Prologue
Since 2011 I was beating my head against the simple problem: my two-way loudspeaker prototype came out great sounding but measured almost poorly. I hope you can trust me there is not much of exaggeration here - it sounded great despite being just one speaker (I did not have room big enough to assemble two of them). My other-half was so impressed that demanded that I build another one and place them in our bedroom immediately after cosmetic uplift (of course). It took longer than expected, but I'm not worried.
Out of all audible subjective qualities of the speaker prototype I'd like to mention great micro-dynamics, very detailed sound on low to moderate level, nice overall bass balance with generous midbass, large presentation with speaker almost disappearing (yes, listening from just mono source!), nice clarity even at low listening levels, pleasant near-field listening despite the solid size of loudspeaker and significant distance between LF/HF drivers. Also great was 'around the corner' listening experience as sound was coherent and pleasant also when listening from nearby room. The most important was the fact hat speaker brought less brain fatigue than I used to have while ever listening to other speakers. Also there was something special about spatial aspects of the overall sound that seemed to be much better then usual. Speaker was assembled and tested in the sonically untreated (except various clutter in shelves working as nice dispersers) and small (2x3x2.5m) listening room. See attached drawing to get the idea of the enclosure design.
So when speaker was finished and initially tweaked we sat there in small, cold workshop me and my wife listening through lots of great music that we both love for a couple of hours like it was the first time we hear them. I named speakers RAD TQWT as RAD also stands for 'joy'. Assuming this sounds too romantic for serious thread, more pragmatic description will follow
Any flaws? Yes. a) limited macrodynamics b) sound being slightly too forward (at this point there was no back tweeter connected and whizzer cones weren't removed yet). When i really wanted to "crank it up" midbass speakers reached cone breakup levels met with SET amp power limits. That's something surely can be addressed with adding more bands and making it 3- or 4-way with separate bass amp, but the goal was to build a best possible two-way. Did we ever wanted to listen much louder? No, it really was big and loud enough at non-distorted levels with just a few watts of applied power.
Since 2011 I was beating my head against the simple problem: my two-way loudspeaker prototype came out great sounding but measured almost poorly. I hope you can trust me there is not much of exaggeration here - it sounded great despite being just one speaker (I did not have room big enough to assemble two of them). My other-half was so impressed that demanded that I build another one and place them in our bedroom immediately after cosmetic uplift (of course). It took longer than expected, but I'm not worried.
Out of all audible subjective qualities of the speaker prototype I'd like to mention great micro-dynamics, very detailed sound on low to moderate level, nice overall bass balance with generous midbass, large presentation with speaker almost disappearing (yes, listening from just mono source!), nice clarity even at low listening levels, pleasant near-field listening despite the solid size of loudspeaker and significant distance between LF/HF drivers. Also great was 'around the corner' listening experience as sound was coherent and pleasant also when listening from nearby room. The most important was the fact hat speaker brought less brain fatigue than I used to have while ever listening to other speakers. Also there was something special about spatial aspects of the overall sound that seemed to be much better then usual. Speaker was assembled and tested in the sonically untreated (except various clutter in shelves working as nice dispersers) and small (2x3x2.5m) listening room. See attached drawing to get the idea of the enclosure design.
So when speaker was finished and initially tweaked we sat there in small, cold workshop me and my wife listening through lots of great music that we both love for a couple of hours like it was the first time we hear them. I named speakers RAD TQWT as RAD also stands for 'joy'. Assuming this sounds too romantic for serious thread, more pragmatic description will follow
Any flaws? Yes. a) limited macrodynamics b) sound being slightly too forward (at this point there was no back tweeter connected and whizzer cones weren't removed yet). When i really wanted to "crank it up" midbass speakers reached cone breakup levels met with SET amp power limits. That's something surely can be addressed with adding more bands and making it 3- or 4-way with separate bass amp, but the goal was to build a best possible two-way. Did we ever wanted to listen much louder? No, it really was big and loud enough at non-distorted levels with just a few watts of applied power.
Topology, components and materials of the speaker
After studying various DYI attempts I compiled it all and came to perhaps somewhat unconventional speaker design.
Topology:
- Two-way symmetric bipole (LF) + dipole (HF) with rear HF driver -3dB attenuated;
- Bass bipole part: two 12" Soviet production paper midbass drivers 4A32 from 70-ties. 95dB sensitivity, Qts around 0.27, Fr 50-60Hz depending on wear-level, 12Ohm each wired in parallel, in-phase magnets facing each other aka push-push configuration);
- 1-st order passive filter for HF, pseudo 2-nd order for LF with central crossover frequency at 1.3kHz;
- K73-9 type 3.9uF polyprophilene condenser in series with HF speakers and
- 1.2mH/0.7Ohm non-reference grade air-core coil inductor in series with LF speakers followed by RC Zobel filter parallel to speakers with RC values empirically chosen to lower Z raise of LF drivers at higher frequencies to exactly match Z rise of HF speakers (measured without series C connected to them), which turned out to be series RC of 10uF + 10Ohm parallel to LF drivers.
Back then I thought of matching Z curves as a new method of empirical simple passive filter design. As I later found out electrical phase matching may give some positive effect only if tweeter resonance is electrically damped with parallel RLC circuit and set amp is used. Of course, speaker gets less critical to cable resistance. Still i didn't worked it out yet - field for more investigation.
- Folded bass pseudo-horn with floor-facing exit aka TQWT. Here I invented traverse TQWT horn braces that also work as bracings for enslosure. The witdth of the enclosure allowed to do it without compromising gradual opening of the folded horn. This approach allowed me to create more wider and more shallow speaker enclosure.
- By gradually tilting speaker it is possible to change size of opening slot near the floor and somewhat adjust bass response to complement particular room mode resonances + some floor gain bonus for low end extension;
- Front side is sloped at about 6 degrees from vertical to lessen floor reflections and to make frontal/back sides non-parallel.
- Two Monacor DT-300 tweeters with complementary 17cm round elliptical WG-300 waveguides are wired in parallel and out-of-phase (later switched to series. 8Ohm, ~97dB sensitivity with WG @2kHz, steep response rolloff below 1.5kHz, gradual rolloff towards 20kHz), smooth, gradually decreasing dispersion above 1.6kHz.
Tweaks:
- Midbass drivers were originally 'fullrange' - paper cones with whizzers. I removed dust caps and whizzers therefore eliminating three unwanted peaks at 4, 8 and 12kHz which resulted in nice 6dB/oct rolloff above 500Hz without any more peaks left to suppress. It also removed unwanted forwardness observed at the very beginning. Also I glued pieces of 5mm thick felt carpet on basket frame insides to reduce amount of MF/HF reflected back to speaker.
- 2 of 4 purchased Monacor HF drivers were rendered unusable by trying to follow Troel Gravesen's mod on these speakers (removal of felt plate and magnetic flux). At the end I figured out that while the mod indeed contributed to more response extension towards 1kHz it decreased damping of tweeter's resonance and made it less pleasant sounding. Out of four Monacors I had one speaker left as new, while other was already half-modified (only felt disc removed and tunnell stuffed with long-fiber cotton instead hence Fr went down from 1.6kHz to around 1kHz). The latter become my front HF driver.
