The "length" should be the axial length (depth) of the open driver, it will be on the order of centimeters.
Sorry, I made a typo, slot depth (from entry to out) = 3.5 cm
Sloth width ~1 mm (seems to be a little over 1 mm, maybe 1.1 mm, I have no way to measure it more precisely)
Oleg, the total area of the air channels can't be 64.5 sq mm. A single 1mm ring in the middle of a 4" diaphragm is around 150 sq mm.
The widths of the individual rings also seem to vary.
- I will try to estimate it from the photo, which seems pretty good for this, in fact
The widths of the individual rings also seem to vary.
- I will try to estimate it from the photo, which seems pretty good for this, in fact
Last edited:
Sorry, I can be obtuse and I can make stupid mistakesOleg, the total area of the air channels can't be 64.5 sq mm. A single 1mm ring in the middle of a 4" diaphragm is around 150 sq mm.
The widths of the individual rings also seem to vary.
The sum length of all the slots is ~64,5 cm (not mm of course), and ~1,1 mm with, so = ~709 sq mm, now it looks real?
I need to make myself another cup of coffee.
700 would be much more real, thanks. The output is around 1000, so there would be some expansion...
The angle of the extension in EXAR 400 is 6° (total). This may be actually better matched than a straight tube, even though the exit wavefront is supposed to be just flat. This probably deserves some further examination. For these numbers it would correspond to roughly a 10° cone.
The angle of the extension in EXAR 400 is 6° (total). This may be actually better matched than a straight tube, even though the exit wavefront is supposed to be just flat. This probably deserves some further examination. For these numbers it would correspond to roughly a 10° cone.
Last edited:
Looks like a good combination. The problem with such a measurement is that it doesn't necessarily reflect the actual source directivity and can be misleading, but of course it gives you an idea about the basic performance, which is so far very impressive indeed.
Would love to hear some big band
Would love to hear some big band
Last edited:
Regarding phase plugs, it's all pretty unintuitive stuff.
For example, the black curve is the area expansion profile of a single phase plug channel, if defined as shown. It actually starts decreasing towards the end of the phase plug. Overall, a straight-wall slot seems suboptimal.
https://www.desmos.com/calculator/yirxbis60w
For example, the black curve is the area expansion profile of a single phase plug channel, if defined as shown. It actually starts decreasing towards the end of the phase plug. Overall, a straight-wall slot seems suboptimal.
https://www.desmos.com/calculator/yirxbis60w
Last edited:
Now I wonder whether the area expansion has anything to do with these (numerically optimized) results:
https://www.researchgate.net/public...s_for_Acoustic_Elements_in_Loudspeaker_Design
- The algorithm used found some curved channel shapes, but it's not clear why exactly it is better that way.
https://www.researchgate.net/public...s_for_Acoustic_Elements_in_Loudspeaker_Design
- The algorithm used found some curved channel shapes, but it's not clear why exactly it is better that way.
Last edited:
Yes, the black curve shows the cross section area as a function of axial length of the channel.
(Actually, it shows 10x smaller value (mm^2), so it fits in the plot.)
Hard to predict by intuition, right?
I suspect that the phase plug in Fig.13 above will be better also in this regard.
(Actually, it shows 10x smaller value (mm^2), so it fits in the plot.)
Hard to predict by intuition, right?
I suspect that the phase plug in Fig.13 above will be better also in this regard.
Last edited:
Maybe not all of them...Overall, a straight-wall slot seems suboptimal.
https://www.desmos.com/calculator/gaejztozmt
So is that optimal? I have my doubts - the rate still decreases, to become increasing again in the horn.
The issue is that the individual channels in such a phase plug will likely be different, even if the entry and exit areas are in the same ratio. How much is this a problem is not clear to me, but I supect we could do better. Another design tool is needed...
The issue is that the individual channels in such a phase plug will likely be different, even if the entry and exit areas are in the same ratio. How much is this a problem is not clear to me, but I supect we could do better. Another design tool is needed...
Last edited:
If I remember correctly I gave you some pointers on how to do this correctly a couple of years ago in this thread
I think with all the knowledge you have developed so far regarding running optimization with ABEC it should be quite doable to design a great working 4-slit phaseplug with lowish compression ratio (IE 1:4) and a nice wide HF opening angle
I think with all the knowledge you have developed so far regarding running optimization with ABEC it should be quite doable to design a great working 4-slit phaseplug with lowish compression ratio (IE 1:4) and a nice wide HF opening angle
Hi Kees, I don't think we ever discussed the expansion rates of the individual channels. From what I remember, we discussed "only" the positioning of their entrances in a compression chamber. This is the unintuitive thing - the channel cross section area changes significantly also with its distance from the axis, even if the channel itself (in a 2D drawing) seems to be only gradually expanding - in fact it's not even in the most simple case.
