Here is the (interpolated) DI for the 25 degree axis using the corrected power response. The blue trace shows the amount of DI contributed solely by the crossover. I'd like to shift the crossover a little to knock over the DI peak at 1k, but I'm caught between a 650Hz tweeter resonance and some cone breakup.
Question, is the DI based on a point or a plane? ie if I need to figure the SPL as the average around a vertical plane then I'd need to weigh in the cancellations around the crossover region.
Question, is the DI based on a point or a plane? ie if I need to figure the SPL as the average around a vertical plane then I'd need to weigh in the cancellations around the crossover region.
Attachments
This is precisely what I do, but its is a complex thing to get right. You need magnitude and phase data at many angles, and the magnitude and phase of the crossover into the speaker load. Then you can sum the two, but only if the field data was taken with a common point of reference, otherwise the sum won;t be correct. Even then, I have done hundreds of these kinds of simulations and it is veryt difficult to get a simulated result that is exactly the same as the actual result. It has taken me years to get this to work. The crossover is very sensitive to phase and the phase is very sensitive to the measurements.
Once you get this to work then you can get the DI from the power and polar response, but only if this all works out correctly. Otherwise the DI will not be correct.
Quite correct. What you have to do is to measure the actual sources in the enclosure that they will be used in. Then you can proceed to simulate the crossover. Trying to model the actual drivers will be very inaccurate at the upper end of the lower driver in the crossover. This will prevent any useful simulations of the crossover.
Years ago I tried to create purely simulated models of drivers to be used in simulations, but I always ended up too far from the real result when I measured that. Now I can measure the drivers and simulate the crossover and the results match almost excatly, within a dB or less. My software now only uses actual available component values not a continuum of values. This has been invaluable in crossover design since I can now do a crossover, order the exact parts that I need and know that things will end up within the tollerance that I am looking for.
I do use simulations of the LF response from bulk parameters like Fs and Q since I cannot do measurements below 200 Hz. But at these frequencies the models are fairly simple and accurate. However, I don't use the classic models of a piston in a baffle, since all of my work is done on free field speakers in boxes and the "baffle step" must be accounted for to get an accurate blending of the simulated result with the measured result.
Crossover simulations using only one axis are very simple, but very limited.
Years ago I tried to create purely simulated models of drivers to be used in simulations, but I always ended up too far from the real result when I measured that. Now I can measure the drivers and simulate the crossover and the results match almost excatly, within a dB or less. My software now only uses actual available component values not a continuum of values. This has been invaluable in crossover design since I can now do a crossover, order the exact parts that I need and know that things will end up within the tollerance that I am looking for.
I do use simulations of the LF response from bulk parameters like Fs and Q since I cannot do measurements below 200 Hz. But at these frequencies the models are fairly simple and accurate. However, I don't use the classic models of a piston in a baffle, since all of my work is done on free field speakers in boxes and the "baffle step" must be accounted for to get an accurate blending of the simulated result with the measured result.
Crossover simulations using only one axis are very simple, but very limited.
Last edited:
I've been measuring my latest build, and I've noticed that measuring from 1m throws the distance between the baffle edge radius and the 0 degree axis out by 1/4 lambda at about 4k compared to my normal listening distance. Although I don't expect much of a problem I thought I'd check in all the same. Will it be an issue?
I have a set of Summas, and I'm happy with them, but I notice that a bit of 'air' is missing in the top octave. You can see this in the measurements of all the Gedlee speakers; they drop off around 16khz.
^^ The above is a measurement of an Abbey from this thread : http://www.diyaudio.com/forums/mult...lee-summa-abbey-kit-build-35.html#post1732496
While reading Stereophile recently, I noticed that JBL is doing something a bit interesting with their K1400 speaker. They have a tweeter running from 8khz and up. Which isn't unusual. What *is* unusual is the intensity of the tweeter; it tips up to +10dB in the very last half an octave. But the power response is basically flat.
