Amazing read! Take the trouble to go through it, it is not that hard to open and worthwhile.
I wish I had the time and space to play like that! At least we don't have termites eating our drivers here in Denver...
My take on your final questions of how it can sound as good as it does with all the problems it has. It has to do with processing. Our brains are amazing and what we "realize" in terms of sound is what our brain can do with the information fed it's ears. There are errors in timing, interferences, omissions, additions and so on, but there are nuances that determine if these are important to our processing or not. Just because it looks bad on the computer screen, doesn't mean that you aren't feeding your brain with all the information it needs to determine what to "hear" and what not to.
I wish I had the time and space to play like that! At least we don't have termites eating our drivers here in Denver...
My take on your final questions of how it can sound as good as it does with all the problems it has. It has to do with processing. Our brains are amazing and what we "realize" in terms of sound is what our brain can do with the information fed it's ears. There are errors in timing, interferences, omissions, additions and so on, but there are nuances that determine if these are important to our processing or not. Just because it looks bad on the computer screen, doesn't mean that you aren't feeding your brain with all the information it needs to determine what to "hear" and what not to.
You are right jrenkin, the most appealing is the conclusion of this experience: how comes a normal subject 
ill can perceive a good imaging through this apparatus?
Of course, statistically it’s not valid but I do my best to be honest. A DSP is a good ally for that, thanks to it’s endless memory: only a few set up produce a nice imaging . The comparison is immediate and fool proof. If I had to work 3 hours, iron solder in hand and then claiming for some improvement, here, I would’nt even trust myself.
These spatial properties don’t come from the room. Recently, I’ve put the system in a classic configuration (takes 2 hours only) that produces almost square waves and that’s average in off axis dispersion. This has some qualities, but it’s not playing music, just HiFi. Before I was biased on it, but now, it’s really kicked out.
I am pretty sure that these properties come mainly from the darn semi circular distribution. The first day I tried it I had a schock so much it was promising. But why ?
The original paper of the OSD publication (10 yrs old !) is a mathematical demonstration, without any reference to the localization processes, strange, as this is supposed to be the target.
To put more trouble in the minds, the pink noise localization that I describe (e.g. a line between the speakers) is classified by Toole as “center image totally destroyed” (confusers and confusion), and you know that we can respect Toole. Anyway, this case must be an exception to the rule.
I wish we could measure spaciousness, imaging and all this stuff, like this we would definitively escape from these psycho-acoustic-feeling-subjective reviews.
For el’Ol, you could say “a safe alternative”, because I’ve been destroying 3 membranes on one year.
About the reduction of size, neodynium could be a better choice (my 1” is already neo but the 2” are 17 cm at the magnet, far too big), the 92 dB efficiency is not an issue, but where do you see the specifications ?
Lab assistant? Maybe more...was the Pacific Connexion.
ill can perceive a good imaging through this apparatus? Of course, statistically it’s not valid but I do my best to be honest. A DSP is a good ally for that, thanks to it’s endless memory: only a few set up produce a nice imaging . The comparison is immediate and fool proof. If I had to work 3 hours, iron solder in hand and then claiming for some improvement, here, I would’nt even trust myself.
These spatial properties don’t come from the room. Recently, I’ve put the system in a classic configuration (takes 2 hours only) that produces almost square waves and that’s average in off axis dispersion. This has some qualities, but it’s not playing music, just HiFi. Before I was biased on it, but now, it’s really kicked out.
I am pretty sure that these properties come mainly from the darn semi circular distribution. The first day I tried it I had a schock so much it was promising. But why ?
The original paper of the OSD publication (10 yrs old !) is a mathematical demonstration, without any reference to the localization processes, strange, as this is supposed to be the target.
To put more trouble in the minds, the pink noise localization that I describe (e.g. a line between the speakers) is classified by Toole as “center image totally destroyed” (confusers and confusion), and you know that we can respect Toole. Anyway, this case must be an exception to the rule.
I wish we could measure spaciousness, imaging and all this stuff, like this we would definitively escape from these psycho-acoustic-feeling-subjective reviews.
For el’Ol, you could say “a safe alternative”, because I’ve been destroying 3 membranes on one year.
About the reduction of size, neodynium could be a better choice (my 1” is already neo but the 2” are 17 cm at the magnet, far too big), the 92 dB efficiency is not an issue, but where do you see the specifications ?
Lab assistant? Maybe more...was the Pacific Connexion.
Last edited:
Specs can only bee seen in the german version of the page.
