EP21 - Single Stereo Speakers (SSS)

My 21st loudspeakers are implementation of Elias Pekonen's Single Stereo Speakers (SSS). This unique concept utilise single, center-positioned main with front and side firing drivers and actively utilise room reflections as important aspect of sound reproduction.

The drivers radiate as follows:

As you can see this is a unique approach to Stereo reproduction.

You can read the full concept at Elias' website.

The build progress are posted at DIYAudio

Topology: Active 2-way
Radiation pattern: Single speaker, front and side-firing
Directivity: Monopole side-front waveplane. Dipole or Cardioid bass 200hz and below
Frequency response: 20hz - 18khz.
Sensititivity: ~83dB
Dimension (H, W, D): TBD
Amplifiers: 4-Channels required (2x stereo amps)
Crossover: One 2x4 MiniDSP

Update 2/11/12 - Project brief posted at DIYAudio
Update 8/11/12 - First design posted
Update 13/11/12 - Drivers arrived, 3x HiVi B3N
Update 23/1/13 - No progress yet ....!
Update 5/7/13 - No progress ... I better pull my finger out :)

Next: Build the cabinet

Upside down / down-firing woofer

Vertical driver orientation like inverted W-Frame used in S19 requires specific requirements as follows:

Old archive from Adire Audio:

Often times, you'd like to mount a driver facing down, such as in a Sonotube™ enclosure.  However, you may not be sure whether the driver you have is suitable for such an arrangement.  Will the driver hold up to horizontal mounting?  Will the relentless pull of gravity be too much, and cause intolerable amounts of sag?Well, there's actually a way to calculate the sag a driver will exhibit when mounted in a downfiring (or upfiring) position!  And you don't need anything more than the Fs, Vas, effective surface area (Sd), and the Xmax of the driver.
To calculate the sag, we first need to calculate the acoustic compliance of the driver.    That is, how soft is the suspension?  This parameter is called Cms, and is related to Vas.  Cms is the acoustic compliance of the driver, and has the units of meters per Newton.  It is calculated as:
Cms = Vas / (1180 * c^2 * (Sd/10000)^2)
bulletVas is the equivalent compliance of the driver, in liters
bulletc is the speed of sound, in m/s (use 343 m/s as a good approximation)
bulletSd is the effective surface area of the driver, in square cm
Note that you can use the following table for rough Sd estimates for drivers:
8"230 cm2
10"330 cm2
12"480 cm2
15"780 cm2
18"1150 cm2
Now that we have the Cms of the driver, we'll need to calculate the effective mass of the driver, Mmd.  For this, we'll need Cms and Fs.
Mms = 1 / ((2*pi*Fs)^2 * Cms)
bulletpi = 3.1415927...
bulletFs is the resonant frequency, in Hz
bulletCms is as calculated above
So, with the Mms and Cms, we're almost there.  To calculate the actual sag, you'll need to multiply the stiffness of the suspension times the mass of the diphragm times the pull of gravity:
Sag = Cms * Mms * g
bulletCms is as calculated above
bulletMms is as calculated above
bulletg is the acceleration of gravity (9.81 m/s2)
This will give us the sag in meters.  So, multiply by 1000 to get to millimeters.  Here's an example, using our Shiva subwoofer:
Vas = 136.6 liters
Fs = 21.6 Hz
Sd = 481 cm^2
So, we calculate the Cms as:
Cms = Vas / (1180 * c^2 * (Sd/10000)^2)
Cms = 136.6 / (1180 * 343^2 * (481/10000)^2)
Cms = 0.0004253
So, now we calculate the Mms:
Mms = 1 / ((2*pi*Fs)^2 * Cms)
Mms = 1 / (2 *  3.1415927 * 21.6)^2 * 0.0004253)
Mms = 0.12766 kilograms
or 127.66 grams.  Lastly, we need to calculate the sag of the driver:
Sag = Cms * Mms * g
Sag = 0.0004253 * 0.12766 * 9.81
Sag = 0.0005326 meters
Sag = 0.5326mm. 
So, that's all fine and dandy.  We can calculate sag.  But what does it mean, and how can it tell us if the driver is OK for horizontal mounting?  Well, as a general rule-of-thumb, we use the following:
If the sag is more than 5% If the sag is more than 5% of the Xmax of the driver, then it's not meant for horizontal mounting.
Simply put, if you lose more than 1/20th of your Xmax from sag, then it shouldn't be mounted that way.  To finish off the example, let's look at Shiva:
Xmax = 15.9mm one way
Sag = 0.5326mm
Percent Sag = (Sag / Xmax) * 100
Percent Sag = (0.5326 / 1
5.9) * 100
Percent Sag = 3.

