Cardomon SB - DIY passive Cardioid with extremely linear frequency and Power Response

[Halbknapp]

Dear ASR Community,

i want to show you the project i am currently working on - a passive Cardioid Bookshelf Speaker with Waveduide Tweeter (Seas Noferro 900/H1025-06 27TF and 3x SB Acoustics SB17NBAC35-8 (2 of which mostly only cancel side and rearwards Sound). The reason i tried this is that most speakers just dont sound right in my living room, particularily they sound dull and muddy even when the on axis frequency response is pretty linear (i am looking at you Neumann KH150 ;-)). This Speaker with cardioid pattern should resolve the muddyness issue, at least so i hoped. I got a tip on how to achive passive Cardioid with using a second woofer in the back from a german forum user called Spatz some years ago (in german hifi forums).

As the on axis and power Response looked pretty promising and Sound was good i have gone full cracy with the passive Crossover to flatten out the responses just to show it can be done. It helps that the Cardioid Design throws up less peaks and dips in room even without windowing the response (at least much more then with conventional Speakers). Still this is like the 100th version of Crossover that is printed here and the parts are ~200$/€ per Speaker - not usind anything special or more expensive then needed or reasonable (no fancy coils or capacitors).

In the picture is a preliminary Crossover as the final one still needs some parts to arrive. The preliminary one has a more wavy frequency response and is corrected with eq (Wiim Pro) to mimic the response given with the final crossover. It is in my crossover test boxes and not tidy at all. The final crossover will look better but still be huge and not fit into the speaker.

How does it sound? Very neutral and uncolored, light and transparent, lively and fresh without being tiring. My wife and daughter wo dont know much about speakers but love Music preferred this speaker against Neumann KH150, Genelec 8341 and a big active 3-Way i made myself as well as another good sounding 2-Way Speaker (Heissmann Acoustics DXT-Mon 182. To be fair i used all Speakers "as is" and without room correction - on the actives i tried the dip switches to maybe improve sound. If set up perfectly and EQd to the room other speakers may still sound better. Not doing this the Cardomon SB has the edge.

However not all is great:
The sensitivity is low (~80db/2,84V but i have to check closer with final crossover).
The absolute Bass output that can be achived is low (and the limiting factor if the speaker is not used with a sub). Harmonic Distortion of the SB17 is superbly low in the Midrange but not so much in the bass - in addition you would get a lot of IMD if you push it hard.
Horizontal dispersion is great but vertical is not. This can not be seen in power Response as there is lobing that fills the gap. Not sure why though as i would expect a bigger Dip in power Response then can be seen in the graph.
Tweeter/Waveguide Response is not as smooth as could be (SB Acoustics Beryllium Waveguide Tweeter, Purify Tweeter, ASCILAB are all better).
Cost - With 4 Drivers per side and a huge Crossover and the realistic need for a sub the speaker is no bargain. Its worth for the sound quality though.
The Design is inconsequential - If a Subwoofer and DSP has to be used anyway for best Sound anyway one could save a lot on Crossover parts and not correct Peaks and Dips in the Crossover but in DSP instead.

I may have fallen into a trap and corrected a measurement error in my crossover (with my setup for measurement all Speakers have a 200Khz peak and 300Khz Dip). Luckily one Resistor fixes that and can easily be changed. As i however tend to listen in spaces where this exact peak and dip applies i propably let the fix in place - just be aware that this may be room correction and not Speaker correction.

For all of these Reasons mentioned i consider this more of a proof of concept and beyond a reasonable DIY Project for anyone else.
I also have to mention that a Dutch &Dutch or Kii 3/7 may just plainly be a better Speaker - propably the same is true for less expensive Cardioid ones like the Mesanovic CDM65 or Buchardt A500.

 

[Fledermaus]

Utterly unreasonable as far as cost and efficiency are concerned, but a superb exercise in style as for linearity and directivity control, hats off for that - especially the impressive implementation of passive-crossed, cancelling side-woofers !

 

[burkm]

@Halbknapp
What are the estimated over all costs in Your terms (EUR) for one speaker including homemade cabinet ?

PS.: I just wondered about the very low sensitivity of the speaker as mentioned: ~80db/2,84V. That "thing" then needs lots of amp power, doesn't it ?

 

[Halbknapp]

@Halbknapp
What are the estimated over all costs in Your terms (EUR) for one speaker including homemade cabinet ?

PS.: I just wondered about the very low sensitivity of the speaker as mentioned: ~80db/2,84V. That "thing" then needs lots of amp power, doesn't it ?

The cost for me was ~450€ per speaker but I had a good deal on the SB Midwoofers. I expect cost for general sourcing to be 600€/$. Optimizing the crossover would get you to 400 for a piece of guess.

Technically speaking the sensitivity is ok in terms of power need but you should have an Amp that could shell out a lot of power on 8 Ohm as a lot of voltage is needed. I currently listen with a Topping PA5II and the actual limiting factor is woofer excursion. With a Subwoofer and active crossover you may run into voltage/gain limit with amplifiers (if not using something like 2x Aiyima A70 mono that does 150W on 8 Ohm which I happen to have).

 

[Halbknapp]

Utterly unreasonable as far as cost and efficiency are concerned, but a superb exercise in style as for linearity and directivity control, hats off for that - especially the impressive implementation of passive-crossed, cancelling side-woofers !

Thanks, as for reason I would actually recommend any other builder to use 4 Ohm woofers and offload as much as possible of eq to a dsp. That way sensitivity could be improved quite some and cost goes down in the crossover.

 

[Wolf]

Bravo!

 

[nc535]

Your PIR and power response curves are admirable so I would expect this speaker to sound good, but I hesitate to call the directivity cardioid or cardioid-like as the polar map is showing that the lower midrange beamwidth is nearly 180 degrees. Look at the Sigberg cardioid development threads or review of the D&D 8C and you will see -6 db beam widths <=90 degrees in this range (from 100-200 Hz up to at least 1 kHz) with a DI of >= 5.

Perhaps it's the simulation at fault - looking at your XO schematic, I would expect a non-zero "Z" parameter on your side woofers representing how far back from the baffle they are placed. Also, you might not have enough amplitude coming from the side drivers to match and cancel with that coming back from the front woofer. Its difficult to tell from the screenshot graphs which FR trace represents the side drivers.

Active cardioid without DSP is really hard as you can't easily trim the level and delay of the side drivers. which is key to optimizing the directivity. If you've already got the amps, then all DSP costs is a Raspberry PI running Camilla DSP plus an 8 channel DAC or sound card.

 

[Halbknapp]

I applaude your observation and you are right. I have a DI of 3 very low going to 5 in the upper midrange. Normal speakers would transition from 0 to 5 instead, cardioid keep the 5 DI. I could have easily modify the speaker to have a DI of 5 or even a bit more from 50 to 250hz but I could not get it to cancel more then 3.x in the area of 300 to 500hz. Improve down low part of the frequency spectrum would make the higher part worse. Therefore I decided to more opt for a smooth DI and power response at the expense of some rear wave canceling. Another reason really was that a higher DI would make the lower frequencies sound too thin without a use of a Subwoofer as the power response would not be tilted down enough.

The phase and timing is with the method used here already perfect in the lower register for cancelation but phase wrap will prevent the 300-500Hz region from working perfect so only a FIR filter can help by setting amplitude and phase separate for best cancelation.

 

[Halbknapp]

In addition the simulation is correct as the side woofers were measured in relation to the front 0 degree point and the offset is in the measurement already as delay (as well as that both woofer responses are measured as one).

As it's not too complex I also added or subtracted delay from the side woofers or applied peq in the simulation but that would not make it better in general, only improve one factor and make another worse

 

[Halbknapp]

@nc535 i Was wrong in feeding the difference between the speaker into the simulation driver Y axis offset as the difference is already in the measurement.
That means the drivers are simulated with double the distance between them and vertical dispersion is displayed wrong. Corrected the PIR again has the expected dip in the crossover region and vertical dispersion is a bit wider then displayed here (again as expected by LR24 Crossover)

 

[nc535]

I'm not sure what you are saying re' the difference is already in the measurement, so I shouldn't comment.

Speaking for myself, I would measure each driver separately, on and around its own axis, so I could draw the Vituix schematic with the driver symbols configured according to their actual positions on the baffle and cabinet walls. That way, there wouldn't be a Yoffset included in the measurement, which at least to me, would be less confusing.

 

[Fledermaus]

Look at the Sigberg cardioid development threads or review of the D&D 8C and you will see -6 db beam widths <=90 degrees in this range (from 100-200 Hz up to at least 1 kHz) with a DI of >= 5.

Here's the 8c horizontal polar according to Erin :

(See here )

I wouldn't say the 200-1k range -6dB beam width is <= 90° - - rather 150°, give or take - close to mine, mind you, FWIW

BTW I have a hard time deciphering @Halbknapp's polar map SPL scaling and contour lines - for me, I got used to a 30-40 dB scale with the -6dB line falling in the yellow color strip, easier to read and to compare.

 

[Ktacos]

Eh, I gotta be honest I don't really see much in the way of cardoid pattern control for the woofer in this diy speaker.

I wouldn't say the 200-1k range -6dB beam width is <= 90° - - rather 150°

Are we looking at the same graph? 8c definitely drops off hard in output beyond 90 degrees, it appears to be nominally 90 degrees for the woofer region.

 

[nc535]

I miss spoke. I said 90 degrees beamwidth but the picture in my head when I said it corresponds to 90 degrees half beamwidth - i.e. the -6db point would no more than 90 degrees off axis, which is what we see at the low end of the D&D's cardioid range.

 

[Fledermaus]

As you see fit, but if you open the link pasted above and hover your cursor over the -6dB line (separating coral red from orangeish colors, see scale) between 200 and 1000 Hz, you'll see the displayed digit oscillating between 70 and 80° (resp. -70 and -80 on the opposite side), which lands you in 150° coverage territory, more or less...

 

[Halbknapp]

Eh, I gotta be honest I don't really see much in the way of cardoid pattern control for the woofer in this diy speaker.



Are we looking at the same graph? 8c definitely drops off hard in output beyond 90 degrees, it appears to be nominally 90 degrees for the woofer region.

Here are is the Picture of horizontal Dispersion normalized with and without the additional rear TMTs. Please note that this Speaker is (a bit) Cardioid down below 100Hz where you would Cross to a Subwoofer anyway and other Designs do exactly that. D&D Cardioid therefore ends a bit above 100Hz.

 

[Halbknapp]

Here is the Polar Chart at 100Hz where the radiation is the widest ("worst Case"). Rear Wave is 10db Down Vs 7db on non Cardioid. But you are totally right that this is not real Cardioid as it does not have the typical Cardioid Form that would show in a much deeper rear 0 and look like a stilistic heart shape.

 

[Halbknapp]

For Comparison - here a stronger Cardioid I could get by just changing the Resistor and Inductance in front of the rear TMTs. HOWEVER the rear 300 Hz radiation is a lot stronger (seen as red Spot at 180°). I consider this worse but it may be up to Test.

 

[Halbknapp]

As mentioned there was an error in the Simulation and the real Power Response and DI looks more lie this (Crossover has been slightly changed too - will update original Post later). The typical dip of PIR and SP from vertical Dispersion in Crossover Region can be observed.

 

[Halbknapp]

BTW I have a hard time deciphering @Halbknapp's polar map SPL scaling and contour lines - for me, I got used to a 30-40 dB scale with the -6dB line falling in the yellow color strip, easier to read and to compare.

As a fun fact the hard to read Scale and Color Palette is the CTA-2032-A Scaling of 50dB and the Klippel Color Palette. I too prefer 40db Scale and default Color scheme of Vituixcad as both are much more revealing but its not ASR Standard.