Materials used:
- Impregnated 21mm Baltic birch plywood cabinet with front panel glued from two sheets for double-thickness;
- 'Naked' WG-300 waveguides attached to HF speakers as-is without any damping mods;
- HF speakers placed on top of LF cabinets decoupled from them by thin foam and magnets supported by thicker foam with possibility to move them back and forth to find out the best position.
Source, amp and measurement rig
- PC with Presonus FireStudio Mobile FireWire soundcard (24/96);
- Tri VP-MINI 300 Mark II Class A SET integrated stereo valve amplifier;
- ARTA v1.7.1 measurement software + measurement box;
- Behringer ECM8000 microphone;
- K&M microphone stand;
- Custom soldered Jack-Jack-RCA-RCA cables for signal interconnects;
- Oehlbach 2.5sqmm loudspeaker cables.
Total investment in speakers including materials was about 400EUR. 2 of 4 bass drivers were a gift from a friend, other pair I got for about 100,- EUR, four Monacors + waveguides around 130,-EUR total, the rest went for Baltic birch plywood and carpenter's work sawing it plus for 2-component wood glue, screws, cables, connectors, binding posts, passive components.
After studying various DYI attempts I compiled it all and came to perhaps somewhat unconventional speaker design.
Topology:
- Two-way symmetric bipole (LF) + dipole (HF) with rear HF driver -3dB attenuated;
- Bass bipole part: two 12" Soviet production paper midbass drivers 4A32 from 70-ties. 95dB sensitivity, Qts around 0.27, Fr 50-60Hz depending on wear-level, 12Ohm each wired in parallel, in-phase magnets facing each other aka push-push configuration);
- 1-st order passive filter for HF, pseudo 2-nd order for LF with central crossover frequency at 1.3kHz;
- K73-9 type 3.9uF polyprophilene condenser in series with HF speakers and
- 1.2mH/0.7Ohm non-reference grade air-core coil inductor in series with LF speakers followed by RC Zobel filter parallel to speakers with RC values empirically chosen to lower Z raise of LF drivers at higher frequencies to exactly match Z rise of HF speakers (measured without series C connected to them), which turned out to be series RC of 10uF + 10Ohm parallel to LF drivers.
Back then I thought of matching Z curves as a new method of empirical simple passive filter design. As I later found out electrical phase matching may give some positive effect only if tweeter resonance is electrically damped with parallel RLC circuit and set amp is used. Of course, speaker gets less critical to cable resistance. Still i didn't worked it out yet - field for more investigation.
- Folded bass pseudo-horn with floor-facing exit aka TQWT. Here I invented traverse TQWT horn braces that also work as bracings for enslosure. The witdth of the enclosure allowed to do it without compromising gradual opening of the folded horn. This approach allowed me to create more wider and more shallow speaker enclosure.
- By gradually tilting speaker it is possible to change size of opening slot near the floor and somewhat adjust bass response to complement particular room mode resonances + some floor gain bonus for low end extension;
- Front side is sloped at about 6 degrees from vertical to lessen floor reflections and to make frontal/back sides non-parallel.
- Two Monacor DT-300 tweeters with complementary 17cm round elliptical WG-300 waveguides are wired in parallel and out-of-phase (later switched to series. 8Ohm, ~97dB sensitivity with WG @2kHz, steep response rolloff below 1.5kHz, gradual rolloff towards 20kHz), smooth, gradually decreasing dispersion above 1.6kHz.
Tweaks:
- Midbass drivers were originally 'fullrange' - paper cones with whizzers. I removed dust caps and whizzers therefore eliminating three unwanted peaks at 4, 8 and 12kHz which resulted in nice 6dB/oct rolloff above 500Hz without any more peaks left to suppress. It also removed unwanted forwardness observed at the very beginning. Also I glued pieces of 5mm thick felt carpet on basket frame insides to reduce amount of MF/HF reflected back to speaker.
- 2 of 4 purchased Monacor HF drivers were rendered unusable by trying to follow Troel Gravesen's mod on these speakers (removal of felt plate and magnetic flux). At the end I figured out that while the mod indeed contributed to more response extension towards 1kHz it decreased damping of tweeter's resonance and made it less pleasant sounding. Out of four Monacors I had one speaker left as new, while other was already half-modified (only felt disc removed and tunnell stuffed with long-fiber cotton instead hence Fr went down from 1.6kHz to around 1kHz). The latter become my front HF driver.
Materials used:
- Impregnated 21mm Baltic birch plywood cabinet with front panel glued from two sheets for double-thickness;
- 'Naked' WG-300 waveguides attached to HF speakers as-is without any damping mods;
- HF speakers placed on top of LF cabinets decoupled from them by thin foam and magnets supported by thicker foam with possibility to move them back and forth to find out the best position.
Source, amp and measurement rig
- PC with Presonus FireStudio Mobile FireWire soundcard (24/96);
- Tri VP-MINI 300 Mark II Class A SET integrated stereo valve amplifier;
- ARTA v1.7.1 measurement software + measurement box;
- Behringer ECM8000 microphone;
- K&M microphone stand;
- Custom soldered Jack-Jack-RCA-RCA cables for signal interconnects;
- Oehlbach 2.5sqmm loudspeaker cables.
Total investment in speakers including materials was about 400EUR. 2 of 4 bass drivers were a gift from a friend, other pair I got for about 100,- EUR, four Monacors + waveguides around 130,-EUR total, the rest went for Baltic birch plywood and carpenter's work sawing it plus for 2-component wood glue, screws, cables, connectors, binding posts, passive components.
Attachments
Last edited:
Lost in measurements
When measuring frontal output (same room mic placed ~60cm from middle of speaker centers, gating 10ms) several measured flaws became obvoious:
- There is a serious response fluctuation of +-5dB at 300-500 Hz with whole midbass range slightly elevated (somehow overall pleasant effect at least in given room dimensions);
- There is a 4dB dip at 1-1.5kHz;
- There is a 2dB wide elevation at 1.5-3kHz;
- There is a gradual rolloff (~3dB/oct) from 3kHz to 14kHz and steeper further on with smoothly narrowing dispersion (second measurement was taken about 30 degrees of-axis).
So measured frontal response unevenness was up to at least 6dB with most pronounced being dip at 1-1.5kHz. It surely was blended with reflected signal due to close distance to walls, so what I was measuring was rather power response (loudspeaker direct + room), and also room modes coming into play with all associated interference plague. I didn't use dual gating for analysis so bass curve is very aproximate.
Things I did to overcome 1kHz dip:
- tried to extend HF driver frequency response below 1.5kHz with mods. It made 1kHz dip less pronounced, but as a result speaker somehow sounded less pleasant;
- tried putting WG on separate 40cm baffle to reduce the dip that way and hence also to reduce 2kHz peak from WG dimensions. It actually made overall sound even worse. Sound became less spacious, more lifeless and boring;
- tried putting HF speaker in same baffle without WG. Most annoying sound, all airiness was lost, become very forward and unbalanced;
- tried disconnecting rear HF speaker and replacing it with resistive equivalent. Killed some of the spaciousness. Sound became too forward instead of hanging somewhere around the loudspeaker. I was confused of what to do next and placed serious doubt on trusting my ears and brain and even taste for even frequency balance.
I was really confused for a time.