But you're right
But you're right
Last edited:
It would be best if phase plug was avoided altogether, see this paper by Gunnes, which briefly touches compression driver phase plugs: https://www.fulcrum-acoustic.com/wp...dspeaker-transient-response-with-dsp-2005.pdf
Or, if there must be a phase plug, I think the inputs on diaphragm side are dictated by stuff in the paper, at least, so I'm not sure how many compression drivers could be upgraded just by swapping better phaseplug, because the whole structure affects. Better measuring drivers have everything well together, while worse measuring ones the problem could be beyond phaseplug swap. Study with the channels might give insight what kind of drivers work better though. Although, it's likely already obvious from measurements, and likely generalizable from there.
Example: This stuff seems to mostly affect per wavelength, as in everything acoustic, so the smaller the diaphragm and phase plug and throat and all the parts are physically, the higher up in frequency any anomaly shows up. As clear from Gunnes paper, there inevitably is some anomaly, so smaller devices work better for highs.
Logically interest is on some bigger devices then, which extend lows with cost of compromized highs. But this is again the same dilemma as with anything, physical size stays static while wavelength does not, so there is some limited good bandwidth for any ideal physical device. So, like with waveguides perhaps the usable bandwidth can be maximized with careful optimization. But, one would likely need to test multiple drivers, which have everything else aligned with the goal and only have a bad phaseplug. For example, if the diaphragm breaks up bad no matter what the phase plug performance doesn't improve? So, I'm not sure how much there is possibility to mod existing drivers with better phase plug, but there is value in the study and information so, please go ahead
Or, if there must be a phase plug, I think the inputs on diaphragm side are dictated by stuff in the paper, at least, so I'm not sure how many compression drivers could be upgraded just by swapping better phaseplug, because the whole structure affects. Better measuring drivers have everything well together, while worse measuring ones the problem could be beyond phaseplug swap. Study with the channels might give insight what kind of drivers work better though. Although, it's likely already obvious from measurements, and likely generalizable from there.
Example: This stuff seems to mostly affect per wavelength, as in everything acoustic, so the smaller the diaphragm and phase plug and throat and all the parts are physically, the higher up in frequency any anomaly shows up. As clear from Gunnes paper, there inevitably is some anomaly, so smaller devices work better for highs.
Logically interest is on some bigger devices then, which extend lows with cost of compromized highs. But this is again the same dilemma as with anything, physical size stays static while wavelength does not, so there is some limited good bandwidth for any ideal physical device. So, like with waveguides perhaps the usable bandwidth can be maximized with careful optimization. But, one would likely need to test multiple drivers, which have everything else aligned with the goal and only have a bad phaseplug. For example, if the diaphragm breaks up bad no matter what the phase plug performance doesn't improve? So, I'm not sure how much there is possibility to mod existing drivers with better phase plug, but there is value in the study and information so, please go ahead
Ah , I thought that I also explained abou tthe expansion rate matching.
What I have done in the past which works well;
-postion the inlets as explained before and definde the openings as for the desired compression ratio
-choose a reasonable channel length (shorter=better for 2nd harmonic wave steepening distortion, but make sure not too make sharp bends)
-create the exit wavefront shape by positioning the channel outputs
-make every channel equal in lenght by curving the channels
now for the channel shape optimazation;
- terminate each channel closed and fully absorbtive at the exit in ABEC
-drive each channel seperately
-probe impedance at inlet and freq response at outlet and adjust the thickness (for instance in 5mm length steps ftom channel inlet to outlet ) of the channels till all impedances/freq responses are equal
it is mandatory that you do this with the channel isolated from the "outside" world because then it becomes one very hard to manage resonant system
What I have done in the past which works well;
-postion the inlets as explained before and definde the openings as for the desired compression ratio
-choose a reasonable channel length (shorter=better for 2nd harmonic wave steepening distortion, but make sure not too make sharp bends)
-create the exit wavefront shape by positioning the channel outputs
-make every channel equal in lenght by curving the channels
now for the channel shape optimazation;
- terminate each channel closed and fully absorbtive at the exit in ABEC
-drive each channel seperately
-probe impedance at inlet and freq response at outlet and adjust the thickness (for instance in 5mm length steps ftom channel inlet to outlet ) of the channels till all impedances/freq responses are equal
it is mandatory that you do this with the channel isolated from the "outside" world because then it becomes one very hard to manage resonant system
Here's a quick sim of the EXAR 400 scaled 1.4 : 1.My first try, I bought EXAR 400 files and scaled them to 1.4
Polars 0 - 180 / 5deg, normalized @10deg
- Home
- Loudspeakers
- Multi-Way
- Acoustic Horn Design – The Easy Way (Ath4)