Note! This document is truncated. Full version with some useless small talk, simulations and prototypes is available only in Finnish.
Target was to make very directive speaker (without horns) to transfer
information from record to ears without disruptive room reflections.
This enables listening over longer distances in bad acoustics without
resonators and special treatment of room surfaces. Sound should be good
with good recordings and bad with bad recordings.
Vertical directivity should be much stronger than average - especially
to directions creating early reflections (-40°/+50°).
Horizontal directivity at midrange should be cardioid or hypercardioid,
and as constant as possible. Attenuation to 180° should be about 15 dB.
Polar response of bass range should be adjustable. Bass should be
weakly pressurizing.
Sufficient vertical directivity at midrange can be created with MMTMM. Tweeter will be ribbon/planar with moderate vertical directivity. Uppe rmidrange will be 2x4" and lower midrange 2x8", both in resistance enclosure.
Bass unit will be combination of dipole and monopole. Monopole driver is located very close to back side of dipole driver to keep polar pattern constant until crossover frequency. Dipole and monopole will have individual power amplifier which enables adjustability of polar pattern to suit for different rooms, placements and listeners. Polar pattern will be adjustable within dipole...hypercardioid...cardioid...(monopole). Dipole frame will be H-frame located on the monopole enclosure. Monopole driver is partly inside of back transmission line. Back side will be damped if frame is not symmetrical H-frame.
Dipole bass | 20-160 Hz | Peerless XXLS 830845 | 1 pcs |
---|---|---|---|
Monopole bass | 20-160 Hz | Peerless XLS 830452 | 1 pcs |
Lower midrange |
160-700 Hz | Seas W22EX001 | 2 pcs |
Upper midrange | 700-2500 Hz | Seas W12CY001 | 2 pcs |
Treble | 2500-20000 Hz | Bohlender Græbener Neo3-PDR | 1 pcs |
Dipole driver is SLS-12". H-fame is non-symmetric with damped back side. Monopole driver is XLS-10" and box is filled with polyester.
Measurements are done outdoors. Measuring distance is 200 cm and mic elevation 90 cm. Smoothing is 1/12 oct. and time window 100 ms. Angles in diagrams are horozontal only.
SLS = dipole, XLS = monopole
Attenuation to 90° is oly 20 dB because of non-symmetrical H.
Without equalization. Phase difrerence between monopole and dipole is almost constant 150°.
Why it's not 180°?
At this point frequency response of drivers are equalized to 180° to create cardioid polar response. Dipole has band-stop filter (210 Hz, -8 dB, Q=1) and level of monopole is decreased 2 dB. Phase difference (150°) has left as it was, and not much attention paid frequencies over 200 Hz.
This is cardioid. Average attenuation within 20...160 Hz:
45° | 90° | 135° | 180° |
---|---|---|---|
-1.3 dB | -5.8 dB | -14.0 dB | -28.9 dB |
Level of monopole is decreased 2.2 dB (compared to cardioid setting) to increase attenuation to 90°. Attenuation increases 1.0 dB, and direction of maximum attenuation rotates from 180° to 135°. Average attenuation within 20...160 Hz:
45° | 90° | 135° | 180° |
---|---|---|---|
-1.7 dB | -6.8 dB | -28.1 dB | -13.4 dB |
Level of monopole is decreased 4.4 dB (compared to cardioid setting) to get even more attenuation to 90°. Attenuation to 90° and 180° are now equal. This is good setting if vertical resonances are as bad as horizontal. This is close to hypercardioid where maximum attenuation is at 120°. Average attenuation within 20...160 Hz:
45° | 90° | 135° | 180° |
---|---|---|---|
-1.8 dB | -8.4 dB | -17.0 dB | -9.1 dB |
Back TL of dipole frame looks too deep because peak frequency (210 Hz) is too low. Dipole part of final speaker will be symmetrical H-frame to enable full dipole and peak at higher frequency.
Polar response of this kind of hybrid cardiod depends on measuring distance. The reasor for this is distance dependent frequency response and level of dipole unit. Dipole is close to pure driver in near field; there's no clear 6 dB/oct. HP slope or dipole peak. Distance dependency of dipole can be reduced by minimizing the frame. However, it's not possible to make ideal cardioid this way. Fortunately it is not necessary because ideal is not necessarily the best.
According my insvestications, the most flexible polar response is hypercardioid where level of monopole is decreased 3...6 dB compared to cardioid setting. It's more immune to placement than ideal cardioid. Also attenuation of vertical resonaces is better.
Dimensions are 240 x 240 x 280 mm. Internal net volume is about 6.5 liters (9 without internal limiters). Side ports are 2.0 mm steel, 308 pcs of ø6 mm holes each. Cotton fabric ('Finlayson Classic' linen) is clued with PVAc inside the plates. Small top port has 88 pcs ø6 mm holes. Leaking area is 90 % of Sd. Box is filled with polyester. There's also extra 6 layers of polyester (Univisio) on the back side of ports.
The result is good - no need to change anything. Attenuation to 180°
would increase by adding some polyester.
Effect of port resistance into impedance response:
- 'Ei koteloa'
= free air
- 'Kotelo, ei portteja'
= driver is in the box without resistance ports
- 'Ei vanua, Finlayson' = resistance ports are on but no polyester
- '6 vanua' = resistance ports on, filling and 6 extra layers of
polyester behind the ports
Dimensions are 150 x 360 x 260 mm. Internal net volume is about 3.8 liters (7.2 without internal limiters). Side ports are 2.0 mm steel, 88 pcs of ø6 mm holes each. Cotton fabric ('Meidän koti' linen) is clued with PVAc inside the plates. Leaking area is 99 % of Sd. Box is filled with polyester. There's also extra 16 layers of polyester(Univisio) on the back side of ports. Volume is too small because 16 layers of polyester was needed to reach target polar response.
Average of horizontal attenuation
Effect of port resistance into impedance response:
- 'Ei koteloa'
= free air
...
- '16 vanua' = resistance ports on, filling and 16 extra layers of
polyester behind the ports
Bohlender & Græbener Neo3-PDR planar with wave guide is installed on 150 x 20 mm baffle. Wave guide is 0.5 mm aluminum. Shape is calculated with the following formula:
Measuring distance 65 cm, time window 4 ms, no smoothing
Average of horizontal and vertical off-axis attenuation
THD is measured with SpectraLAB 4.32. Frequency step is 1/6 oct, voltage 1 Vrms, which dosn't reveal reaction with high sound pressures.
Worst results as spectrum:
Harwood Acoustics UR2.0, 7071 Hz
Harwood Acoustics UR2.0, 10 kHz
Visaton MHT12, 2500 Hz
I'm very disappointed with these ribbons and planars. However, B&G Neo3-PDR is really good.
Measuring distance (100 cm) is from mic to nearest midrange cone
- time window 20 ms, smoothing 1/3 oct.
Enclosure drawing:
Wtc7_dwg.pdf
(version 6, 29.06.2005)
Some old versios: Wtc7a_dwg.pdf, Wtc7b_dwg.pdf, Wtc7c_dwg.pdf
Resistance ports (1:1):
Resistanssilevyt.pdf
Material is local birch. Thickness of dipole baffle is 30 mm, upper midrange and tweeter box 20 mm and the rest 25 mm. Boxes are oiled/waxed with OsmoColor; first layer with pine pigment and second with clear (3101).
Total height is only 143 cm.
Speaker is active filtered with two Behringer DCX2496; one for the
woofers and second for the MMTMM. It needs 10 power amplifier channels.
Power amps are Rotel RMB-1066 and Yamaha XM6150. Cable hell...
Parameter files are not available anymore.
Crossover topology has strong effect to vertical directivity with MMTMM construction. Drivers are ideal in this simulation. Diameter of tweeter is only 20 mm.
LR2 has the strongest vertical directivity.
- living room, concrete walls and roof
- measuring distance 200 cm
Measuring distance 120 cm, time window 20 ms:
Average of horizontal attenuation within 30°...180°
Measuring distance 150 cm, woofers 1 cm:
Measuring distance 150 cm, woofers 1 cm:
SPL 96 dB/1m, measuring distance 50 cm:
Measuring distance 50 cm, time window 5 ms:
Tonal balance? Very good, but bass range was somehow hidden below
midrange.
Rhythmic? Not so
Dynamic? Bass no, midrange and tweeter ok
Timing? Not so good; too much phase distortion weakening the
transients. Weak attack of piano and kick drum.
Soundstage? Sharp, logical and forward (not behind speaker line).
Speakers can't create/reproduce soundstage.
Presence? Depends on record; sometimes excellent but usually not.
Floorless/floating images when vocals were near field recorded.
Tune? Good
Bass tracing? Some disintegration at upper bass...lower midrange, but
usually good.
Resolution? Very good but not full. DCXs were original and using analog
inputs which dropped resolution.
This project was very complex and challenging. Weakest points were too mild attack and too small size of bass radiator. Importance of time alignment was not so clear to me. Many gurus are still telling fairy tales that it's not audible. Maybe it is't by the ear drum but at leat I can feel it with my body. Size of bass radiator should be bigger than midrange. Otherwise midrange rules received sound information over the bass. I would call it unbalanced resolution and projection.
Page updated 08.11.2010