When measuring frontal output (same room mic placed ~60cm from middle of speaker centers, gating 10ms) several measured flaws became obvoious:
- There is a serious response fluctuation of +-5dB at 300-500 Hz with whole midbass range slightly elevated (somehow overall pleasant effect at least in given room dimensions);
- There is a 4dB dip at 1-1.5kHz;
- There is a 2dB wide elevation at 1.5-3kHz;
- There is a gradual rolloff (~3dB/oct) from 3kHz to 14kHz and steeper further on with smoothly narrowing dispersion (second measurement was taken about 30 degrees of-axis).
So measured frontal response unevenness was up to at least 6dB with most pronounced being dip at 1-1.5kHz. It surely was blended with reflected signal due to close distance to walls, so what I was measuring was rather power response (loudspeaker direct + room), and also room modes coming into play with all associated interference plague. I didn't use dual gating for analysis so bass curve is very aproximate.
Things I did to overcome 1kHz dip:
- tried to extend HF driver frequency response below 1.5kHz with mods. It made 1kHz dip less pronounced, but as a result speaker somehow sounded less pleasant;
- tried putting WG on separate 40cm baffle to reduce the dip that way and hence also to reduce 2kHz peak from WG dimensions. It actually made overall sound even worse. Sound became less spacious, more lifeless and boring;
- tried putting HF speaker in same baffle without WG. Most annoying sound, all airiness was lost, become very forward and unbalanced;
- tried disconnecting rear HF speaker and replacing it with resistive equivalent. Killed some of the spaciousness. Sound became too forward instead of hanging somewhere around the loudspeaker. I was confused of what to do next and placed serious doubt on trusting my ears and brain and even taste for even frequency balance.
I was really confused for a time.
Attachments
- try to suspend the tweeter(s) with rubber bands ( the black coated ones)
Just build a headframe; the two tweeters can be made as one unit by placing the magnets on a short piece of tube ( I think you already did that).
What you reported corresponds to what I hear when tweeters are dependant on the enclosure vibrations .
The rubber band(s) must be at nearly full lenght stretched: this allows the load to oscillate at subsonic frequency, which is the purpose of doing it.
Just build a headframe; the two tweeters can be made as one unit by placing the magnets on a short piece of tube ( I think you already did that).
What you reported corresponds to what I hear when tweeters are dependant on the enclosure vibrations .
The rubber band(s) must be at nearly full lenght stretched: this allows the load to oscillate at subsonic frequency, which is the purpose of doing it.
Striking back!
- First I mainly attributed initial setup pleasantness to correct phase or rather time alignment at crossover frequency which I tried to explain in different thread and received lots of good criticism and also food for thought but no remedy for my initial confusion. Adjusting frontal HF driver placement initially did gave me a great deal of satisfaction by easily finding best HF speaker time alignment in respect to LF within +-0.5cm spot. My ears were pleased, but as we saw measurements were far form ideal.
- I then came to thought that first-order passive filters could play some importance to better blending of sound between HF/LF drivers. First it seemed counter-intuitive as more 'blur' would be expected with wave source being smeared between two speakers that are at distance from each other. In my humble opinion if crossing of drivers is done as low as 1kHz brain suffers less from the "blur" as below 2kHz it quickly gets harder to locate the "dual" source and brain (probably!) just isn't looking for spatial cues it just "doesn't bother" and takes it "as is". Also close placement will help. A simple test to prove or deny this theory would be to blindly compare biamped 1-st and 4-th order active filters on the same speakers. Must be super easy with amps equipped with programmable DSP based filtering like Hypex.
- Some suggested that there is no difference for HF to be connected as dipole or bipole. My ear said the opposite. I measured that given waveguided HF speakers in bipole setup have brought weird acoustic artefacts like very pronounced side lobes showing up as high as at 4kHz. So yes, non-correletion is possible if crossing is done very high and very steep. But why do it? I found an answer at friends house - his listening room ir rather wide than long. Connecting his two pair of Mission bookshelf speakers as bipoles (one on top of the other, upper one turned facing rear wall). It increased pleasantness by giving more midrange reflections through some side lobing and from rear wall in otherwise very boring sounding room. So first conclusion towards optimal directivity here - the further the side walls are from speakers the more pleasant becomes wider dispersion towards sides. Bipole may be more preferable here compared to other types of speakers unless they have very hard toe-in angle (more than 45 degrees) which may not contribute to evenness of direct signal for obvious reasons. Still side lobes must follow particular pattern. More on this on final thoughts.
- While comparing HF dipole vs bipole dispersion I noticed that for the dipole there is a dispersion peak at 1kHz where it becomes almost omni-, while on other frequencies it gave very nice figure 8 dispersion pattern. It seemed that due to low-order passive filter tweeter phase at resonance bends at 45 degrees and back following its electric phase response to the sides. This happens as ridiculously low (but in my opinion necessary due aforementioned spatial cues ceasing below 1kHz) passive filter crossover frequency and series capacitor's impedance is preventing HF speaker resonance to be fully damped by low output impedance of the amplifier to be attenuated enough. SET amp having low damping factor probably made it even worse. Anyway I realized that I now have wide emphasis of 1kHz on side response while still having dip in frontal mix of measured direct and power response. And it still sounds quite right! I didn't know I cold love reflected 1kHz so much. If what I enjoy is more spatial cues from otherwise hard to locate frequency band, then perceptual brain feels probably not only unburdened but given delusional increased easiness of hearing. Like having a small hearing aid apparatus that increases resolution Field for more research here.
- Following suggestion by someone (was it SpeakerDave?) I bought and read book by Floyd Toole "Sound Reproduction: The Acoustics and Psycho-acoustics of Loudspeakers and Rooms" Being quite fresh (2008) it brought a lot of demystification and valuable info mostly based on reference scientific experiments, so I really recommend the book to anyone that likes to gain more knowledge. At first there was no clear answer for my "very small-room non-liner response pleasantness" case. After re-reading most enjoyable parts I found a couple of things that I missed initially:
-an experiment that shows that for majority of people most pleasant D/R ratio for in room listening is up to 1 which means that most people prefer blend of direct and reflected signal in 1:1 proportion provided that reflections are significantly delayed compared to direct signal. That may explain dipole/supercardioid associated phenomena and seemed to answer why dipole HF sound was so open and spacious;
- head-related transfer function (see HRTF) that was showing almost 20dB(!) perceived dip at 2kHz in anechoich chamber when listening is made at exactly same distance to stereo source speakers and similar 6-10dB dip when listening in moderately reflective environment so whole stereo listening is flawed, welcome multichannel or ambisonics (hey, but I have one speaker here, allright?);
- equal loudness diagrams that in critical and upper midrange almost resembled my measured response.. inverted! Here in section "Relevance to sound level measurement and noise measurement" we read "Although the*A-weighting*curve, in widespread use for*noise*measurement, is said to have been based on the 40-phon Fletcher–Munson curve, research in the 1960s demonstrated that determinations of equal-loudness made using pure tones are not directly relevant to our perception of noise.[3]*This is because the cochlea in our inner ear analyzes sounds in terms of spectral content, each "hair-cell" responding to a narrow band of frequencies known as a*critical band. The high-frequency bands are wider in absolute terms than the low frequency bands, and therefore "collect" proportionately more power from a noise source. However, when more than one critical band is stimulated, the outputs of the various bands are summed by the brain to produce an impression of loudness. For these reasons Equal-loudness curves derived using noise bands show an upwards tilt above 1*kHz and a downward tilt below 1*kHz [emphasis mine] when compared to the curves derived using pure tones." My cautious conclusion here is: when reflected signal meets direct signal in comparable proportions it should be considered as direct signal mixed with spatially added (though pleasant) noise. That means either correcting both direct and reflected or just reflected signal response (better as it doesn't ruin direct signal response). And it must be corrected differently at different reproduced lodness levels and different D/R ratios. To get exact correction pattern as afunction of room size more research on that topic coud be handy. Still I believe it can be put as simply as - attenuate high frequencies in tight rooms slightly above 1kHz and raise midbass under 1kHz. With midbass baffles wider than 40cm (baffle step frequency lowered) combined with controlled directivity horns or waveguides (smoothly narrowing dispersion as a function of frequency) it all may come out very natural. Which is very close to what I have measured in my tiny room at quiet to moderate listening levels.