The first measurement above is the on-axis response of the K1400; the second measurement is basically the power response. More details are available in the text of the article*.
Just thought I'd post that, in the event that there are other Gedlee customers who've been a bit curious about adding a supertweeter. It might be worth experimenting with what JBL did there. You might be able to add a bit of 'air' to the top end of the Gedlee speaker if you're sensitive to that type of thing. (I know that 80% of the people out there wouldn't notice 20khz, but I seem to be sensitive to it. On headphones I can hear it quite distinctly.)
I'm guessing that JBL is able to use a very high intensity for that last octave because the room is so absorptive at these frequencies. And we're so insensitive to them. IE, it really doesn't take a lot to absorb 18khz, so to get flat power response in that octave you need unusually high intensity.
* measurements here : JBL Synthesis 1400 Array BG loudspeaker Measurements | Stereophile.com
^^ The above is a measurement of an Abbey from this thread : http://www.diyaudio.com/forums/mult...lee-summa-abbey-kit-build-35.html#post1732496
While reading Stereophile recently, I noticed that JBL is doing something a bit interesting with their K1400 speaker. They have a tweeter running from 8khz and up. Which isn't unusual. What *is* unusual is the intensity of the tweeter; it tips up to +10dB in the very last half an octave. But the power response is basically flat.
The first measurement above is the on-axis response of the K1400; the second measurement is basically the power response. More details are available in the text of the article*.
Just thought I'd post that, in the event that there are other Gedlee customers who've been a bit curious about adding a supertweeter. It might be worth experimenting with what JBL did there. You might be able to add a bit of 'air' to the top end of the Gedlee speaker if you're sensitive to that type of thing. (I know that 80% of the people out there wouldn't notice 20khz, but I seem to be sensitive to it. On headphones I can hear it quite distinctly.)
I'm guessing that JBL is able to use a very high intensity for that last octave because the room is so absorptive at these frequencies. And we're so insensitive to them. IE, it really doesn't take a lot to absorb 18khz, so to get flat power response in that octave you need unusually high intensity.
* measurements here : JBL Synthesis 1400 Array BG loudspeaker Measurements | Stereophile.com
Last edited:
John
First the drop in those older plots is a result of an anti-aliasing filter in the measurements and it not accurate, but yes, the response above 18 kHz (where there is actually a large peak)drops just like a DE250 does on any other device. But the real question is if the downside of using a super tweeter (crossover, more lobing etc.) will be the cure being worse than the disease. IMO it would be.
That the newer DE250's do have more high end than the older ones is also true.
First the drop in those older plots is a result of an anti-aliasing filter in the measurements and it not accurate, but yes, the response above 18 kHz (where there is actually a large peak)drops just like a DE250 does on any other device. But the real question is if the downside of using a super tweeter (crossover, more lobing etc.) will be the cure being worse than the disease. IMO it would be.
That the newer DE250's do have more high end than the older ones is also true.
The funny thing is i know people for which is OS profile too hot at HF. I mean in sense that powerresponse dont fall up in highs. I think that everybody here agree that directivity must somehow rise with frequency and completely flat powerresponse will always sound too bright. Even JBL/harman agree and they even make falling to HF on-axis response as general recommended target. Its just a way that music is mastered at studios - they dont have flat power response monitors - such a monitor doesnt exist.
I make a LOT of experiment and completely agree with mr. Geddes that you cannot properly integrate supertweeter with WG. It is just not possible. /why do think JBL make new M2 without STW?/ If you make sudden changes in PR it is always sounds weird. And when you manage it without changes it will sound exact like without STW. If you use bigger throat or different profile /like i do/ i can understand you concern, but 1inch OS WG is about as wide as it get at HF - maybe proportionaly /mid to HF/ wider than 1 inch dome direct radiator.
You can make that 10db rise experiment very easy for your self just try EQ some music in PC. If you dont play from PC preprocess some files and burn it on CD.
