Reasoning with compromises and trade-offs: why there is more or less compression ratio on compression drivers? which ratio would suit your application and why? if you choose the wrong one it is very likely it has adverse effects. But, if the whole driver has better trade-offs everything accounted it doesn't matter. What matters is knowing what different compression ratios giveth and taketh to be able to choose right one in grand scheme of things 
heading of to read some myself, coffee time >
edit. mabat, good you added more content on the post. Had to think second why diaphragm size and throat size do not correlate with compression ratio, that was intuitively the first thought for me as well few minutes ago. Compression happens in the chamber. The phase plug slots do not contract to the throat but expand. Expansion starts immediately at the chamber entrance to slots and continues through the phase plug, through any conical throat section, through any waveguide attached.
heading of to read some myself, coffee time >edit. mabat, good you added more content on the post. Had to think second why diaphragm size and throat size do not correlate with compression ratio, that was intuitively the first thought for me as well few minutes ago. Compression happens in the chamber. The phase plug slots do not contract to the throat but expand. Expansion starts immediately at the chamber entrance to slots and continues through the phase plug, through any conical throat section, through any waveguide attached.
Last edited:
Me neither its not even listed on any datasheets if I remember. But, it probably has to do extending any drivers bandwidth or SPL capability, which probably comes with cost in SPL / distortion / high frequency problems etc. This is what I was after, one can expect worse (looking) highs with big driver than for smaller one, for example. If small driver is designed for extended lows there probably needs to be more volume on the chamber for xmax which brings down acoustic lowpass which mush increase slot area to get the highs back or something like that and compression ratio sails along. Otherwise we would have 0.5" throat drivers with massive diaphragms with very wide bandwidth, but in reality the nice bandwidth is always limited.
its not even listed on any datasheets if I remember. But, it probably has to do extending any drivers bandwidth or SPL capability, which probably comes with cost in SPL / distortion / high frequency problems etc. This is what I was after, one can expect worse (looking) highs with big driver than for smaller one, for example. If small driver is designed for extended lows there probably needs to be more volume on the chamber for xmax which brings down acoustic lowpass which mush increase slot area to get the highs back or something like that and compression ratio sails along. Otherwise we would have 0.5" throat drivers with massive diaphragms with very wide bandwidth, but in reality the nice bandwidth is always limited.
Last edited:
I understand compression ratio as a mere byproduct of a phase plug design. I doubt it has a major effect per se, as there seem to be things with a lot more stronger effects, like the resonances inside the compression chamber and their coupling to the rest, not mentioning the non-rigid vibrations of the diaphragm itself.
It is not surprising that the internal dimensions remain a blackbox. For the typical use case of compression drivers, it counts that they do compress and hence increase SPL, and customers might then choose their driver from i. e. SPL capabilities vs. distortion properties. I just picked up this criterion from some online reviews, I think it was even stated in a Anselm Franke review, but maybe it is a non-criterion, okay. With may laymen reasoning I just thought more compression will amplify resonances and hence distortion.
Increasing throat size is just effort trying to extend bandwidth of a tweeter. While its fun problem trying to fix issues ends of transducer bandwidth there is eventually some limit, diminishing returns. Holygrailing I often feel frustrated thinking about some loudspeaker detail and seeing the trade-offs, argh. It is less effort just to accept physics as it comes, wavelength changes, and design with the physics and not against it and be angry about it. Now focus can be directed to what is actually more important, what is more audible "issue" in system level than something else, this is the real problem. And what the actual requirements are one is trying to meet, which could be false. For example some 1" driver could well be more than enough given that real listening level at home was << 80db. If one knows limit is 80 + headroom, then its rather easy to test few drivers how good they sound with the listening level and use the best one. Perhaps they all did, no need to try and extend the bandwidth and ruin top octave for no good reason.
If there is need for low crossover tweeter then there might be some error on top octave and that's it, figure out which is more important and be happy. If there is some relatively simple way to somewhat better the performance, like the ring insert, then nice, but to get performance past that expect higher crossover point and less SPL. Its a tug of war. Tug of war with our own though processes as well, physics is what it is Why we even need low crossover for a treble? why not high? where is the best bandwidth to xo for audibility?