- My other guess is that increasing playback volume in small room may require even more HF attenuation. My observation is that in very small room any speaker with increasing loudness becomes too harsh sooner, perhaps cone breakups and THD/IMD are only half of the problem other significant half being unpleasant tonal balance at midrange when given loudness and reflection power being close to equal to that of direct signal while recognition capability of resulting at such "noise" level declining. This thought actually came to me only now while writing. One way to test it would be to bring my speaker to a bigger room and listen to it there. If strong room size dependence to mid-high attenuation proves to be true, then room-size dependent loudness correction must be adopted as an industry standard, all contemporary multimedia file standards must be updated to include mandatory RMS (average loudness) tag to get correction applied in correct amount, and also some calibration towards the room size and speaker sensitivity must be done for each loudness corrector in room it is supposed to be operational. Not an easy task
- First I mainly attributed initial setup pleasantness to correct phase or rather time alignment at crossover frequency which I tried to explain in different thread and received lots of good criticism and also food for thought but no remedy for my initial confusion. Adjusting frontal HF driver placement initially did gave me a great deal of satisfaction by easily finding best HF speaker time alignment in respect to LF within +-0.5cm spot. My ears were pleased, but as we saw measurements were far form ideal.
- I then came to thought that first-order passive filters could play some importance to better blending of sound between HF/LF drivers. First it seemed counter-intuitive as more 'blur' would be expected with wave source being smeared between two speakers that are at distance from each other. In my humble opinion if crossing of drivers is done as low as 1kHz brain suffers less from the "blur" as below 2kHz it quickly gets harder to locate the "dual" source and brain (probably!) just isn't looking for spatial cues it just "doesn't bother" and takes it "as is". Also close placement will help. A simple test to prove or deny this theory would be to blindly compare biamped 1-st and 4-th order active filters on the same speakers. Must be super easy with amps equipped with programmable DSP based filtering like Hypex.
- Some suggested that there is no difference for HF to be connected as dipole or bipole. My ear said the opposite. I measured that given waveguided HF speakers in bipole setup have brought weird acoustic artefacts like very pronounced side lobes showing up as high as at 4kHz. So yes, non-correletion is possible if crossing is done very high and very steep. But why do it? I found an answer at friends house - his listening room ir rather wide than long. Connecting his two pair of Mission bookshelf speakers as bipoles (one on top of the other, upper one turned facing rear wall). It increased pleasantness by giving more midrange reflections through some side lobing and from rear wall in otherwise very boring sounding room. So first conclusion towards optimal directivity here - the further the side walls are from speakers the more pleasant becomes wider dispersion towards sides. Bipole may be more preferable here compared to other types of speakers unless they have very hard toe-in angle (more than 45 degrees) which may not contribute to evenness of direct signal for obvious reasons. Still side lobes must follow particular pattern. More on this on final thoughts.
- While comparing HF dipole vs bipole dispersion I noticed that for the dipole there is a dispersion peak at 1kHz where it becomes almost omni-, while on other frequencies it gave very nice figure 8 dispersion pattern. It seemed that due to low-order passive filter tweeter phase at resonance bends at 45 degrees and back following its electric phase response to the sides. This happens as ridiculously low (but in my opinion necessary due aforementioned spatial cues ceasing below 1kHz) passive filter crossover frequency and series capacitor's impedance is preventing HF speaker resonance to be fully damped by low output impedance of the amplifier to be attenuated enough. SET amp having low damping factor probably made it even worse. Anyway I realized that I now have wide emphasis of 1kHz on side response while still having dip in frontal mix of measured direct and power response. And it still sounds quite right! I didn't know I cold love reflected 1kHz so much. If what I enjoy is more spatial cues from otherwise hard to locate frequency band, then perceptual brain feels probably not only unburdened but given delusional increased easiness of hearing. Like having a small hearing aid apparatus that increases resolution
Field for more research here.- Following suggestion by someone (was it SpeakerDave?) I bought and read book by Floyd Toole "Sound Reproduction: The Acoustics and Psycho-acoustics of Loudspeakers and Rooms" Being quite fresh (2008) it brought a lot of demystification and valuable info mostly based on reference scientific experiments, so I really recommend the book to anyone that likes to gain more knowledge. At first there was no clear answer for my "very small-room non-liner response pleasantness" case. After re-reading most enjoyable parts I found a couple of things that I missed initially:
-an experiment that shows that for majority of people most pleasant D/R ratio for in room listening is up to 1 which means that most people prefer blend of direct and reflected signal in 1:1 proportion provided that reflections are significantly delayed compared to direct signal. That may explain dipole/supercardioid associated phenomena and seemed to answer why dipole HF sound was so open and spacious;
- head-related transfer function (see HRTF) that was showing almost 20dB(!) perceived dip at 2kHz in anechoich chamber when listening is made at exactly same distance to stereo source speakers and similar 6-10dB dip when listening in moderately reflective environment so whole stereo listening is flawed, welcome multichannel or ambisonics (hey, but I have one speaker here, allright?);
- equal loudness diagrams that in critical and upper midrange almost resembled my measured response.. inverted! Here in section "Relevance to sound level measurement and noise measurement" we read "Although the*A-weighting*curve, in widespread use for*noise*measurement, is said to have been based on the 40-phon Fletcher–Munson curve, research in the 1960s demonstrated that determinations of equal-loudness made using pure tones are not directly relevant to our perception of noise.[3]*This is because the cochlea in our inner ear analyzes sounds in terms of spectral content, each "hair-cell" responding to a narrow band of frequencies known as a*critical band. The high-frequency bands are wider in absolute terms than the low frequency bands, and therefore "collect" proportionately more power from a noise source. However, when more than one critical band is stimulated, the outputs of the various bands are summed by the brain to produce an impression of loudness. For these reasons Equal-loudness curves derived using noise bands show an upwards tilt above 1*kHz and a downward tilt below 1*kHz [emphasis mine] when compared to the curves derived using pure tones." My cautious conclusion here is: when reflected signal meets direct signal in comparable proportions it should be considered as direct signal mixed with spatially added (though pleasant) noise. That means either correcting both direct and reflected or just reflected signal response (better as it doesn't ruin direct signal response). And it must be corrected differently at different reproduced lodness levels and different D/R ratios. To get exact correction pattern as afunction of room size more research on that topic coud be handy. Still I believe it can be put as simply as - attenuate high frequencies in tight rooms slightly above 1kHz and raise midbass under 1kHz. With midbass baffles wider than 40cm (baffle step frequency lowered) combined with controlled directivity horns or waveguides (smoothly narrowing dispersion as a function of frequency) it all may come out very natural. Which is very close to what I have measured in my tiny room at quiet to moderate listening levels.
- My other guess is that increasing playback volume in small room may require even more HF attenuation. My observation is that in very small room any speaker with increasing loudness becomes too harsh sooner, perhaps cone breakups and THD/IMD are only half of the problem other significant half being unpleasant tonal balance at midrange when given loudness and reflection power being close to equal to that of direct signal while recognition capability of resulting at such "noise" level declining. This thought actually came to me only now while writing. One way to test it would be to bring my speaker to a bigger room and listen to it there. If strong room size dependence to mid-high attenuation proves to be true, then room-size dependent loudness correction must be adopted as an industry standard, all contemporary multimedia file standards must be updated to include mandatory RMS (average loudness) tag to get correction applied in correct amount, and also some calibration towards the room size and speaker sensitivity must be done for each loudness corrector in room it is supposed to be operational. Not an easy task
Getting closer to the source
- At this point it came to me that same 1kHz dispersion stuff is happening downwards in LF/HF blending zone. So by reducing zobel resistor to 2Ohms (sorry, tubes!) and hence shifting electrical and acoustical phase of LF driver towards better blending with frontal tweeter with already shifted phase response due undamped resonance I managed to eliminate dip at 1kHz. Was it better? Well, maybe slightly, but by sacrificing some significant other parts of tonal balance. So I switched back. I was tempted to do correction more politely from opposite side by adding LCR impedance correction circuit in parallel to HF units, but my second kid was born at that time and suddenly spare time was available no more.
- I was still curious about elevation at 1.5-3kHz and possible relation to listening preference. So I re-read some of Linkwitz papers and dropped a mail to Floyd Toole with a question: what if slight emphasis around 2kHz could compensate to ill effects related to HRTF and hence give more balanced perception and elevated listening pleasure. He promptly and kindly responded (great guy, isn't he?) that (as stated in his book) HRTF associated stereo flaw ans associated bluntness of perceived sound is audible only in very narrow sweet spot (perhaps +-1cm of a head movement or even less). He was probably right, so I dropped this possible explanation for pleasantness from slightly elevated 1.5-3kHz as not related and it is still unresolved for me. The things I can think of is that reflection pattern in ordinary room may be absorbing this range more than other bands and even more in a well damped environment that breaks evenness of otherwise close-to-even power response. Or, perhaps our brain is just happy receiving slightly more pronounced spatial cues in this very sensitive midrange region.
- I also tried connecting simple gainclone chip-amp to try non-valve version of the same speaker. Increase in bass 'tightness' (less dip) at resonant frequency due better damping was noticeable. Unfortnately 'bigness' of the midrange and highs collapsed, mids become much less of a sweet honypot. Sound now was coming from the speakers not some ephemera cloud surrounding it. I've heard same differences with other speakers, so no big surprises here, still somewhat disappointing.
- While being most of the time in a constant hurry serving one or both toddlers and keeping up with everyday business I came to thought that I must try Econowave to get better blending with LF driver and more even power response. So I ordered a pair of 18sound XT1086 horns and BMC BMS 4538-8 compression drivers. I attenuated them to match the level of LFs and used same 1-st order filter. Judging by how attached horn smoothed jagged impendance curve of naked BMC driver i guessed they could be a good match. Then I tried connecting it as a substitute of frontal Monacors.. and didn't like it much. Nearfield listening pleasure was gone. Narrower vertical dispersion was not singing happily towards my face as did WG-300 which is a round one and smaller size. While delivering outstanding dynamics previous spatialness and ease were somewhat decreased, sound becoming focused as highs were more forward. Conlusion - deeper and larger horns are better suited for large rooms, mid- to far-field listening. For small rooms and near- to mid-field shallow round waveguide seems to be much better companion.
- Meanwhile I found something worth by peeking into this focused man's thread Provided small image is taken from Zwiker's and Fastl's book "Psychoacoustics: facts and models". So I'm attaching the picture that gave me another good thinking before immersing into lately habitual irregular night's sleep.
What we can see here is suggested equalization for in-room diffuse sound-field defined as a function of open-air response. Quite resembling my measured near-field response which is most likely sum of direct and reflected signals except the difference at the very top end and elevated midbass in my case, which I find rather nice and full than annoying.
- To get more details on picture I ordered the Second edition of the book from ebay second-hand. It came as written-off from Munich university after probably being replaced by Third edition. I didn't know I'd be visiting the Munich area quite soon. Seems that some items in life are working like magnets to some important places, at least for me
Unfortunately there wasn't much of explanation of this EQ curve except analysis of loudness curves which were just about inverse depending of the loudness level. My conclusion from this picture is that for in-room listening for frontally lineary responding speaker the best dispersion pattern curve would follow this suggested equalization line, or directivity index curve following just inverted version of it. From my measurements seems that I was just lucky to stumble upon the optimal combination of different aspects while trying to build a rather non-conventional design and to get very close to this.
- Unless going for 8-channel multi-amping with correcting EQs that will match the line to rear speakers amped separately in dipoles or bipoles (I'd like to hear that!) this seems to be possible to also accomplish acoustically, still with design variations for different room sizes. For example my guess is that Earl Geddes highly praised speakers are getting to this optimally diffused sound-field EQ by combining wider toe-in with controlled dispersion, so some of direct signal is bounced off opposite wall, perhaps giving required diffused sound field elevation at 2-3kHz and perhaps even within the right amount of attenuation. Copied-by-many Econowave speaker also may be complementary to this curve to some degree given the room is big enough. It would be moderately easy to measure if you have an-echoic chamber or ground-plane measurement setup. Unfortunately not many can have access to that kind of measurement rig.
- One more thing: according to the end part of correction curve line we need to elevate 10-20kHz range in that same diffuse sound-field. Also if room is treated or stuffed with furniture covered with fabric we can expect more losses here (I always preferred leather in the car seating as better complementing with the HF sound so EQ compensation must be perhaps twice as high as shown. This intuitively asks for extended durability of such speakers or having very small horns (~1.5cm in diameter), or combination of both. I tried this type of sound-field correction by just adding small rear-and-upwards facing Vifa tweeters with small enough series capacitor to my friend's monopole Mission speakers. The improvement on spatial presentation was audible but the feeling was like "we wanted more of it!". So this is something yet to be done - a nice and inexpensive way to upgrade any existing system to get better sound inside the room.
- I remember myself reading somewhere at Linkwitz's site that humans tend to prefer high frequencies to be coming from slightly above ear level (perhaps genetically encoded sentiment from our ancestors for bird's noises in primordial forest?). I totally agree with that. Added rear tweeter can help with that through ceiling and back wall reflections if elevation of speakers isn't possible for some reason.