Honesty i even think that everything 15khz and up just doesnt matter. And hear 19khz at nominal level. When you experiment with such a things you must use very steep filters than dont affect response below. For example when you hear 2th order low pass at 15khz you just hear changed response below 15khz. You must use at least 10th order. - try it.
Even then the response as shown it looks more like error for me. It will be quite hard to obtain such a response with passive crossover and compression drivers.
I make a LOT of experiment and completely agree with mr. Geddes that you cannot properly integrate supertweeter with WG. It is just not possible. /why do think JBL make new M2 without STW?/ If you make sudden changes in PR it is always sounds weird. And when you manage it without changes it will sound exact like without STW. If you use bigger throat or different profile /like i do/ i can understand you concern, but 1inch OS WG is about as wide as it get at HF - maybe proportionaly /mid to HF/ wider than 1 inch dome direct radiator.
You can make that 10db rise experiment very easy for your self just try EQ some music in PC. If you dont play from PC preprocess some files and burn it on CD.
Honesty i even think that everything 15khz and up just doesnt matter. And hear 19khz at nominal level. When you experiment with such a things you must use very steep filters than dont affect response below. For example when you hear 2th order low pass at 15khz you just hear changed response below 15khz. You must use at least 10th order. - try it.
Even then the response as shown it looks more like error for me. It will be quite hard to obtain such a response with passive crossover and compression drivers.
Last edited:
I believe that some of the "air" effect is the traditional beaming by a direct radiator and its poor HF power response. This will certainly sound different than a flat (or slightly falling) power response in a constant directivity waveguide. But which is correct? Well the answer to that is obvious I think, but there might be a "preference" or "expectation" for the beaming sound, but it is not "correct".
I too believe that > 15 kHz is a total waste of time and that these frequencies are not resent in nature in general because of air absorption. People quote their ability o "hear" them, but I suspect that they just "detect" something. These frequencies cannot be a major component of music.
I too believe that > 15 kHz is a total waste of time and that these frequencies are not resent in nature in general because of air absorption. People quote their ability o "hear" them, but I suspect that they just "detect" something. These frequencies cannot be a major component of music.
I was playing around with an iOS app and the Earpods, the app would allow you to generate frequencies from 20~20KHz, one can determine what frequency they can hear by sliding down the frequency. Well, something interesting happened:
I turned the volume to max, starting at 20KHz and started to slide down the frequency, by wife was sitting about 2 Meters from me, and suddenly turned towards me wondering what was going on, when I got to 15KHz, she was annoyed. Seeing her facial expression, I turned the volume down. I could not hear anything, but my wife was obviously not happy with it present even though she was 2 meters away from me, and I was listening through the Earpods.
I turned the volume to max, starting at 20KHz and started to slide down the frequency, by wife was sitting about 2 Meters from me, and suddenly turned towards me wondering what was going on, when I got to 15KHz, she was annoyed. Seeing her facial expression, I turned the volume down. I could not hear anything, but my wife was obviously not happy with it present even though she was 2 meters away from me, and I was listening through the Earpods.
Earl,
As an adult, I never heard above 18KHz, but now I certainly miss the upper 3000 Hz that I no longer can hear.
While it is true that 20KHz is attenuated by 9 dB at 50 feet due to air absorption, the majority of pop music sources are recorded at distances of inches, and listened back at distances where HF attenuation is slight.
Granted, for old men like us who barely can hear 15KHz, what goes on above that is of little consequence, but to dismiss the upper harmonics that exist in music as "a total waste of time" for those that do have hearing that extends to 20 kHz or beyond is akin to a blind person saying that sight is unimportant.
Art
A cardioid woofer/mid with properly designed waveguide/tweeter section would qualify as having flat power response. And I can assure you it does not sound bright or coloured in any way (although a dipole does, but that's because it has other inherent problems).