Original plan was to use conventional box woofer with short front horn. Prototype was done and tested, but it was not completely satisfactory in our living room. Bass range was more uneven and less integrated than with cardioid woofer. Mid-bass had dip which separated lower bass from upper frequencies. This phenomenon is common with conventional box speakers; reflections from front wall and corners sucks mid-bass where front part of the room is dominant sound reflector and speakers are located 80...130 m from front wall. Location of vents far from the cone and close to front wall increased non-integrity I think. Prototype had lots of authority and capacity as vented, but that fun was lost when vents were plugged. Correspondingly, closed box was very fast; excess group delay close to nothing and rising bass balance runs like a rabbit.
Front horn increased requirement of equalization which was not a surprise. Horn was slightly too deep dropping frequency response already below low-pass (X/O) frequency. Sound was kicking and bass notes were felt on the skin with low pressure.
I was also planning a passive crossover for the woofer, but prototype was not compatible with it. Lower midrange produced way too much power to burn in resistor. This would be a good choice somewhere else and crossed 200...300 Hz.


So, I gave up quite fast and picked woofer enclosure from KS-2125. That box with two 18" drivers is too large for 400 Hz crossover, but careful equalization and tilting slightly back - if necessary - should make the trick. I was also worried about intermodulation. Resistance enclosure uses more excursion than vented/closed which may cause fuzzy lower midrange when cone operates close to xmax.

Horn-loaded midrange has been a long-term dream. I had to do something decent before my dream is totally faded. I have started to suffer forwardness of high frequencies which is possible consequence if tweeter is not (enough) omni-directional point source; dome, short ribbon or planar. Horns and wave guides are not causing troubles with all microphone techiques, but all sound engineers are not aware of this or don't care, I guess.
StereoLab SL250-2 was selected as main horn. Slightly bigger horn would have been better for 300-400 Hz crossover but SL190 was not yet available when I purchased the horns. BMS 4590 coaxial driver was selected to enable testing without separate 3rd way. BMS 4550 and SL600-1 are optional supplements for 3rd front way or rear tweeter. Design is initially coaxial front + rear tweeter, but may change if combination is not satisfactory.

Afterwards I was informed by StereoLab that Klang&Ton has published two designs with SL250+4590 (4/2010 "Big Bang", 2/2011 "Tron"). I have looked the pictures, measurements and filter schematics. Crossover will be different for sure because I'm not able to imitate anything, regardless of quality.

Technical specifications

General 3-way, resistance enclosure WW,
horn-loaded MTT
Dimensions WW: 480 x 940 x 528 mm (W x H x D)
MTT: 445 x 590 x 520 mm
Volume WW: 90+90 liters
X/O frequencies 410 Hz and 6500 Hz
acoustically 24 dB/oct and ~30 dB/oct
Drivers BMS 18N840v2
BMS 4590 + SL250-2, front
BMS 4550-16 + SL600-1, rear
Impedance 4 / 10 ohm (WW / MTT)
Sensitivity MTT: 100 dB/2.83V/1m


Woofer enclosure

See KS-2125
There are two changes for KS-1804. Woofer cabinets are filled with polyester, 1.0 kg/box. Bi-amping option has been added to enable one power amplifier per 18" driver. Individual power amps allow also active delay for upper woofer to eliminate distance difference (about 90 mm). This option is not tested.

Horn holder

IMG_0951.JPG IMG_0954.JPG IMG_0956.JPG

Enclosure drawing: KS-1804.pdf


Crossover is semi-active.

Woofer section

It's low-passed and equalized with Behringer DCX2496, modified with DCX active upgrade by Linear Audio and SRC/clock by Frank Oettle.

Parameter file: KS1804TA.xpc (13.12.2011)
Download Behringer DCX-Remote16a.exe to investigate this file.
Gain of power amp is +27.6 dB (Apoll Acoustics MA-1.0).

Upper BMS 18N840 has all-pass filter to produce about 86 mm time delay, compensating different distances from woofers to listening point (upper woofer is closer). Phase alignment improves integration around lower X/O frequency.


Horn section

Horns are high-passed with Behringer DCX2496 to decrease power and voltage requirements of power amplifier. It has also passive filter including basic equalization for mid-range, sensitivity dropping from 115 to 100 dB/2.83V/1m and high-pass for rear tweeter. Sensitivity of 4590+SL250 is too high for many amplifiers with unbalanced interconnections and high gain. Gain requirement of power amp is only 6...10 dB  as full active, which is difficult to find and not so rational to compose with line level attenuator.

I'm using modified DCX as DAC and pre amp to have common volume control for woofers and horns.
Gain of power amp is  +19.0 dB (Accuphase A-20V, Gain switch = -9 dB).