IT Speaker Shop - MD9550 | Gradient Select Tief-Mitteltonlautsprecher :: Gradient Select :: Lautsprecher ::
IT Speaker Shop - MD9550 | Gradient Select Tief-Mitteltonlautsprecher :: Gradient Select :: Lautsprecher ::
Hello,
Link works for me, I must be closer from Germany.
Generally speaking, making a HF dipole is more complicated than removing the back chamber of a compression driver. Revisit the dedicated Rudolf's thread to take the dimension of the debate.
For making LQ, if the priority is the omni dispersion, it's not very difficult. (If the priority is dispersion + phase, here I failed). It means that the drivers I use are not the only formula. Maybe small wide range with beautifully openened baskets ?
OTOH, the Gradient can be open back, but isn't it just a resonating vent? (cf dynaudio D54 that was sealed). Without phase plug, it's almost impossible to make a working WG and the back wave must suffer.
If we stay in the dipole tweeter thread, posts 96 and 99 have to be read twice...
Link works for me, I must be closer from Germany.
Generally speaking, making a HF dipole is more complicated than removing the back chamber of a compression driver. Revisit the dedicated Rudolf's thread to take the dimension of the debate.
For making LQ, if the priority is the omni dispersion, it's not very difficult. (If the priority is dispersion + phase, here I failed). It means that the drivers I use are not the only formula. Maybe small wide range with beautifully openened baskets ?
OTOH, the Gradient can be open back, but isn't it just a resonating vent? (cf dynaudio D54 that was sealed). Without phase plug, it's almost impossible to make a working WG and the back wave must suffer.
If we stay in the dipole tweeter thread, posts 96 and 99 have to be read twice...
Of course this driver is not optimized for dipolar radiation pattern, it's open back because this allows more installation flexibility in a car than a driver with large back chamber. But the dynamical dipole speaker trend is not being taken very seriously by the industry and one has to take what one gets. An alternative could be the new BG Neo 8 which has improved output compared to older versions.
bravissimo! that is VERY impressive! though not very diy-cloneable
there are two unconventional approaches that You might also find interesting:
Stereolith - stereolith.com
http://www.audiopro.com/media.php?id_file=527
can You please do us a favour and try the central midtweeter section in stereolith-like configuration instead of two dipoles?
just put them in a kind of a box - I wonder whether there would be any difference and if so how different would it be
and can You please tell us what are the crossover points of Your OSD array?
best regards!
graaf
The two face to face HF dipoles form a linear quadrupole in the center, but not entirely so, only for mono signal.
So, I guess this combo would somewhat enhance the common (mono) portion in the stereo signal in the lateral (left-right) direction - that's where the main lobe of the LQ is pointing. While the listening position is in the 'cancellation lobe' of this LQ, so, less mono signal is coming over here - is this some kind of acoustically cross talk cancellation?
About the xover setting, I remember there's a chapter of "Filter" somewhere near the end of the paper....
So, I guess this combo would somewhat enhance the common (mono) portion in the stereo signal in the lateral (left-right) direction - that's where the main lobe of the LQ is pointing. While the listening position is in the 'cancellation lobe' of this LQ, so, less mono signal is coming over here - is this some kind of acoustically cross talk cancellation?
About the xover setting, I remember there's a chapter of "Filter" somewhere near the end of the paper....
perhaps but I doubt - something different must be going on because of the time shift and resulting phase shifts
yes, probably
CLS;2422922 About the xover setting said:yes, thank You
so the points are around 180, 1200 and 5200 Hz
looks like most of the music is reproduced by omni LQ section at 32 degrees
best,
graaf
Hello,
If I wear a hat I would lift it for your efforts!
This is not a loudspeaker but a psychoacoustic stimulator! It does many things right.
The problem is that stereo was invented before the psychoacoustics research really took a fly. Now we have stereo systems. But the other one not.
Don't worry, as usually the general acceptance follows 10-20 years after the leading edge
- Elias
If I wear a hat I would lift it for your efforts!
This is not a loudspeaker but a psychoacoustic stimulator! It does many things right.
The problem is that stereo was invented before the psychoacoustics research really took a fly. Now we have stereo systems. But the other one not.
Don't worry, as usually the general acceptance follows 10-20 years after the leading edge
- Elias
Hello guys,
Graaf, it’s a long time that I run after the wiring of the stereolith. I’ve only seen hype about it.
The day I have a good info, I’ll do the experiment, maybe with Supravox 215 that I have here, (because I would’nt be seen as a grooty sound maker. ROTFL).
OK, serious time now: I was not thinking of it, but you are right, stereolith and my tweeter thing must have something in common: a frontal and centered source that can give lateralized clues.
In my case, no witness except me, but a lot for stereolith.
So, it’s not a dream, and I would be very interested by cloning a stereolith.
For the way the center stereo LQ works, I have only assumptions, nothing strongly demonstrated.
The following constatations only are safe:
Following the frequency range allowed by the filter, there is a transfer of “clues making” from the HF center combo to the mids. You have seen on the filter (resulting curves on axis only) of my doc that the 2” are finally high passed very high (I did it for other reasons, narrowing the soundstage, but the fact is…).
So, it’s very possible that the mids have finally a great responsibility in lateralization clues, can we say “as in stereolith” ?ah ah…
OTOH, we have to integrate here that between 1500 and 3000, all our hearing localization systems get confused.
Complicated isn’t it?
I don’t feel comfortable to speak of CTC. In the best case, maybe a reduction only.
Last point, when the promoters of speakers like Audiopro claim for a very wide sweet spot, here it’s almost a monoseat optimal place. The soundstage is not so much different than the production of a conventional stereo set. No wide angle view here.
Graaf, it’s a long time that I run after the wiring of the stereolith. I’ve only seen hype about it.
The day I have a good info, I’ll do the experiment, maybe with Supravox 215 that I have here, (because I would’nt be seen as a grooty sound maker. ROTFL).
OK, serious time now: I was not thinking of it, but you are right, stereolith and my tweeter thing must have something in common: a frontal and centered source that can give lateralized clues.
In my case, no witness except me, but a lot for stereolith.
So, it’s not a dream, and I would be very interested by cloning a stereolith.
For the way the center stereo LQ works, I have only assumptions, nothing strongly demonstrated.
The following constatations only are safe:
- · Side walls main reflections arrive at the listener after 15 ms so the Hass effect locks the listener to the “direct source”, could it be 10 dB lower
- · At 5 meters, the listener is not anymore in the recessive positive central lobe but in the lateral negative lobes, more or less confused in a quasi spheric wave
Following the frequency range allowed by the filter, there is a transfer of “clues making” from the HF center combo to the mids. You have seen on the filter (resulting curves on axis only) of my doc that the 2” are finally high passed very high (I did it for other reasons, narrowing the soundstage, but the fact is…).
So, it’s very possible that the mids have finally a great responsibility in lateralization clues, can we say “as in stereolith” ?ah ah…
OTOH, we have to integrate here that between 1500 and 3000, all our hearing localization systems get confused.
Complicated isn’t it?
I don’t feel comfortable to speak of CTC. In the best case, maybe a reduction only.
Last point, when the promoters of speakers like Audiopro claim for a very wide sweet spot, here it’s almost a monoseat optimal place. The soundstage is not so much different than the production of a conventional stereo set. No wide angle view here.
So...
OSD looks like W........M...TT...M........W, and our hearing is not so good at pointing out the direction of mid-low frequency sound. So how about a big mono horizontal WMTMW ?
Or, slow it down a little, how about only one mono "T" in the center? What would be the difference compared to 2 very close "TT"?
OSD looks like W........M...TT...M........W, and our hearing is not so good at pointing out the direction of mid-low frequency sound. So how about a big mono horizontal WMTMW ?
Or, slow it down a little, how about only one mono "T" in the center? What would be the difference compared to 2 very close "TT"?
Hi CLS,
You just have been reinventing the commercial append of OSD: Opsodis.
It's exactly what you said, horizontal sound bar, and it's target is the TV market, where the money is.
In the same category there is the Polk's SDA and other Yamaha products.
One mono T, maybe, but as my center LQ is not perfect (directivity peak on axis),
it's possible that some of it's "stereo properties" come from this flaw.
There are always hidden factors when the approach is not scientific but empiric.
You just have been reinventing the commercial append of OSD: Opsodis.
It's exactly what you said, horizontal sound bar, and it's target is the TV market, where the money is.
In the same category there is the Polk's SDA and other Yamaha products.
One mono T, maybe, but as my center LQ is not perfect (directivity peak on axis),
it's possible that some of it's "stereo properties" come from this flaw.
There are always hidden factors when the approach is not scientific but empiric.
The first trail with 'quasi'-omni tweeters
The word "trial" is almost always coming with "error", so very likely, this is it.
I admire the concept of OSD very much, and I finally got some spare time to try it (well, sort of... ). Trouble is, for practical reason, my existing system is located very near the wall. This brings me all kinds of problems...
