KS-483 will be built mainly for HT front speaker. Secondary job is to be a spare speaker in 2-channel system when main loudspeaker project has faced it's final destiny - chain saw etc., or for some reason can't be located continuously in the living room (WAF of course!). Speakers should be capable for background music, and therefore listening range should be as wide as possible. This is advantage also while HT session, because audience is not in the sweet spot. Final location will be very close to front wall (partly brick), and therefore adequate attenuation to back direction is required. Speakers should be flat enough to allow some space to walk.

These requirements should be full-filled with Sonus Faber Elipsa/Stradivari -style - wide and heavily rounded speakers. Pair of W22EX001 and W15CY001 were found in the store, which allowed implementation very much like PMS Excel of Troels Gravesen. T25CF001 was selected as a tweeter, because it was ready in the store.

Technical specifications

General Floor standing, 3-way, vented
Dimensions 500 x 1000 x 237 mm (W x H x D)
Volume 40.5 + 7.5 liters
Tuning 32 Hz, vent 2x Ø47/130 mm
X/O frequencies 400 and 2600 Hz
Sensitivity 86 dB/2.83V/1m (estimated)
Drivers Seas T25CF001 (2007)
Seas W15CY001 (2004)
Seas W22EX001 (2005)
Impedance 8 ohm, min. 4.6 ohm @ 670 Hz




Few changes were made compared to PMS Excel. Circular wall contains 13 separate pcs of 25 mm thick MDF, to forge as wide and durable contacts as possible without epoxy cast. Corners of front wall are on elliptical arc - not on constant radius. Because of elliptical shape and limited amount of separate bars, corners of the front wall are sharper and width differences between flat parts are bigger than PMS Excel. This may generate some unwanted diffraction. Corners will be rounded some day.
Internal volumes and bracing are also different than PMS Excel. Enclosure for W15 has enlarged to 7.5 liters to allow sealed with Qtc=0.6. There is no vent at the moment. "Impossible to see the future is" (Yoda). Vertical bracing tries to minimize vibrations of front and back walls. Vents are located on the back side, a bit lower than woofer. Tuning frequency is 30 Hz. Tweeter is located 5 mm and mid 10 mm higher than PMS's.
These changes do not mean that I think there is something wrong in the PMS Excel. This is just different - an individual approach to same speaker. I do everything my own way.

Enclosure drawing: KS-483.pdf (30.03.2008)

IMG_1282.JPG IMG_1283.JPG
IMG_1284.JPG IMG_1288.JPG
IMG_1291.JPG IMG_3404.JPG
IMG_3408.JPG IMG_3409.JPG
IMG_3416.JPG IMG_3419.JPG


Filter schematic, version 4:

Crossover is very much like PMS Excel. The biggest difference is in the mid, where two main break-up nodes (8.4 + 11.2 kHz) are filtered with double notch with cross-connecting resistor R2051. That resistor decreases Q and attenuation of both notches, and reduces over-shooting around, and especially between nominal frequencies. This configuration does not contain less components, but they are not on the signal's path. Acoustical low-pass is more than 12 dB/oct and highest peak is -28 dB, which is ok. See measurement without tweeter.
Board lay-out is not optimal. At least L2041, L2061 and L3021 may interfere. See http://www.troelsgravesen.dk/coils.htm


Filter is built on chopping board (Orthex), using hot glue and cable ties. Terminals are regular lo-tech shit from grocery, to aspire maximum street credibility. Internal wiring is Supra Classic 2.5.

Parts list, version 4:

Resistors R P pcs
R3011 4.7 ohm metal oxide 4 W 2
R2051 47 ohm metal oxide 4 W 2
Capacitors C U pcs
C2061 1.0 uF MKT 160 V 2
C2041 2.7 uF MKT 160 V 2
C2041,C2061 3.3 uF MKT 160 V 4
C3011 5.6 uF MKT 160 V 2
C1021,C3031 10 uF MKT 160 V 4
C1021,C2011 47 uF MKT 160 V 4
Inductors L D R pcs
L2061 0.047 mH air 0.71 mm ? 2
L2041 0.06 mH air 0.71 mm ? 2
L3021 0.22 mH air 0.71 mm 0.39 ohm 2
L2031 0.82 mH air 1.4 mm 0.26 ohm 2
L2021 3.9 mH IT Corobar 1.4 mm 0.42 ohm 2
L1011 4.7 mH IT Corobar 1.4 mm 0.52 ohm 2


Crossover version 4

Simulated frequency response, measured drivers:

Simulated phase response:

Simulated impedance response. Drivers in the enclosure, vent 1x Ø47/100 mm (not final):


Crossover version 3

Frequency response in room, 250 cm, on-axis, adaptive filtered PIR + 300 ms, no smoothing:

Frequency response, 100 cm, 0-120°, adaptive filtered PIR + 300 ms, no smoothing:

Waterfall, 100 cm, 0°, 4 ms, no smoothing:

Near field responses of W22EX001:

Impedance response, vents 2x Ø47/160 mm, not final damping:

Harmonic distortion

90 dB / 1 m
Some box or other mechanical resonances (crossover base plate etc.) around 100...150 Hz.
2.6 kHz crossover frequency is still a bit high for 8.4 kHz break-up node of W15CY001.


These speakers have lot's of body. This feature is very often just an advantage, because many recordings - especially 70/80's - have very thin and dickless balance, probably due to vintage or boomy sounding monitors in small studios. Strong upper bass and lower midrange generate some vibrations in walls, windows and furniture, creating very lively and presence (they are here) sound.

Sound stage is stable but not razor-sharp when listening distance is more than 230 cm. Shorter distances with thick carpets on the floor create very sharp and detailed stereophonic illusion. Size of instruments and vocalist is a bit bigger than average. I feel sound stage as high, although most of the lower frequencies lie down. Maybe bigger projection of images make also sound stage higher. Images from upper mid are projected above tweeters' level. Tilt angle changes elevation of these images.

Sound is mostly clean, but the tweeter may have some harshness at highest octave. T25CF001 has damping-related resonance at 17 kHz.
There is quite good amount of authority, slam and attack for me.

Listening range is very wide. Tonal balance is surprisingly neutral and bright enough even 90° off-axis. Speaker's front wall acts like wide and flat wave guide, preventing diffractions and sharp turn-down in off-axis responses around 1 kHz - like many regular 2/3-way box speakers do.


Distance from tweeter to tweeter 240...260 cm. Listening distance 220...240 cm. Gap between front wall and speaker's back wall 20 cm or more. Tilt backwards 4°. Height of stands+spikes at least 75 mm, otherwise high bass and low mid may be boosted, creating boomy sound.
Speakers can be located near the front wall because of cardioid polar pattern from lower midrange. Location can be adjusted according bass level, where the best low and mid-bass level is achieved. Walls may suck mid-bass (or low-bass) when wall material and speaker/listener placement make bad combination.


These speakers sound very unique; rhythmic, lively and presence. They can handle many kind of recordings and music genres from classic to hard rock, focusing to music much more than observation of recordings with high resolution microscope.

Vifa XT25TG30 or Scan-Speak R2904/70000 could be the best option if treble preference is more clean and soft. At least it is closest to implementation of Sonus Faber Elipsa.

Page updated 29.09.2008