So, Shiva would be OK for horizontal mounting.  And now you can calculate the suitability of your favorite driver, too!  Just run the numbers, and if you're less than 5% of Xmax, you're sitting pretty!

Work in Progress

Random notes of work in progress

18/12/12 - Observations and measurement of Nao Note

NaO (and Orions) commercial kit offering designed by experts in the field. It is a reference design. Particulary with Nao Note variant where it is desired to maintain true dipole polar response to high frequencies.

DIYAudio link: http://www.diyaudio.com/forums/multi-way/163072-nao-note-preview-16.html

9/11/12 - S22 Improving high frequency directivity

9/10/12 - Cardioid using W-Frame
I proposed that sealing one of the slots in W-Frame dipoles could yield Cardioid response.

14/9/12 - Z-Frame Dipole Subwoofer
Small dipole loudspeakers are competing requirements due to the need for pushing air with large woofers.
Here is an idea to squeeze the footprint even more by using opposing force. 

After three coats of Tung Oil

Routing and fitting baffle to 'furniturise' S19.
Perfect fit for both Seas DXT and Vifa XT25 tweeters. Uncoated baffles.

Mahogany (Eastern) test finish. 1 coat of Tung Oil

S20-Z Compact Open Baffle using Zaph Kit

This design was announced as winner of Zaph Design Contest. The design had gone through some changes but you can read the original submission.

Follow the build discussions at HTGuide Forum.

Compact and economical Open Baffle design, similar dimensions to a typical 5" hifi floorstander!

While open baffle loudspeakers provide the outmost transparency and openness due to absence of box, they are normally very large and expensive to build. Zaph had done the hard work of providing an excellent 2-way loudspeaker kit, and we simply take this kit as a basis for this design.

The loudspeakers need additional woofers to reach 20hz, and any 10” with decent xmax will do. Extremely large subwoofers, in fact, are not advisable due to size constraints. I use Jaycar 10” woofers bought for about $15 each.

Design : 3-way Open Baffle using Hybrid crossovers
Frequency Response : 20Hz – 20kHz
Polar response : Dipole 20hz to 1.8khz, then forward directional
Crossovers : 1.8khz M-T, 220hz W-M (Asymmetric). With dipole correction
Placement : 80cm minimum from front walls, 50cm from sides
Size (H x W x D): 97cm x  23cm x 27cm
Tweeters: Vifa DQ25SC16-04
Midbass: Zaph|Audio ZA14W08
Woofers: 4x 10"

The loudspeakers require 4 channels of amplification and use hybrid design to achive this:

Midrange-Tweeter section uses ZA5.2TM kit’s crossover, passing the signal for dipole eq first. The Dipole EQ is quite simple: a notch filter and lowpass-shelving filter

Bill of Material
Zaph ZA5.2TM Kit (Madisound), pair $200
10” woofers, 4x $80 - $100
XO / dipole eq options:
  - MiniDSP $100, or  
  - DIY op-amp based $30-$40  

MiniDSP Crossover Settings

Woofer Section:

Image - 220Hz LR4 Crossover
Image - 330Hz/-16dB/2.8Q quarterwave resonance notch filter (measured)
Image - 20Hz/0.7Q Linwitz Transform biquad correction (measured)
Image - 6db/oct Dipole rolloff EQ

The above Linkwitz Transform is for Jaycar CW2139. It will be different for other woofers, depending on the original Fs and Qts of the driver.

A very good value Woofer is Dayton Audio SD270A-88. Low Fs, good xmax, good price and reliably sourced from PartsExpress. If that is the chosen woofer, the Linkwitz Transform Biquad will be as follows:

Using Dayton SD270A-88 Sub
(Credit to SVEN from DIYAudio for providing the measurements):
Image - Linkwitz Transform for Dayton SD270A-88.
Image - Dayton SD270A-88 subwoofer FR modelling

Note: If you have other woofers in mind, please provide me with .frd nearfield measurement on the cabinet mouth and I will gladly publish the LT biquad parameters.

Midbass & Tweeter Section

Image - 120Hz LR4 Asymmetrical crossover
Image - 650Hz/-4.2dB/1.4Q Notch for dipole peak EQ (measured)
Image - 6db/oct Shelving-Lowpass dipole rolloff EQ
Image - Overall response of the panel. Note the dip at 3.8khz which needs to be investigated

Most critical of these settings is the 650Hz dipole notch. It is directly related to the front-rear separation and hence the baffle width. The final loudspeaker build had a baffle width of exactly 21.5cm. If your baffle is wider, the notch will be moved to lower frequency. The notch frequency can only be reliably measured outdoor/anaechoic conditions.

The crossover is 120hz asymmetrical to compensate floor reflections. This puts large demand on the 5" midbass. I play loud enough to fill in a 4.5m x 8m living room with no issues. But for extreme SPL, you may want to change it to 220Hz.

A null is observed at 3.8khz which deserve further investigation.

Other Settings

Image - 2π to 4π radiation space transition
Image - delay and attenuation

As with any other loudspeaker design, the top half of S20Z fires to full space (4π), while the bottom half fires to half-space (2π). This needs to be compensated accordingly using shelving highpass filter.

The combination of CW2139 and Zaph kit does not need further attenuation. With SD270A-88 above it might need minor tuning.

10/7/12 - Drivers shipped and on their way to down under !
17/7/12 - Drivers arrived.
29/7/12 - First dipole sub built and measured.
12/8/12 - Woodworking completed. The final design is actually smaller than planned.

15/8/12 - Preliminary crossover system and listening impressions observed.
14/9/12 - Outdoor measurements taken.
19/10/12 - Crossover configuration and measurements posted
9/7/13 - Other narrow dipole sub possibilities: Z Frame, Z Frame2, W-Frame, M-Frame
10/7/13 - wmcarpenter's adaption using Zaph MTM kit and 4x Bipole woofers
12/7/13 - danvprod's build progress using Markaudio fullrange driver

Audio-Technica ATH-AD700 Correction Circuit

These are mid-priced (~$150) "Open-Air" headphones which have strong following among head-fi enthusiasts. Among the general consensus are high-resolution, airy and transparent sound. I heard them first when JP Howard  brought them to my house a few years ago and was immediately impressed.

I guess being open-aired they do not have to deal with back pressure from the typical cups, and for me the appeal is dipole-like neutrality. There is less fatique because the ears would not be pressurised by the bass too much. The disadvantage is, of course, other people will hear the songs you are playing!

Upon hearing the headphones after their arrival, I noticed they sound too bright for my liking. Hence I measured them as per above photo.

From the impulse response we can see that the headphones exhibit strong resonances at 5khz and 10khz with a rising trend to the frequency response. This explains the bright sound of the headphones.

Raw nearfield measurement of ATH-AD700

Although looking non-flat, these are actually typical response for headphones. They are basically 2" fullrange transducers and would experience breakup at some point. Compare with Stereophile's measurements of AKG K530, Audio-Technica ATH-AD700, Beyerdynamic DT440, and Grado SR80 here.

The response can be improved by utilising passive EQ consisting of 1.56mH inductor. Effectively a 1st order lowpass filter with corner frequency of ~5khz.

Schematic diagram of ATH-AD700 Passive EQ.
The parts can be obtained from Jaycar in Australia, or any other electronics shops.
note: you can also use single 1.5mH inductor from SoundLabsGroup.

Finished Passive EQ

Measured acoustic output of ATH-AD700 when equalised with Passive EQ.
There is very little noticeable loss as higher gauge inductors (0.8mm/20AWG) are used.

Perceptually the headphones now sounds very natural. 

Of course as with all headphones they still exhibit high spatial distortion. That is ... the sound seems to emanate from inside the head. So they are not a true replacement for proper loudspeakers.

Discussions on this EQ are in HeadFi and SNA.

Other measurements of interests:

2nd, 3rd, and 4th Harmonic distortions

Burst response


Below DSP eq was Archived on 29/5/12.
Passive EQ above is preferred due to its simplicity.

These can be improved by applying two notch filters. I do not have separate headphone amplifier on-hand so the equalization process utilised external pink noise. Not ideal, but close. I would revisit this later on.

EQ1: fo = 47780hz, Q = 1.86, Gain = -2.2dB
EQ2: fo = 11,669hz, Q = 1.10, Gain = -8.2dB

Pink noise measurement of ATH-AD700
Blue = Before EQ, Black = After EQ

I use Foobar2000 VST Plugin and TDAe EQ2004P parametric equalizer to prototype the notches.


Application of DSS filter for single-tweeter Dipoles
The imbalances with single firing tweeter is due to different radiation between front and back. And normally the tweeter levels are adjusted by ear to correct it to a certain degree (e.g. Linkwitzlab Phoenix). But what if instead changing tweeter levels we use the DSS filter ?

Class D for Tweeters?

Class D amplifiers have been known to be superior option for Bass and Midrange duties due to their efficiencies and damping factor. But what about for Tweeters?

I measured ClassD and Gainclone amplfiers at the *acoustic output* (loudspeakers), instead of amplifiers.  I conclude that Class D is at least as good as Gainclones/Class AB. The distortions of both types of amplifiers are insignificant compared to loudspeaker/transducer distortions.  Please download the .pdf here, or follow my discussions in SNA here and here. (7/4/2012)

Compression Driver showdown: B&C DE250 vs. Selenium D220Ti

Arguably the two most popular drivers for controlled-directivity DIYers. D220Ti is the darling of Econowave, and DE250 is used extensively in Gedlee Speakers. But which one is better? 
Read .pdf here and summary/discussions here(26/5/2011)

Effects of Crossover Points on Loudspeaker Directivity

Crossover points needs to be chosen accurately for seamless integration of driver directivity. Here are examples of Econowave loudspeakers crossed at 2.5khz, 1.5khz, and 1khz. Read here.

Constant-directivity vs. Dipole Loudspeakers
Which is the better transducer ?

S19 4-Way Dipole Radiator

S19 dipole-radiating speakers

The quality of dipole loudspeaker accuracy is improved by using a small 5" midrange driver from 400hz to 2khz. The tradeoffs are higher complexity and more difficult design process to take account of polar response changes.

Push-pull dipole subwoofer section

The dipole subwoofers are w-frame to benefit from vibration cancellation. They are long-throw Dayton RSS256HF-8. The woofers are Seas Excel W22EX001 and W15CH001, and the Tweeters are Seas DXT.

The w-frame sub's cancellation topology really works. It's quite mesmerising to see the drivers pumping air furiously yet there is no cabinet vibrations. Seen below is how clean they are crossed at 120hz. The subs are -3db at 20hz.

Please follow the build progress at HTGuide Forum.

Topology: Active 4-way
Radiation pattern: Dipole
Directivity: Dipole from 20hz - 2khz then forward-directional
Frequency response: 20hz - 20khz.
Sensititivity: ~88dB
Dimension (H, W, D): 117 x 31 x 31 (cm)
Amplifiers: 8-Channels required (4x stereo amps)
Crossover: Two 2x4 MiniDSP

Update 12/3/12
Outdoor measurements of early prototype to investigate the frequency response of small midrange.

Update 25/3/12 - Seas Excel W15CH001 received and tested
Update 2/4/12 - First prototype of S19M (magnesium)
Update 20/4/12 - S19/MT - Titanium dome Tweeters (Vifa DQ25) used for improved resolution.
Update 31/8/12 - Started the final design of the cabinet and wood choices
Update 2/9/12 - Cabinets completed
Update 23/1/13 -  I have been living for quite a while with these speakers. I havent' had any desires to build something new !
Update 28/6/13 - Frequency response in the critical 1khz-20khz region showing ±1db (XT25)

Update 13/9/13 - The 4x Dayton Reference subs requires large power to drive them to xmax.

Update 21/7/15 - Amateur video of S19