Bringing the tweeters of both channels to a very close distance, I guess the directivity should be very very wide, omni is preferable. Why? I can't explain, imagining 2 narrow beams shooting at me right from the front makes me uncomfortable. It's all by instinct.
How can I get a wide spreading HF from my existing drivers? Oh well, this must be laughable, but it works to some extent. I just put 2 tweeter horns (Eminence APT) face to face, and 'hopefully' the sound between them would be 'squeezed' and escaping side way, thus forming a radial radiation.
I know it's far from ideal. This is just the easiest way to get a wide spreading HF from the stuff on hand. There must be a lot of lobing problems in every directions, but insteretingly, I can't detect any interference lobes by ears. Moving laterally, the HF sounds stay consistent. So at least it's pretty omni in the horizontal plane. And as expected, moving up and down beyond a certain range, the HF sounds collapse. I'm OK with this.
Next, how can I use this against the wall? By the wave simulation, putting a point source near a boudary produces a lot of interference lobes. So this must be very bad. Again, it's very interesting to find it's not entirely so in reality...
I tried reducing the rear refections by foam pads. Again, it's funny, it seems not much differece to my ears in lobing or smearing. Without absorption, it sounds slightly brighter overall.
The major 'loss' is the sensitivity of the tweeters. At first I separated the tweeters by 2.5cm (the distance between the felt pads around the small horns). This is probably too close, and it sounds somewhat suffocating. Later I added the distance to 3.5cm, then it sounded better. Nevertheless I still needed to re-EQ the system to get a reasonably flat response. Compared to the previous setting (HF horns pointing at me directly), I added 4.5dB above 6kHz, and supress mid to higher mid (under 5kHz) also by 4.5dB. (The LF was also turned down, but with analog volume pot, so no numbers) So it's total 9dB loss of subjective loudness at the listening position.
That's all for the tweeter only. What about the OSD? Before a major re-layout, now I can only try moving the tweeters close together (to 30cm c-t-c). The mid horns and woofers are still at their original locations. The woofers are 3.1m apart, and it's slightly further between the recessed mid horns. With listening position at the center of couch, they almost form a perfect equilateral triangle.
The crossover point between tweeter and mid is 3kHz or so. This should be very 'scary' for most audio hobbyists to separate tweeter and mid by a distance of 1.7m at such xover frequency. Ha ha... Let's just try it. How bad can it be? Explosion? So why worry?
OK, how does this smattering OSD sound? It's a very intereting presentation. The sweet spot is almost everywhere in the room. The tonal balance stays more stable across the room than my previous large toe-in (ordinary 3way) setup. The fiddled quasi-omni tweeter works pretty well in this regard.
And I found it's pretty recording-dependent. On some recordings, the sound stages seem jam-packed with the images spreading only a little wider than mono. OTOH, on some other recordings, it sounds very spacious, images spread everywhere and are in no relation with the close-located tweeters. But I haven't experienced sounds coming outside of the mid horns, yet. It's very interesting that the HF 'sparkle' is also very recording-dependent. On some, by memory, I feel the leading edges of HF seem softened or even muffled, so they sound somewhat darker. But on some others, there's no such 'illness' effect.
I also found the best listing position should be further back by about a meter or so. Overall the sounds are blent better at the position behind my desk which is behind the couch. Maybe the tweeter and mid are indeed too far away to each other.
Now this is only a first small step into the OSD concept. It's far from ideal but very interesting and also very listenable. I have no problem living with this for some time.
Next step will be quite large. I have to relayout the whole system, bring the mids closer and push the woofers further apart. Or maybe I'll dump the mid horn and make it cardioid. Xover is another big topic and need reconsidering as a whole... I have very little spare time and build things slowly, so all those won't happen very soon.
It's already very long. Any comments are welcome.
The word "trial" is almost always coming with "error", so very likely, this is it.
I admire the concept of OSD very much, and I finally got some spare time to try it (well, sort of... ). Trouble is, for practical reason, my existing system is located very near the wall. This brings me all kinds of problems...
Bringing the tweeters of both channels to a very close distance, I guess the directivity should be very very wide, omni is preferable. Why? I can't explain, imagining 2 narrow beams shooting at me right from the front makes me uncomfortable. It's all by instinct.
How can I get a wide spreading HF from my existing drivers? Oh well, this must be laughable, but it works to some extent. I just put 2 tweeter horns (Eminence APT) face to face, and 'hopefully' the sound between them would be 'squeezed' and escaping side way, thus forming a radial radiation.