Filter schematic of horn section, version 1:

Capacitors: MKP, 250 V or more

Coils (by Graditech):
L1021, L1041; air core, wire 1.00 mm
L2021, L3021; air core, wire 0.71 mm

R1021, R2011, R3011: 10W
R1041, R1051, R2031; 5W

Crossover is located below BMS 4590 on the back side. Vertical plate is 184 x 116 mm, and horizontal 182 x 127 mm.


Frequency response of horn section (without rear tweeter).

Frequency response of rear tweeter. Phase is close to front tweeter i.e. treble is initially bipolar. Could also be dipolar because distance between front and back tweeters is much longer than wave length.

Phase response of horn section. Red = front tweeter, green = rear tweeter

Impedance response of horn section.

Filter gain of midrange; BMS 4590/M + SL250
- includes 1st order active high-pass @ 488 Hz

Filter gain of front tweeter; BMS 4590/T + SL250

Filter gain of rear tweeter; BMS 4550-16 + SL600


Frequency response

On-axis response
- measuring distance 220 cm, on horn axis
- time window 3.4 ms, no smoothing

Room response
- measuring distance 250 cm, 5º off-axis
- red: time window 300 ms, smoothing 1/6 oct.
- cyan: time window 3.33 ms, smoothing 1/24 oct.

Frequency response with floor (and roof) reflections
- measuring distance 250 cm, 5º off-axis
- time window 5 ms
- no smoothing (DFT)

Individual frequency responses (cyan = rear tweeter)
- measuring distance 100 cm, 0º
- time window 300, 10, 5, 5 ms
- smoothing 1/3, 1/6, 1/24 , 1/24 oct.

Off-axis responses of horn section 0...90º, steps 5º
- measuring distance 150 cm
- time window 5 ms
- smoothing 1/12 oct.
- 0º response normalized to 0 dB

Power average of horn section ±180º red, on-axis green
- measuring distance 145 cm
- time window 6 ms
- smoothing 1/12 oct.

Note: mic not calibrated

Same without rear tweeter: KS-1804-MTT_FR_145cm_0+360deg-avg_6ms_1p12_wo-RT.png
Rear tweeter doubles the power at 20k as expected.

BMS 4590/M + SL250
- measuring distance 150 cm, on-axis
- time window 10 ms
- smoothing 1/24 oct.

BMS 4590/T + SL250
- measuring distance 150 cm, on-axis
- time window 5 ms
- smoothing 1/24 oct.

BMS 4550-16 + SL600
- measuring distance 134 cm, on-axis
- time window 10 ms
- smoothing 1/24 oct.

Step response

- measuring distance 250 cm, 5º off-axis

Excess group delay

- room measurement at 250 cm, 5º off-axis

Harmonic distortion

- measuring distance 100 cm, on-axis
- SPL 100 dB, smoothing 1/6 oct.
<400 Hz on woofers' axis, >400 Hz on SL250 axis
- rear tweeter muted

Burst decay

- room measurement at 250 cm, 5º off-axis


Without room acoustics treatment

- room measurement at 250 cm, 5º off-axis
KS-1804_STI_250cm_05deg.png KS_T30.png

After room acoustics treatment with some hanging carpets

KS-1804_STI_250cm_05deg_carpets.png KS_T30_carpets.png

Impedance response

BMS 4590/M + SL250

BMS 4590/T + SL250

BMS 4550-16 + SL600

2 pcs 18N840 parallel in resistance enclosure


Lower treble has slightly metallic sound which is probably consequence of driver's resonance at 3 kHz. This is audible e.g. with acoustics strings and clean electric guitar. Peak is possible to cut by an extra notch in passive filter, or parameter equalizer as active. There is -1.6 dB Q=5.0 notch at 3.04 kHz in the DCX at the moment.
Sound of horn section is forward, but the feature is not annoying at all when level adjustment of woofer section is ok.
There is some sibilance in super tweeter, but it seems to be related to record/voice. High resolution and forwardness of speaker - as well as sound of power amplifier - probably underline features of some records and female vocals. Accuphase I'm using for horns is slightly edgy and bright for this application. It could be worth to minimize this feature by a parameter eq; about -1.1 dB Q=7.1 @ 7.3 kHz.
Bass range is dry and "humble" e.i. not making a big noise of itself. Authority and pressure at lower bass could be stronger, but this is highly related to volume, shape and construction materials of listening room. Our >150 m3 space is quite difficult to pressurize with other than huge vented systems. Capacity of woofer section and amplifiers are usually limiting maximum pressure way before horn section.
BMS coaxial driver has some audible distortion with extreme sound pressures (requiring ear protection anyway). Don't expect to hear clean violin groups at 110 dB.

Everything else is good and positive, but I won't make such a long list.


Woofer section with 2x 15" as vented or closed would be easier to implement for a common man. Woofer section of K&T's Big Bang is one possibility. It requires passive high-pass for horn - of course.

Created 01.09.2011
Updated 27.08.2012