Conversely, one could embrace physics and start thinking sacrificing stuff for good reason. Like extra SPL capability of compression driver for better top end of dome tweeter. Or bass extension for better diffraction performance, just add another bass box to reclaim. Sacrificing stuff is making progress comfortably instead of being angry with trading-off stuff that really isn't that important. Like you wanted to hold on to something that really could go with no regrets.
Perhaps use a dome tweeter with no throat and phase plug issues at all to get nice top octave, done. Or forget about top octave all together until its place in relative order of importance is found. Perhaps ear is so sensitive to distortion or dynamics or something horn systems do better than dome tweeters that the top octave just doesn't matter that much. Perhaps the wide bandwidth sounds better even though its ragged frequency response, why? Or perhaps its all just nostalgia and all we need to do is concentrate on listening and thinking what makes into better sound instead of trying to maximize bandwidth, for example. Perhaps this all is just nerds engineering and fooling around while all the fun could be had with what ever one gets hands on? Well, what ever makes fun time Friday, jolly weekend everyone!
I often feel frustrated thinking about some loudspeaker detail and seeing the trade-offs, argh. It is less effort just to accept physics as it comes, wavelength changes, and design with the physics and not against it and be angry about it. Now focus can be directed to what is actually more important, what is more audible "issue" in system level than something else, this is the real problem. And what the actual requirements are one is trying to meet, which could be false. For example some 1" driver could well be more than enough given that real listening level at home was << 80db. If one knows limit is 80 + headroom, then its rather easy to test few drivers how good they sound with the listening level and use the best one. Perhaps they all did, no need to try and extend the bandwidth and ruin top octave for no good reason.If there is need for low crossover tweeter then there might be some error on top octave and that's it, figure out which is more important and be happy. If there is some relatively simple way to somewhat better the performance, like the ring insert, then nice, but to get performance past that expect higher crossover point and less SPL. Its a tug of war. Tug of war with our own though processes as well, physics is what it is
Why we even need low crossover for a treble? why not high? where is the best bandwidth to xo for audibility?Conversely, one could embrace physics and start thinking sacrificing stuff for good reason. Like extra SPL capability of compression driver for better top end of dome tweeter. Or bass extension for better diffraction performance, just add another bass box to reclaim. Sacrificing stuff is making progress comfortably instead of being angry with trading-off stuff that really isn't that important. Like you wanted to hold on to something that really could go with no regrets.
Perhaps use a dome tweeter with no throat and phase plug issues at all to get nice top octave, done. Or forget about top octave all together until its place in relative order of importance is found. Perhaps ear is so sensitive to distortion or dynamics or something horn systems do better than dome tweeters that the top octave just doesn't matter that much. Perhaps the wide bandwidth sounds better even though its ragged frequency response, why? Or perhaps its all just nostalgia and all we need to do is concentrate on listening and thinking what makes into better sound instead of trying to maximize bandwidth, for example. Perhaps this all is just nerds engineering and fooling around while all the fun could be had with what ever one gets hands on? Well, what ever makes fun time
Friday, jolly weekend everyone!
Last edited:
Just to keep things on track, the throat ring insert has really nothing to do with compression ratio.
1) It removes the curvature mismatch otherwise present at the junction of the original conical exit and a WG, i.e. removes this reflection, making the overall response smoother;
2) It transforms/diffracts the wavefront into a wider angle than what's possible with the original (bigger) throat;
3) It can increase the acoustic loading a small bit (not that it's really important here but people like to hear that ).
Basically, it moves all the acoustic problems closer to the diaphragm and to a smaller space, shifting them up in frequency as a result.
1) It removes the curvature mismatch otherwise present at the junction of the original conical exit and a WG, i.e. removes this reflection, making the overall response smoother;
2) It transforms/diffracts the wavefront into a wider angle than what's possible with the original (bigger) throat;
3) It can increase the acoustic loading a small bit (not that it's really important here but people like to hear that
).Basically, it moves all the acoustic problems closer to the diaphragm and to a smaller space, shifting them up in frequency as a result.