I know it's far from ideal. This is just the easiest way to get a wide spreading HF from the stuff on hand. There must be a lot of lobing problems in every directions, but insteretingly, I can't detect any interference lobes by ears. Moving laterally, the HF sounds stay consistent. So at least it's pretty omni in the horizontal plane. And as expected, moving up and down beyond a certain range, the HF sounds collapse. I'm OK with this.
Next, how can I use this against the wall? By the wave simulation, putting a point source near a boudary produces a lot of interference lobes. So this must be very bad. Again, it's very interesting to find it's not entirely so in reality...
I tried reducing the rear refections by foam pads. Again, it's funny, it seems not much differece to my ears in lobing or smearing. Without absorption, it sounds slightly brighter overall.
The major 'loss' is the sensitivity of the tweeters. At first I separated the tweeters by 2.5cm (the distance between the felt pads around the small horns). This is probably too close, and it sounds somewhat suffocating. Later I added the distance to 3.5cm, then it sounded better. Nevertheless I still needed to re-EQ the system to get a reasonably flat response. Compared to the previous setting (HF horns pointing at me directly), I added 4.5dB above 6kHz, and supress mid to higher mid (under 5kHz) also by 4.5dB. (The LF was also turned down, but with analog volume pot, so no numbers) So it's total 9dB loss of subjective loudness at the listening position.
That's all for the tweeter only. What about the OSD? Before a major re-layout, now I can only try moving the tweeters close together (to 30cm c-t-c). The mid horns and woofers are still at their original locations. The woofers are 3.1m apart, and it's slightly further between the recessed mid horns. With listening position at the center of couch, they almost form a perfect equilateral triangle.
The crossover point between tweeter and mid is 3kHz or so. This should be very 'scary' for most audio hobbyists to separate tweeter and mid by a distance of 1.7m at such xover frequency. Ha ha... Let's just try it. How bad can it be? Explosion? So why worry?
OK, how does this smattering OSD sound? It's a very intereting presentation. The sweet spot is almost everywhere in the room. The tonal balance stays more stable across the room than my previous large toe-in (ordinary 3way) setup. The fiddled quasi-omni tweeter works pretty well in this regard.
And I found it's pretty recording-dependent. On some recordings, the sound stages seem jam-packed with the images spreading only a little wider than mono. OTOH, on some other recordings, it sounds very spacious, images spread everywhere and are in no relation with the close-located tweeters. But I haven't experienced sounds coming outside of the mid horns, yet. It's very interesting that the HF 'sparkle' is also very recording-dependent. On some, by memory, I feel the leading edges of HF seem softened or even muffled, so they sound somewhat darker. But on some others, there's no such 'illness' effect.
I also found the best listing position should be further back by about a meter or so. Overall the sounds are blent better at the position behind my desk which is behind the couch. Maybe the tweeter and mid are indeed too far away to each other.
Now this is only a first small step into the OSD concept. It's far from ideal but very interesting and also very listenable. I have no problem living with this for some time.
Next step will be quite large. I have to relayout the whole system, bring the mids closer and push the woofers further apart. Or maybe I'll dump the mid horn and make it cardioid. Xover is another big topic and need reconsidering as a whole... I have very little spare time and build things slowly, so all those won't happen very soon.
It's already very long. Any comments are welcome.
Attachments
Uh....
OK, the odd looking and probably ridiculous tweeters wouldn't be attractive. Previously, my system was odd enough, so more oddness doesn't matter for me anyway. But this is not just about oddness.
No measurement charts? I know I know, it's almost against the rule here :lol: It's just because now I have limited access to a proper set of instrument and software. I'd like to encourge any of you, who has a system with movable units, to give this a try. Just arrange the units more or less in the fashion of: W....M..T.T..M....W laterally. Simple enough (for the first step...).
And give it a good listen, or measurement You'll find something very interesting. And maybe, this will be a beginning of a revolution.
OK, the odd looking and probably ridiculous tweeters wouldn't be attractive. Previously, my system was odd enough, so more oddness doesn't matter for me anyway. But this is not just about oddness.
No measurement charts? I know I know, it's almost against the rule here :lol: It's just because now I have limited access to a proper set of instrument and software. I'd like to encourge any of you, who has a system with movable units, to give this a try. Just arrange the units more or less in the fashion of: W....M..T.T..M....W laterally. Simple enough (for the first step...).
And give it a good listen, or measurement
You'll find something very interesting. And maybe, this will be a beginning of a revolution.- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.