- Out of curiosity I also ordered and read the book "Get better sound" by Jim Smith. Being almost totally opposite character to Floyd Toole, Jim is rather pragmatic experimentalist, not a scientist. Several times in the book you will read something like "I don't know why, but it works". If Floyd's book was read before this one you'll be able to answer to most of the questions except some doubtful stuff like "when pulling amp slightly out of the rack the sound gets a bit warmer". If you can tolerate and skip such observations (there aren't very many of this type), the book give indeed some good advice on how to set up your favorite listening spot. Also it uncovers some low-level signal resolution related advantages of analog over digital recording. Best advice from the book regarding dispersion imo was to put some small trees near walls on both sides of the room halfway to speakers to effectively scatter first sidewall reflection. Usage of scattering panels was encouraged also by Toole with coverage of some of the vendors and panel types. Generally this just helps to make diffuse sound-field even more diffuse and enjoyable
- At this point it came to me that same 1kHz dispersion stuff is happening downwards in LF/HF blending zone. So by reducing zobel resistor to 2Ohms (sorry, tubes!) and hence shifting electrical and acoustical phase of LF driver towards better blending with frontal tweeter with already shifted phase response due undamped resonance I managed to eliminate dip at 1kHz. Was it better? Well, maybe slightly, but by sacrificing some significant other parts of tonal balance. So I switched back. I was tempted to do correction more politely from opposite side by adding LCR impedance correction circuit in parallel to HF units, but my second kid was born at that time
and suddenly spare time was available no more.- I was still curious about elevation at 1.5-3kHz and possible relation to listening preference. So I re-read some of Linkwitz papers and dropped a mail to Floyd Toole with a question: what if slight emphasis around 2kHz could compensate to ill effects related to HRTF and hence give more balanced perception and elevated listening pleasure. He promptly and kindly responded (great guy, isn't he?) that (as stated in his book) HRTF associated stereo flaw ans associated bluntness of perceived sound is audible only in very narrow sweet spot (perhaps +-1cm of a head movement or even less). He was probably right, so I dropped this possible explanation for pleasantness from slightly elevated 1.5-3kHz as not related and it is still unresolved for me. The things I can think of is that reflection pattern in ordinary room may be absorbing this range more than other bands and even more in a well damped environment that breaks evenness of otherwise close-to-even power response. Or, perhaps our brain is just happy receiving slightly more pronounced spatial cues in this very sensitive midrange region.
- I also tried connecting simple gainclone chip-amp to try non-valve version of the same speaker. Increase in bass 'tightness' (less dip) at resonant frequency due better damping was noticeable. Unfortnately 'bigness' of the midrange and highs collapsed, mids become much less of a sweet honypot. Sound now was coming from the speakers not some ephemera cloud surrounding it. I've heard same differences with other speakers, so no big surprises here, still somewhat disappointing.
- While being most of the time in a constant hurry serving one or both toddlers and keeping up with everyday business I came to thought that I must try Econowave to get better blending with LF driver and more even power response. So I ordered a pair of 18sound XT1086 horns and BMC BMS 4538-8 compression drivers. I attenuated them to match the level of LFs and used same 1-st order filter. Judging by how attached horn smoothed jagged impendance curve of naked BMC driver i guessed they could be a good match. Then I tried connecting it as a substitute of frontal Monacors.. and didn't like it much. Nearfield listening pleasure was gone. Narrower vertical dispersion was not singing happily towards my face as did WG-300 which is a round one and smaller size. While delivering outstanding dynamics previous spatialness and ease were somewhat decreased, sound becoming focused as highs were more forward. Conlusion - deeper and larger horns are better suited for large rooms, mid- to far-field listening. For small rooms and near- to mid-field shallow round waveguide seems to be much better companion.
- Meanwhile I found something worth by peeking into this focused man's thread Provided small image is taken from Zwiker's and Fastl's book "Psychoacoustics: facts and models". So I'm attaching the picture that gave me another good thinking before immersing into lately habitual irregular night's sleep.
What we can see here is suggested equalization for in-room diffuse sound-field defined as a function of open-air response. Quite resembling my measured near-field response which is most likely sum of direct and reflected signals except the difference at the very top end and elevated midbass in my case, which I find rather nice and full than annoying.
- To get more details on picture I ordered the Second edition of the book from ebay second-hand. It came as written-off from Munich university after probably being replaced by Third edition. I didn't know I'd be visiting the Munich area quite soon. Seems that some items in life are working like magnets to some important places, at least for me
Unfortunately there wasn't much of explanation of this EQ curve except analysis of loudness curves which were just about inverse depending of the loudness level. My conclusion from this picture is that for in-room listening for frontally lineary responding speaker the best dispersion pattern curve would follow this suggested equalization line, or directivity index curve following just inverted version of it. From my measurements seems that I was just lucky to stumble upon the optimal combination of different aspects while trying to build a rather non-conventional design and to get very close to this.
- Unless going for 8-channel multi-amping with correcting EQs that will match the line to rear speakers amped separately in dipoles or bipoles (I'd like to hear that!) this seems to be possible to also accomplish acoustically, still with design variations for different room sizes. For example my guess is that Earl Geddes highly praised speakers are getting to this optimally diffused sound-field EQ by combining wider toe-in with controlled dispersion, so some of direct signal is bounced off opposite wall, perhaps giving required diffused sound field elevation at 2-3kHz and perhaps even within the right amount of attenuation. Copied-by-many Econowave speaker also may be complementary to this curve to some degree given the room is big enough. It would be moderately easy to measure if you have an-echoic chamber or ground-plane measurement setup. Unfortunately not many can have access to that kind of measurement rig.
- One more thing: according to the end part of correction curve line we need to elevate 10-20kHz range in that same diffuse sound-field. Also if room is treated or stuffed with furniture covered with fabric we can expect more losses here (I always preferred leather in the car seating as better complementing with the HF sound
so EQ compensation must be perhaps twice as high as shown. This intuitively asks for extended durability of such speakers or having very small horns (~1.5cm in diameter), or combination of both. I tried this type of sound-field correction by just adding small rear-and-upwards facing Vifa tweeters with small enough series capacitor to my friend's monopole Mission speakers. The improvement on spatial presentation was audible but the feeling was like "we wanted more of it!". So this is something yet to be done - a nice and inexpensive way to upgrade any existing system to get better sound inside the room. - I remember myself reading somewhere at Linkwitz's site that humans tend to prefer high frequencies to be coming from slightly above ear level (perhaps genetically encoded sentiment from our ancestors for bird's noises in primordial forest?). I totally agree with that. Added rear tweeter can help with that through ceiling and back wall reflections if elevation of speakers isn't possible for some reason.
- Out of curiosity I also ordered and read the book "Get better sound" by Jim Smith. Being almost totally opposite character to Floyd Toole, Jim is rather pragmatic experimentalist, not a scientist. Several times in the book you will read something like "I don't know why, but it works". If Floyd's book was read before this one you'll be able to answer to most of the questions except some doubtful stuff like "when pulling amp slightly out of the rack the sound gets a bit warmer". If you can tolerate and skip such observations (there aren't very many of this type), the book give indeed some good advice on how to set up your favorite listening spot. Also it uncovers some low-level signal resolution related advantages of analog over digital recording. Best advice from the book regarding dispersion imo was to put some small trees near walls on both sides of the room halfway to speakers to effectively scatter first sidewall reflection. Usage of scattering panels was encouraged also by Toole with coverage of some of the vendors and panel types. Generally this just helps to make diffuse sound-field even more diffuse and enjoyable
Attachments
Munich High-End: notes from the field
My dear kids just had grown-up to the age when they can be left with grandpas-mas and I decided to finally calibrate my inner hearing against the sound coming from "reference grade" systems. So I went to Munich High-End show this year. One and-a-half days at the show I was running through the halls but mostly stopping for awhile in dedicated listening rooms in so called "Atrium" zones and making notes of 107 short auditions in total.
- First day was mostly disappointing at the beginning, including nearly unbearable sound including some of renowned brands I won't mention here for ethical reasons. Munich MOC trade fair complex isn't providing probably the best environment for sound judgement, but even considering that only 1/10 of the rooms were as Jim Smith would probably say "carefully prepared for auditioning". Lack of dispersers and scatterers made lots of systems sound really unbalanced in a given room.
- While noting also subbass level (aka bottom end) I accounted that it is too much room-mode dependent and given the unequal conditions among vendors I didn't take it into account for final judgement.
- Subjective criteria I used for judgement were presence or lack of:
○ Airiness, openness of the sound (matched time coherence in mid- and HF range);
○ Clarity (low THD/IMD%, good micro-dynamics);
○ Extended frequency range (smooth response with good top HF and bottom LF);
○ Lack of artifacts (lack of audible resonances within whole spectrum except subbass);
○ Large, panoramic soundstage (good dispersion, pleasant interaction with the room);
○ Macro dynamics (clarity and balance at being loud).
- Now we see that three out of six criteria are very room and baffle-design dependent especially the last one. So my ratings are very approximate.
- Besides some important stuff that i missed for different reasons on 10 point scale I did gave 21 systems rating of 7 as sounding "great", 10 systems with 8 as "especially pleasant sounding" and only one system with 9 out of 10. The latter is MBL's flagship model 101 Extreme with loads of amp power, huge dedicated stereo subs, and guess what - auditioning room so densily stuffed with palm trees that it smelled like a tropical house in some Bothanic garden. Trees were all around the walls and also in lines in the middle of the room behind listening seats. Perhaps this is how renowned true omni- speaker manufacturer deals with the Zwicker's suggested diffusion EQ. The sound was a bit metallic so I couldn't really give 10, every other aspect was flawless.
- All the rest ratings below 7 had either one on more clearly audible flaws within the frequency range of lower midbass to highest heights and I didn't spend much of the time waiting for more "revealing" record to come.
- So here come 10 systems that scored at least 8/10 plus 1 pair of very pleasant sounding headphones:
Genelec 8260A
Kawero Vivace
Living Voice Vox Olympian
mbl 116F
Lansche Audio (model?)
Rockport Technologies Altair II
T+A Solitaire
TOTEM Acoustic Element
Tune Audio Anima
ULTRASONE Edition 8 (headphones)
Wilson Audio Alexia
- And also 21 of 7/10 (which is also a really good mark by all means with just one or too slightly audible flaws).
ADAM Artist 5
Amphion Either Helium 520Cs placed vertically or new MTM-type model
ASW Magadis
AUDIO PHYSIC Avantera
AURALiC DAC, preamp, amp, atskaņo uz Audeze
Ayon Audio Blackfire XS
Canton ???
Crystal Cable Arabesque
DLS New round baffle on-wall model w on-wall sub
Dynaudio ???
ELAC MicroMagic
Estelon ???
Göbel High End Epoque
Kaiser Akustik ??
Kharma Exquisite
Lawrence Audio Sonic Art
Magnat Quantum Signature
Mc Intosh Home Theater
Soundkaos Wave 40
TRIANGLE Magellan Cello II
Zingali Acoustics Client Name 1.5 Evo
- Some models weren't familiar for me, sorry about that.
- If you search for images of these speakers it is amazing to see so different baffles and drivers altogether having very pleasant sound in their own very special way. One thing that encourages me regarding our topic - those having the best dispersion were either using waveguided dome tweeters, like Amphions (being so outstanding in spatial aspects that I felt almost like home or waveguided AMT/RAAL/other planar speakers. WG for Amphions seemed identically shaped as WG-300 and baffle witdth was exactly same as WGs or just a bit wider: 17-18cm. I basically not a fan of MTM but as the speakers were lifted slightly above ear level the panorama was just right, wide and spacious.
- So whats so magic about 17cm baffle width? My guess is that it gives us pleasant 1kHz side-wise elevation through baffle step. 17cm = F3 of 680Hz so above that goes elevation up to 2-3kHz that gives it some dip in front and most probably increased 0.6-1.5kHz output to the sides just like on Zwicker's attenuation curve. Yet to be confirmed. Many of the best speakers in the show had baffles right this wide.
- By the way, what's the baffle width of upper midrange speaker of Linkwitz's LX521 and NaO Note RSII?
- I hope you found some of the information useful.
Jānis Irbe,
Riga, Latvia
My dear kids just had grown-up to the age when they can be left with grandpas-mas and I decided to finally calibrate my inner hearing against the sound coming from "reference grade" systems. So I went to Munich High-End show this year. One and-a-half days at the show I was running through the halls but mostly stopping for awhile in dedicated listening rooms in so called "Atrium" zones and making notes of 107 short auditions in total.
- First day was mostly disappointing at the beginning, including nearly unbearable sound including some of renowned brands I won't mention here for ethical reasons. Munich MOC trade fair complex isn't providing probably the best environment for sound judgement, but even considering that only 1/10 of the rooms were as Jim Smith would probably say "carefully prepared for auditioning". Lack of dispersers and scatterers made lots of systems sound really unbalanced in a given room.
- While noting also subbass level (aka bottom end) I accounted that it is too much room-mode dependent and given the unequal conditions among vendors I didn't take it into account for final judgement.
- Subjective criteria I used for judgement were presence or lack of:
○ Airiness, openness of the sound (matched time coherence in mid- and HF range);
○ Clarity (low THD/IMD%, good micro-dynamics);
○ Extended frequency range (smooth response with good top HF and bottom LF);
○ Lack of artifacts (lack of audible resonances within whole spectrum except subbass);
○ Large, panoramic soundstage (good dispersion, pleasant interaction with the room);
○ Macro dynamics (clarity and balance at being loud).
- Now we see that three out of six criteria are very room and baffle-design dependent especially the last one. So my ratings are very approximate.
- Besides some important stuff that i missed for different reasons on 10 point scale I did gave 21 systems rating of 7 as sounding "great", 10 systems with 8 as "especially pleasant sounding" and only one system with 9 out of 10. The latter is MBL's flagship model 101 Extreme with loads of amp power, huge dedicated stereo subs, and guess what - auditioning room so densily stuffed with palm trees that it smelled like a tropical house in some Bothanic garden. Trees were all around the walls and also in lines in the middle of the room behind listening seats. Perhaps this is how renowned true omni- speaker manufacturer deals with the Zwicker's suggested diffusion EQ. The sound was a bit metallic so I couldn't really give 10, every other aspect was flawless.
- All the rest ratings below 7 had either one on more clearly audible flaws within the frequency range of lower midbass to highest heights and I didn't spend much of the time waiting for more "revealing" record to come.
- So here come 10 systems that scored at least 8/10 plus 1 pair of very pleasant sounding headphones:
Genelec 8260A
Kawero Vivace
Living Voice Vox Olympian
mbl 116F
Lansche Audio (model?)
Rockport Technologies Altair II
T+A Solitaire
TOTEM Acoustic Element
Tune Audio Anima
ULTRASONE Edition 8 (headphones)
Wilson Audio Alexia
- And also 21 of 7/10 (which is also a really good mark by all means with just one or too slightly audible flaws).
ADAM Artist 5
Amphion Either Helium 520Cs placed vertically or new MTM-type model
ASW Magadis
AUDIO PHYSIC Avantera
AURALiC DAC, preamp, amp, atskaņo uz Audeze
Ayon Audio Blackfire XS
Canton ???
Crystal Cable Arabesque
DLS New round baffle on-wall model w on-wall sub
Dynaudio ???
ELAC MicroMagic
Estelon ???
Göbel High End Epoque
Kaiser Akustik ??
Kharma Exquisite
Lawrence Audio Sonic Art
Magnat Quantum Signature
Mc Intosh Home Theater
Soundkaos Wave 40
TRIANGLE Magellan Cello II
Zingali Acoustics Client Name 1.5 Evo
- Some models weren't familiar for me, sorry about that.
- If you search for images of these speakers it is amazing to see so different baffles and drivers altogether having very pleasant sound in their own very special way. One thing that encourages me regarding our topic - those having the best dispersion were either using waveguided dome tweeters, like Amphions (being so outstanding in spatial aspects that I felt almost like home
or waveguided AMT/RAAL/other planar speakers. WG for Amphions seemed identically shaped as WG-300 and baffle witdth was exactly same as WGs or just a bit wider: 17-18cm. I basically not a fan of MTM but as the speakers were lifted slightly above ear level the panorama was just right, wide and spacious.- So whats so magic about 17cm baffle width? My guess is that it gives us pleasant 1kHz side-wise elevation through baffle step. 17cm = F3 of 680Hz so above that goes elevation up to 2-3kHz that gives it some dip in front and most probably increased 0.6-1.5kHz output to the sides just like on Zwicker's attenuation curve. Yet to be confirmed. Many of the best speakers in the show had baffles right this wide.
- By the way, what's the baffle width of upper midrange speaker of Linkwitz's LX521 and NaO Note RSII?
- I hope you found some of the information useful.
Jānis Irbe,
Riga, Latvia
Last edited:
The problem is that when you open with a premise that goes against everything that I know it is really hard to take the rest serious. Stating that the speaker that you made yourself sounds great is such a classic foundation for an entirely erroneous discussion that no one is going to accept it.
Keep the discussion based in non-subjective facts and you might get somewhere. Lead off and follow through with subjectivism and all is doomed.
Hi Jānis,
I haven't read everything here yet. But you may find my suggestion about how to eq. wide dispersion speakers interesting (not sure if you have seen it before).
Interim Conclusions
I haven't read everything here yet. But you may find my suggestion about how to eq. wide dispersion speakers interesting (not sure if you have seen it before).
Interim Conclusions
Yes. The gating applied to results is 10ms so room interaction with measurements must be significant.
But that's the whole point - if combined direct+reflected measurement follows desired power response according to suggested psycho-acoustical in-room correction, and my ear indeed finds it pleasant, then I see it as a good proof that that's how corrections should be done in general.
Dear Earl, with all my respect that's totally up to you. I included some measurements further on plus I don't think it goes "against", rather "along". Or I don't get which statement you mean exactly.
Most room related concepts are based on findings done in anechoic chamber or IEC 268-13 standard room (26' x 16' x 10') or both. Most of people live in smaller rooms and I believe that's a field that asks for more studies towards correction and optimal dispersion pattern of the loudspeakers. Based on my subjective perception I think I found the right direction but you're right that this fact alone doesn't prove it. That's why I looked for explanation and possibly found it in diffuse sound field correction suggested by Zwicker/Fastl. At least it looks reasonably similar to what I got by occasion.
Last edited:
Unfortunately I don't have PIR files anymore just screenshots from analysis window. I'm going to redo them anyway. I believe applied gating on analysis window was 10ms.
Excellent, thanks! We came to the same observations but different ways.
"The effects are proportional to the volume setting" - are you sure about this? I guess with increase of reproduction volume (especially in smaller rooms) reflections could build up to be percieved to a larger degree than direct signal so more of attenuation can be required with increased volume.
"What correlation with the HF dip contributes to the sensation that makes you feel as if you are in a larger room" - I guess because high frequencies have faster DD (double-distance) decay rate compared to mids. It was mentioned in Toole's book. That's just how big rooms sound: HF reflections are attenuated as they lose their energy travelling long distances.
When it's my turn ?
Oh sorry, I posted the explanation (pic) of my example on another thread
http://www.diyaudio.com/forums/full-range/236880-floating-speaker-enclosures.html
Last edited:
Without gating out the room reflections, you have no idea what the first-arrival (anechoic) response is, which makes crossover design impossible.
Set the left edge of the time window a bit to the left of the initial rise-time, and the right edge just before the first room reflection. That might only give you 3 mSec, which in turn only gives about 300~400 Hz frequency resolution, but that's good enough to see if the drivers are reasonably flat and if the crossover is flat.
The rule-of-thumb in crossover design is to use sharp enough slopes so that severe ripples are at least 15 to 20 dB below the average response level. Since this is a difficult requirement to meet without 24 dB/octave (or steeper) slopes, it's an incentive to use drivers that have smooth rolloff regions, at least for an octave above the intended crossover frequency.
Set the left edge of the time window a bit to the left of the initial rise-time, and the right edge just before the first room reflection. That might only give you 3 mSec, which in turn only gives about 300~400 Hz frequency resolution, but that's good enough to see if the drivers are reasonably flat and if the crossover is flat.
The rule-of-thumb in crossover design is to use sharp enough slopes so that severe ripples are at least 15 to 20 dB below the average response level. Since this is a difficult requirement to meet without 24 dB/octave (or steeper) slopes, it's an incentive to use drivers that have smooth rolloff regions, at least for an octave above the intended crossover frequency.
When it's my turn ?
Oh sorry, I posted the explanation (pic) of my example on another thread
http://www.diyaudio.com/forums/full-range/236880-floating-speaker-enclosures.html
Sorry, I missed your initial post. Thanks for suggestion! I don't think it is related to vibrations, and foam layers I used provides a certain degree of decoupling.
Right. Still I like to use 1-st order if possible for its excellent transient response and lack of ripple at crossover point. Unfortunately there aren't many midbass units around that doesn't exhibit sharp peaks on upper midrange. I was lucky to get these old paper units. Here is one of them measured alone with whizz cone still in place, then removed, and then improvised phase plug added. Green line is where I stopped tweakings.
Attachments
That's another measurement you need to be gating properly, like Lynn is describing. Thinking about in-room sound and measuring it is all well and good, but you need to measure the direct output without reflected sound to do the sorts of things you're posting about, like finding flaws in the crossover and messing with phase plugs.
Last edited: