few industry leaders embrace Active Cardioid Systems

Guess what Genelec made with their top-of-the-line speakers and subs:

W371A SAM™ Woofer System - Genelec.com

An absolute revelation in low frequency performance. With the W371A, you can finally wave goodbye to the detrimental influences of your room on LF reproduction.

www.genelec.com

8381A SAM™ Adaptive Point Source Main Monitor - Genelec.com

Our 8381A empowers engineers and creators to perceive every sonic detail, offering true-to-life sound level and dynamics – with the full flexibility of free-standing monitor placement and advanced room adaption.

www.genelec.com

The sub specifically tailors to a cardioid pattern, and the monitors' 4 midranges are basically doing the same thing to narrow directivity forward.

 

[smowry]

Guess what Genelec made with their top-of-the-line speakers and subs:

W371A SAM™ Woofer System - Genelec.com

An absolute revelation in low frequency performance. With the W371A, you can finally wave goodbye to the detrimental influences of your room on LF reproduction.

www.genelec.com

8381A SAM™ Adaptive Point Source Main Monitor - Genelec.com

Our 8381A empowers engineers and creators to perceive every sonic detail, offering true-to-life sound level and dynamics – with the full flexibility of free-standing monitor placement and advanced room adaption.

www.genelec.com

The sub specifically tailors to a cardioid pattern, and the monitors' 4 midranges are basically doing the same thing to narrow directivity forward.

Yes voodooless, the 8381A implements a full range Active Cardioid monitor. It represents an upgrade from the Ones' 8361A + W371A. "Directivity is everything." Ilpo was among the few.

 

[Halbknapp]

I am pretty sure there are not many diy cardioids because of the cost involved, complexity (more drive units needed) and knowledge needed (I know of no documented, easy to replicate kit).

Other then the rather costly Hypex amps there are not many easy to use dsps and amps that can do the needed fir corrections for adjustment of phase. Even the Hypex amps are rather limited if phase has to be adjusted per channel.

There is a commercial 4 channel dsp Amp that mesanovic and others use but it is not available for diy.

 

[voodooless]

do the needed fir corrections for adjustment of phase

You don't necessarily need FIR to do active cardioid speakers. The passive version is also a minimum phase system, so in an active configuration, that should be enough as well. Delay will get you most of the way there. FIR may help you in certain configurations, though, and it may help you get a slightly better result.

 

[Halbknapp]

You don't necessarily need FIR to do active cardioid speakers. The passive version is also a minimum phase system, so in an active configuration, that should be enough as well. Delay will get you most of the way there. FIR may help you in certain configurations, though, and it may help you get a slightly better result.

I have done a full passive cardioid speaker where woofers do the cancelation (instead of resistance enclosure) and it could be done with active electronics just the same but it is much easier to keep the correct phase relation if fir filters are used and you can tune the phase separately. Delay will only work for a certain frequency but above and below the phase relation is frequency response dependent which may just not fit for the application intended

 

[ZolaIII]

Cardioid patern is a no brainer and goes with make less problem to have to deal with later and increased focus over the range they take the patern is just that minimising room impact. I guess it's not easy to achieve and some tunel modeling DSP is still needed even with selected drivers. So for DIY take most popular known driver from such commercially successful speakers as that will be 5.5" one dig more to find at least 6.5" one so that it doesn't uper bass correctly on it's own. Form a good digital model (crucial part which will take most time) and use all you can to bring it up to driver itself. Optimise, optimise and simplify, burp it all together in convolution FIR (pay attention to minimise latency!) and pack it up to DSP. Sub's can remain close enclosure and ideally 10" (for the time domain and same rule). You can try to focus on DIY forums who did it and how far he got. Rule is to use every advantage you can get in order to get where you need to be and there is no shortcuts and I am getting increasingly annoyed with people who don't want to make effort regarding positioning or wanting to do minimal accustic treatment. Cardioid patern will make problem smaller but it won't magically disappear. However using two out of three is better than nothing at all so cardioid + FIR is cure for illness to those who by all means refuse accustic treatment with a slap and forced to use FIR and to those who still refuse half deacent positioning ther is no hope. For those who anyway can achieve good focus in highs and nice mids (the nice rooms more mythological category in most parts of the world) there isn't much need for such speakers.

 

[P.L.]

I have done a full passive cardioid speaker where woofers do the cancelation (instead of resistance enclosure) and it could be done with active electronics just the same but it is much easier to keep the correct phase relation if fir filters are used and you can tune the phase separately. Delay will only work for a certain frequency but above and below the phase relation is frequency response dependent which may just not fit for the application intended

FIR filters are not necessarily required for phase adjustment either.
A sufficient amount of independent phase correction can be achieved using only all-pass filters.
The example below was also optimized using only IIR filters and constant time delay. (a DIY 4-way active cardioid design like Kii Three)

I think what truly matters in active cardioid implementation is how the optimization is practically performed, and I will soon cover this topic I used in detail in a separate thread.

 

[ZolaIII]

@P.L. Stop looking at convolution traditionally, adopt it as kernel with sum of all alterations passed in one and for convenience.

 

[voodooless]

@P.L. Stop looking at convolution traditionally, adopt it as kernel with sum of all alterations passed in one and for convenience.

You can do that if you have a vast number of taps at your disposal. If not, an IIR/FIR mix is generally still the best way to go. Audyssey and Dirac do this hybrid approach as well.

 

[nc535]

If you use Camilla DSP on Rpi4, you will have a vast number of taps at your disposal. My simulation told me 4K taps at 96 khz per driver was enough if I added PEQs in the low end where there were errors for not having more taps.

 

[voodooless]

If you use Camilla DSP on Rpi4, you will have a vast number of taps at your disposal. My simulation told me 4K taps at 96 khz per driver was enough if I added PEQs in the low end where there were errors for not having more taps.

That will give you only 24 Hz resolution, so indeed, for bass duty that is generally not accurate enough. A Pi4 will do much more than 4k taps, though.

A nice trick is to first downsample to a lower sample rate, so for bass, you can get away with maybe 8 kHz or less and then still use the 4k taps, giving you 2 Hz resolution, then upsample again to the original rate.

 

[nc535]

that is why the error correcting PEQs

I know of that trick. Don't know if I can do that in Camilla. Perhaps if MiniDSP would implement in Flex, I wouldn't need Camilla

 

[voodooless]

that is why the error correcting PEQs

Exactly

I know of that trick. Don't know if I can do that in Camilla. Perhaps if MiniDSP would implement in Flex, I wouldn't need Camilla

I don't think Camilla can do that. It could probably be implemented as a special convolution kernel that already has the sample rate conversion built in (2x, 4x, 8x, 16x reduction for instance). Since this can be synchronous, it should be relatively simple. But given the number of taps even a Pi can do, it's probably not super useful.

MiniDSP doesn't support it for sure. You can however do it with similar DSP hardware. SigmaStudio does offer some tools to do this.

 

[ZolaIII]

You can do that if you have a vast number of taps at your disposal. If not, an IIR/FIR mix is generally still the best way to go. Audyssey and Dirac do this hybrid approach as well.

That would be optimising it in a one way. You wage latency never the less as simpler it gets to taps number to get resolution and upsampling helps. Played with it for a while way back. Thinking about doing it again with James DSP.

GitHub - Audio4Linux/JDSP4Linux: An audio effect processor for PipeWire and PulseAudio clients

An audio effect processor for PipeWire and PulseAudio clients - Audio4Linux/JDSP4Linux

github.com

Now day's computal power is getting cheap on general purpose CPU cores. I hope more people will jump to this and futured discussions especially regarding the model so that we can make most of it.

 

[Ktacos]

General lack of affordable dsp amps is why I haven't done anything like cardoid that can actually sit inside a home. Still waiting for the market to catch up there. Being stuck with off the shelf drivers for the most part also puts a bit of a damper on projects when one sees what can be done with custom stuff, like waveguides actually designed to mate with the tweeter rather than diy options that are just ok.