Last edited:
btw. just read a paper http://boson.physics.sc.edu/~kunchu...isalignment-of-acoustic-signals---Kunchur.pdf whose results indicate people can detect few microsecond timing difference between two transducers in the test setup, 2-3mm in distance. Test signal was around 7kHz.
It was mentioned in the paper that even though the very short delay has frequency response effects on very high frequency even people who cannot hear that high could still perceive the difference. For this reason the top octave frequency response doesn't feel very important compared to temporal stuff, both sacrificed with compression driver I think (on the top octave) because of smear in phase plug. Or, if the phase plug of one driver is better so that there isn't that much reflection on delayed sound echoing out than on some other then it could sound better even if it had ragged top octave response. We just don't hear that well frequency response up high, especially with age.
But, I was left puzzled how this could be they hear any difference of few millimeters as listening distance was multiple meters so some room effect expected. Also, what structure the drivers were attached on to? Did their output interact with each other, reflection and diffraction like physical objects should? Were they really detecting the path length difference or just detecting acoustic fingerprint of the test setup changing or something like that?
One evening I was astounded not to hear any difference otherside of the room when I was joggin delay of my tweeters with DSP, millisecond delay had about no difference on music that was playing, let alone microseconds. Didn't repeat it with near field listening, but the delay was working as I could hear difference when the delay got distinct, > 10ms it was easy to hear, below that, not much. Perhaps reflections dominated direct sound and it didnt matter? Perhaps something is broken? Perhaps ear is not sensitive to timing difference as physical setup is not changing? perhaps the system is so bad its impossible to hear difference? Perhaps the crossover around 1-2kHz just isn't very audible? most likely number of things. Fun and exciting and frustrating all at the same time And gives perspective on stuff, whats your listening distance and setup compared what you hear? Is it there is something wrong in the system or is the listening environment setup wrong for it?
Well, I'm so sceptical on all this stuff I really need to hear everything my self. If I don't hear difference, then the cheaper / easier / better looking option wins, or needs more investigation. Conversely, what ever sounds better wins no matter price or size or shape or tech, and needs more investigation
It was mentioned in the paper that even though the very short delay has frequency response effects on very high frequency even people who cannot hear that high could still perceive the difference. For this reason the top octave frequency response doesn't feel very important compared to temporal stuff, both sacrificed with compression driver I think (on the top octave) because of smear in phase plug. Or, if the phase plug of one driver is better so that there isn't that much reflection on delayed sound echoing out than on some other then it could sound better even if it had ragged top octave response. We just don't hear that well frequency response up high, especially with age.
But, I was left puzzled how this could be they hear any difference of few millimeters as listening distance was multiple meters so some room effect expected. Also, what structure the drivers were attached on to? Did their output interact with each other, reflection and diffraction like physical objects should? Were they really detecting the path length difference or just detecting acoustic fingerprint of the test setup changing or something like that?
One evening I was astounded not to hear any difference otherside of the room when I was joggin delay of my tweeters with DSP, millisecond delay had about no difference on music that was playing, let alone microseconds. Didn't repeat it with near field listening, but the delay was working as I could hear difference when the delay got distinct, > 10ms it was easy to hear, below that, not much. Perhaps reflections dominated direct sound and it didnt matter? Perhaps something is broken? Perhaps ear is not sensitive to timing difference as physical setup is not changing? perhaps the system is so bad its impossible to hear difference? Perhaps the crossover around 1-2kHz just isn't very audible? most likely number of things. Fun and exciting and frustrating all at the same time
And gives perspective on stuff, whats your listening distance and setup compared what you hear? Is it there is something wrong in the system or is the listening environment setup wrong for it?Well, I'm so sceptical on all this stuff I really need to hear everything my self. If I don't hear difference, then the cheaper / easier / better looking option wins, or needs more investigation. Conversely, what ever sounds better wins no matter price or size or shape or tech, and needs more investigation
Last edited:
That's a terrible way to design speakers. Listening tests can be so misleading as there is a strong bias to hear something, whether there or not.
If I understand mabat correctly, he does not see a sound quality compromise when reducing the aperture of the compression driver. I am a bit surprised that there should not be an unwanted effect, but my knowledge in physics is mostly absent so I tend to believe.
The ring insert must be printed with a high resolution SLA printer?
I tried to disassemble one of the CDX1-1747 today to further check on its internal construction, but sadly, one of the four Allen key headed screws kept slipping and I have to source a perfectly fitting driver now, if I ever want to be able to move the screw again. The CDX1-1747 also has a rudimentary phase plug that features three bars, maybe there is more to it to see only when opened up.
I think it would be a nice feature with a wide dispersion 120-140 degrees waveguide, if the VHF was at least this 80 degrees the simulation promises.
The ring insert must be printed with a high resolution SLA printer?
I tried to disassemble one of the CDX1-1747 today to further check on its internal construction, but sadly, one of the four Allen key headed screws kept slipping and I have to source a perfectly fitting driver now, if I ever want to be able to move the screw again. The CDX1-1747 also has a rudimentary phase plug that features three bars, maybe there is more to it to see only when opened up.
I think it would be a nice feature with a wide dispersion 120-140 degrees waveguide, if the VHF was at least this 80 degrees the simulation promises.
Hence intense investigation
You are absolutely right though about bias.edit. oops, this became quite long post. The long part is just some late night text why I would write that listening is important.
I often write about how I, as hobbyist and enthusiast trying to learn things, and hardest part is to connect concepts to perceived sound. Its almost impossible to connect concepts to perceived sound in my room, with my setup, without experience. Eventually it needs to be listened to get perspective. Example
I have very good speakers right now, best I've had, mabat waveguides and all, tuned to room, multiple prototypes already, spinorama stuff going on and confortable with it and have found multiple errors and figured them out, so some experience and some confidence. I get very good sound I've been looking for up to neighbor complaint levels, but the very good stereo image extends only up to about 2.2 meter equilateral listening triangle, doesn't quite reach practical listening spot which is a bummer. "Sound" is good on the whole house, but the sharp stereo image gets veiled a bit where I like to sit. If I take chair and put it front of sofa I can listen to good sound, or move the sofa, no problem. Its living room situation and a big family so I'd like it really be practical. Nitpicking perhaps, but its interesting hobby so nitpicking is allowed
Task is now to extend the listening triangle so that very good sound can be enjoyed on the sofa. I've figured out it is most likely just direct to reflected sound ratio that gets too low, over some threshold in hearing and detail is lost. Few more prototypes for different coverage angle are in the works. Plan is to compare these and listen how coverage angle affects this effect. Yeah I could test in different rooms but I need more prototypes anyway, to test manufacturing, ring insert and so on
Now, having spent some time on the subject, while not actually knowing what is causing it I now have some perspective to it. If someone shows a picture with speakers other side of a reflective room and says the speakers image well I know what they are talking about. If it takes very narrow coverage angle speakers listened up close to get good stereo image I know they probably haven't heard good phantom center ever and just talk about how big the mush in front is. Now I can relate their message to my situation and consider if it is relevant or not. Suddenly everyones opinions and knowledge on things that often conflict is much easier to put into perspective for the part I have some own experience.
It would be arrogant to say I know how good or bad someones system is, or know better than others or know anything, hopefully that is not the impression, but I've gotta say I know a lot better than me few years back. And its due to trying to connect visuals and text and knowledge to perceived sound and practical experience that comes with it. Or just experience, where listening is crucial part of.
With modern tools like ATH, BEM, VituixCAD, easy to access to measurement equipment, knowledge available, its not too hard to make rather good speakers acoustically without hearing single sound from it. Only problem is to find out how all the graphs I've been looking at and the concepts the system is based on sounds in my application and in my room with and practical situation, how hearing system works and how to exploit it. Connecting graphs to perceived sound, what matters and what doesn't.
For the top octave stuff above, I have no idea since I don't have that much experience on it. But my theory is there needs to be suitably wide bandwidth of good sound from any good system and if tweeter ruins top octave, how ever, its lost and the speaker is not as good as it could. Better not lose top octave, simple as that
Currently my top octave is fine compared to other things, like bass response which could be more even, or the aforementioned stereo image veil that sets in if too far away from speakers. Plan is to buy another driver and listen if it is better somehow, as I see resonance in graphs for the current one, although not sure if I hear any of it. Perhaps I do but doesn't seem to bother in any way.
Last edited:
- I've found some older experiments with Bliesma T34B, here's the code for its shape if anyone wanted to try their luck. I haven't found anything that I would be really satisfied with but probably haven't tried hard enough (also I had no optimizer at the time)...
I only hope it still works. It should be also easily possible to add various "phase equalizers" in front of the dome in the script.
It's all described in 6.5 in the User Guide: http://www.at-horns.eu/release/Ath-4.8.2-UserGuide.pdf#page=40
Code:
; Bliesma T34B
Source.Contours = {
point p1 7 0 3
point p2 0 17 0.6
point p4 -1 18 0.6
cpoint c1 -17.139 0
cpoint c2 0 18
arc p1 c1 p2 1.0
arc p2 c2 p4 0.75
arc p4 c2 WG0 0.25
}
Source.Velocity = 2I only hope it still works. It should be also easily possible to add various "phase equalizers" in front of the dome in the script.
It's all described in 6.5 in the User Guide: http://www.at-horns.eu/release/Ath-4.8.2-UserGuide.pdf#page=40
Last edited:
...Only quickly put in a ST260-like free-standing "plate" -
Now someone could try to make it a bit wider
Now someone could try to make it a bit wider
Code:
R-OSSE = {
R = 120.0
r0 = 19
a0 = 30
a = 55
k = 0.7
r = 0.3
b = 0.8
m = 0.7
q = 3
}
Source.Contours = {
point p1 7 0 3
point p2 0 17 0.6
point p4 -1 18 0.6
cpoint c1 -17.139 0
cpoint c2 0 18
arc p1 c1 p2 1.0
arc p2 c2 p4 0.75
arc p4 c2 WG0 0.25
}
Source.Velocity = 2
Mesh.AngularSegments = 8
Mesh.LengthSegments = 40
Mesh.SubdomainSlices =
Mesh.WallThickness = 5
ABEC.MeshFrequency = 32000
ABEC.NumFrequencies = 60
ABEC.SimProfile = 0
ABEC.SimType = 2
ABEC.f1 = 200
ABEC.f2 = 20000
ABEC.Polars:SPL = {
MapAngleRange = -120,120,49
Distance = 1.5 ; [m]
NormAngle = 0
}
Output.ABECProject = 1
Output.STL = 0
Report = {
Title = "ST260-T34B"
Width = 1000
Height = 1000
NormAngle = 10
;GnuplotCode = pmap-circ.gpl
MaxAnglePM = 180
MaxRadius = 350
}Yeah, I would say that if top octave performance is critical then something like that would be obvious choise over compression drivers.
It wouldn't work with very low crossover, so both would enable different kinds of systems, bigger or smaller mid.
If one needs very narrow coverage then the compression driver work better at system level, better sound overall even if the top was little bit less than ideal.
One should choose set of compromises, which ever enables best fit system for given application.
It wouldn't work with very low crossover, so both would enable different kinds of systems, bigger or smaller mid.
If one needs very narrow coverage then the compression driver work better at system level, better sound overall even if the top was little bit less than ideal.
One should choose set of compromises, which ever enables best fit system for given application.
Given the very high breakup of the Bliesma tweeter, the above would be indeed an extremely smooth solution.
Could be approximated with an OSSE profile and put into a rounded box as well.
Could be approximated with an OSSE profile and put into a rounded box as well.
Last edited: