I got similar results with a modified 8 inch Lowther PM2 in a 60x60x54 cm horn, very uneven response. The best measuring driver was the JBL CMCD. Much thanks to its phase plug. But tonally, a bit weak.
The fullrange Lowther had everything going for it, light cone, huge magnet, but I think it is the suspension that is too weak for a fullrange to be a good «compression driver». And the paper is often flimsy, at least the whizzer cone.
The fullrange Lowther had everything going for it, light cone, huge magnet, but I think it is the suspension that is too weak for a fullrange to be a good «compression driver». And the paper is often flimsy, at least the whizzer cone.
And your point is, exactly?
Why did you quote me there?
I was referring to Holland and Newell's article, which was clearly about horns for compression drivers, and where the authors made some generalizations that were IMHO not well supported by the evidence.
You posted your sim and measurements of a short horn loading an undisclosed (and whizzer-cone equipped?) 8" fullrange. Not sure how this is relevant or in which way it either proves or disproves what I was saying.
I'm honestly confused - not trying to start an argument!
Well, I think maybe someone will find it interesting but ok not in this thread maybe. I can only say that horn like this must be properly equalised. Longer horns usually have "horn sound" because of directivity sonogram character. No, 8" driver is enclosed to see low end response compared to Hornresp sim.
Drawing conclusions that aren't wholly supported by the data is unfortunately all too common, but in cases like these where the experiments and results are shared quite openly the reader can decide how much weight to give.
Once («in band camp») I build a copy of the Avantgarde Trio midbass horn and lengthened it to fit a JA6681B. It must have been over 1m in diameter and 110cm deep. Incredible details down to about 200Hz, but the shorter replica of the Trio midrange horn was much cleaner, but from 550Hz and up, so no wow-factor in lower midrange details.
Last edited:
Response measurement of throat adapter Prototype #2.
In Post #186 two objectives were laid out for converting a circular to a rectangular surface area.
1. Getting the correct surface area profile for the objective hypex T and Fc. One can achieve this easily by adjusting the width between the adapter walls along its profile. The walls can be vertical (no convexity) with roundovers near the roof and floor.
2. Shaping the spherical wavefront to a cylindrical wavefront. Vertical walls are problematic here, since speed of sound being constant, the synchronized wave edge on the circular entrance then arrives at different times on the vertical exit wall, and thus does not produce a cylindrical wave front at the exit. This is more tedious to solve and involves tradeoffs.
Adapter Prototype #2 (Posts #169, #191 and #201) implements the idea of equal length arcs on the side walls from circular entrance to straight vertical exit (Post #168). Its convex walls (bumps when viewed from inside or dimples when viewed from outside) aim to satisfy #2 above.
Shown below is measurement of 3D-printed adapter Prototype #2 compared to my first printed adapter that had vertically straight walls and a hypex flare profile (Post #127 to #133). The two adapters were measured on the same horn and driver unit without filtering or EQ. Their flare rate profiles differ slightly, so this is not a perfect experiment, but because the on-axis measurements are close to identical (as shown below), I believe that the off-axis differences can be attributed to the vertical convexity of the Prototype #2 walls. Recall that the driver exit is 1.4 inch and horn entrance 40 x 50 mm. Psychoacoustic filtering was applied to all curves to make differences between adapters more visible.
Measured response of the convex wall versus vertical wall adapters measured near the left edge and middle elevation of the horn mouth. Reference Prototype #1 is shown in green and new Prototype #2 in orange.
Near the left and top edge (diagonal corner) of the horn mouth. This produces worst-case measurements that are seldom published.
And for completeness, on central axis of the horn mouth (top curves), and near the centre and the top edge of the horn mouth (bottom curves).
Observations
Looking at the horizontal off-axis measurements, the improvements provided by Prototype #2 exceeded my expectations. The frequency response is flatter and more extended at high frequencies (3 to 14 kHz). When not in the "sweet spot", a change of this magnitude should be easy to hear.
On-axis and vertical off-axis measurements show little differences in frequency response, nor in transient/ waterfall response (not shown).
In all measurements, the lower frequency extends slightly more for Prototype #2. Most likely this is by virtue of slightly closer matching of area flare rates (driver 8 to adapter 6.9, and adapter exit 12.7 to horn 10). The benefits of this flare rate matching appear to outweigh the more severe roof and wall flare tangent angular discontinuities compared to there reference adapter.
Bottom line is I am very pleased with Prototype #2. Prototype #3 will likely amount to minor dimensional tweaks and be printed in pair so I can listen in stereo.
Pierre
PS Since some expressed interest, I started thinking about an adapter for the 50 mm TAD driver and Athos Yuichi A290 horn.
In Post #186 two objectives were laid out for converting a circular to a rectangular surface area.
1. Getting the correct surface area profile for the objective hypex T and Fc. One can achieve this easily by adjusting the width between the adapter walls along its profile. The walls can be vertical (no convexity) with roundovers near the roof and floor.
2. Shaping the spherical wavefront to a cylindrical wavefront. Vertical walls are problematic here, since speed of sound being constant, the synchronized wave edge on the circular entrance then arrives at different times on the vertical exit wall, and thus does not produce a cylindrical wave front at the exit. This is more tedious to solve and involves tradeoffs.
Adapter Prototype #2 (Posts #169, #191 and #201) implements the idea of equal length arcs on the side walls from circular entrance to straight vertical exit (Post #168). Its convex walls (bumps when viewed from inside or dimples when viewed from outside) aim to satisfy #2 above.
Shown below is measurement of 3D-printed adapter Prototype #2 compared to my first printed adapter that had vertically straight walls and a hypex flare profile (Post #127 to #133). The two adapters were measured on the same horn and driver unit without filtering or EQ. Their flare rate profiles differ slightly, so this is not a perfect experiment, but because the on-axis measurements are close to identical (as shown below), I believe that the off-axis differences can be attributed to the vertical convexity of the Prototype #2 walls. Recall that the driver exit is 1.4 inch and horn entrance 40 x 50 mm. Psychoacoustic filtering was applied to all curves to make differences between adapters more visible.
Measured response of the convex wall versus vertical wall adapters measured near the left edge and middle elevation of the horn mouth. Reference Prototype #1 is shown in green and new Prototype #2 in orange.
Near the left and top edge (diagonal corner) of the horn mouth. This produces worst-case measurements that are seldom published.
And for completeness, on central axis of the horn mouth (top curves), and near the centre and the top edge of the horn mouth (bottom curves).
Observations
Looking at the horizontal off-axis measurements, the improvements provided by Prototype #2 exceeded my expectations. The frequency response is flatter and more extended at high frequencies (3 to 14 kHz). When not in the "sweet spot", a change of this magnitude should be easy to hear.
On-axis and vertical off-axis measurements show little differences in frequency response, nor in transient/ waterfall response (not shown).
In all measurements, the lower frequency extends slightly more for Prototype #2. Most likely this is by virtue of slightly closer matching of area flare rates (driver 8 to adapter 6.9, and adapter exit 12.7 to horn 10). The benefits of this flare rate matching appear to outweigh the more severe roof and wall flare tangent angular discontinuities compared to there reference adapter.
Bottom line is I am very pleased with Prototype #2. Prototype #3 will likely amount to minor dimensional tweaks and be printed in pair so I can listen in stereo.
Pierre
PS Since some expressed interest, I started thinking about an adapter for the 50 mm TAD driver and Athos Yuichi A290 horn.
Attachments
This isn't that weird, BEM has limitations when it comes to the accuracy of modelling cone depth, below a certain frequency the depth gives a more accurate answer but above it a flat piston can be a better approximation.
Jorg Panzer has a paper on it
http://www.randteam.de/_Docs/Pub/Aes112 Radiation impedance cones.pdf
Flare rate modelling with sensitivity analysis of the TAD TH-4001 Horn clone sold by Athos Audio.
This horn by Athos Audio does not necessarily have the same dimensions as the TAD-made horn. Judging by the TAD drawing, there could be small differences. I don't know.
The Athos Audio horn entrance measures 40 x 50 mm. Five expansion "cells" having identical dimensions make up the first part of the expansion. The side walls have the same curvature as the fins. I estimate that the fins are 155 mm long and measure 11 mm at their widest. Cell width close to this mid-fin-length point is 22.5 mm and at the end of the fins 55 mm. The mouth measures 660 x 230 mm. I estimated flare rate first without and then with the round-to-rectangular 20 mm long adapter that comes with this horn. The f and T estimates below were obtained through an iterative search software.
A. Best fit hyperbolic expansion, rectangular 40 x 50 horn entrance
f = 318.888, T = 0.868346, MAP Error = 0.607996 %
x = 0, S = 2000, est = 2000, m = 9.92556
x = 77.5, S = 4500, est = 4554.41, m = 10.8221
x = 155, S = 11000, est = 10865.5, m = 7.54322
x = 385, S = 155000, est = 155000
B: Same as A, here S_0 accounting for 2 mm corner roundovers at horn entrance
f = 317.306, T = 0.924858, MAP Error = 0.352883 %
x = 0, S = 1928, est = 1928, m = 10.3258
x = 77.5, S = 4500, est = 4525.65, m = 10.8221
x = 155, S = 11000, est = 10907.4, m = 7.54322
x = 385, S = 155000, est = 155000
C: Same as B, x_0 moved 3 mm within rectangular opening (est. wavefront sagitta)
f = 315.683, T = 0.974347, MAP Error = 0.164333 %
x = 0, S = 1942, est = 1942, m = 10.6599
x = 74.5, S = 4500, est = 4517.71, m = 10.8221
x = 152, S = 11000, est = 10971, m = 7.54322
x = 382, S = 155000, est = 155000
D: Same as C, defining x_0 to lie at the round TAD adapter entrance
f = 328.097, T = 0.594403, MAP Error = 4.66926 %
x = 0, S = 1963, est = 1963, m = -0.537772
x = 20, S = 1942, est = 2283.73, m = 10.6599
x = 94.5, S = 4500, est = 4532.85, m = 10.8221
x = 172, S = 11000, est = 10447.8, m = 7.54322
x = 402, S = 155000, est = 155000
E: Same as C, fin width estimate increased to 13 mm, cell width estimate narrowed to 20.5 mm
f = 316.78, T = 0.959122, MAP Error = 2.61705 %
x = 0, S = 1942, est = 1942, m = 9.58836
x = 74.5, S = 4100, est = 4491, m = 11.7924
x = 152, S = 11000, est = 10897.5, m = 7.54322
x = 382, S = 155000, est = 155000
Observations
The cutoff frequency of this horn is around 320 Hz, as specified.
Without the round throat adapter, the flare rate in the portion closest to the rectangular entrance of the horn is approximately m = 10. Between the end of the fins and the horn mouth, flare rate m reduces to 7.5.
Factoring in the round corners of the rectangular entrance (B), and estimating a cylindrical wavefront (C and E), the expansion is almost exponential (hypex constant T approaching 1).
Including the round horn adapter in the estimates (D) yields an overall hyperbolic expansion with T = 0.6.
To increase the fidelity of the flare rate estimate right at the entrance would require the exact fin profile, which I don't have. On the bright side, varying the dimensions as I did above does not change the results a lot.
Bottom line is that a throat adapter with exit flare rate close to m = 10 should match this horn well.
This horn by Athos Audio does not necessarily have the same dimensions as the TAD-made horn. Judging by the TAD drawing, there could be small differences. I don't know.
The Athos Audio horn entrance measures 40 x 50 mm. Five expansion "cells" having identical dimensions make up the first part of the expansion. The side walls have the same curvature as the fins. I estimate that the fins are 155 mm long and measure 11 mm at their widest. Cell width close to this mid-fin-length point is 22.5 mm and at the end of the fins 55 mm. The mouth measures 660 x 230 mm. I estimated flare rate first without and then with the round-to-rectangular 20 mm long adapter that comes with this horn. The f and T estimates below were obtained through an iterative search software.
A. Best fit hyperbolic expansion, rectangular 40 x 50 horn entrance
f = 318.888, T = 0.868346, MAP Error = 0.607996 %
x = 0, S = 2000, est = 2000, m = 9.92556
x = 77.5, S = 4500, est = 4554.41, m = 10.8221
x = 155, S = 11000, est = 10865.5, m = 7.54322
x = 385, S = 155000, est = 155000
B: Same as A, here S_0 accounting for 2 mm corner roundovers at horn entrance
f = 317.306, T = 0.924858, MAP Error = 0.352883 %
x = 0, S = 1928, est = 1928, m = 10.3258
x = 77.5, S = 4500, est = 4525.65, m = 10.8221
x = 155, S = 11000, est = 10907.4, m = 7.54322
x = 385, S = 155000, est = 155000
C: Same as B, x_0 moved 3 mm within rectangular opening (est. wavefront sagitta)
f = 315.683, T = 0.974347, MAP Error = 0.164333 %
x = 0, S = 1942, est = 1942, m = 10.6599
x = 74.5, S = 4500, est = 4517.71, m = 10.8221
x = 152, S = 11000, est = 10971, m = 7.54322
x = 382, S = 155000, est = 155000
D: Same as C, defining x_0 to lie at the round TAD adapter entrance
f = 328.097, T = 0.594403, MAP Error = 4.66926 %
x = 0, S = 1963, est = 1963, m = -0.537772
x = 20, S = 1942, est = 2283.73, m = 10.6599
x = 94.5, S = 4500, est = 4532.85, m = 10.8221
x = 172, S = 11000, est = 10447.8, m = 7.54322
x = 402, S = 155000, est = 155000
E: Same as C, fin width estimate increased to 13 mm, cell width estimate narrowed to 20.5 mm
f = 316.78, T = 0.959122, MAP Error = 2.61705 %
x = 0, S = 1942, est = 1942, m = 9.58836
x = 74.5, S = 4100, est = 4491, m = 11.7924
x = 152, S = 11000, est = 10897.5, m = 7.54322
x = 382, S = 155000, est = 155000
Observations
The cutoff frequency of this horn is around 320 Hz, as specified.
Without the round throat adapter, the flare rate in the portion closest to the rectangular entrance of the horn is approximately m = 10. Between the end of the fins and the horn mouth, flare rate m reduces to 7.5.
Factoring in the round corners of the rectangular entrance (B), and estimating a cylindrical wavefront (C and E), the expansion is almost exponential (hypex constant T approaching 1).
Including the round horn adapter in the estimates (D) yields an overall hyperbolic expansion with T = 0.6.
To increase the fidelity of the flare rate estimate right at the entrance would require the exact fin profile, which I don't have. On the bright side, varying the dimensions as I did above does not change the results a lot.
Bottom line is that a throat adapter with exit flare rate close to m = 10 should match this horn well.
Having now listened to five adapters on the same driver and horn, I can confirm that the adapter plays more than just a plumbing role. Although adapters don't make as much difference in sound as the driver or the horn - as long as they are approximately the right length - they do alter the speaker's sound.
In the frequency domain
As observed earlier, the adapter impacts the frequency response on and off axis in measurable and audible ways. The result can be positive (e.g. smoothing the response) or negative. There are too many acoustic mechanisms to get into here (standing waves, dispersion, etc.). But generally I find that this impact in the frequency domain is not too concerning to us, fearless DIYers, as we can tweek our filter and voice another piece of our system to compensate.
Soundstage
Commenting briefly on the TH4001 horn, I appreciate how it positions the drivers closer together than the round AH425, and as a result, how musical instruments sound better integrated, e.g. notes don't appear to climb up or down in space according to their frequency. Tonality remains acceptable even if I move my head somewhat close to the speakers' plane (a hard test). The image shifts left or right within the sweet spot (unavoidable), yet the tonality remains mostly unchanged. The fins of the TH4001 rule here, and the adapters don't seem to make much difference.
In the time domain
My most educative session happened when listening to the composite adapter(*) in one speaker and Prototype 2 in the other. My initial impression was that something was wrong in Prototype 2. This speaker sounded lifeless compared to the one fitted with the composite adapter. Then, after a few tracks, I caught myself wondering how strange it is that mixing engineers always put the "reverb" in only one and the same channel (the composite adapter one). As it turned out, I could hear what I believe is "unintended reverberation" caused by flare mismatch in the composite adapter. Perhaps this arises due to standing waves in the presence frequency range. In all fairness, the resulting effect is not unpleasant to the ear. In front of me stood a "wet" and a "dry" speaker. Had the two speakers been "wet" (composite adapters) I don't think I would have been displeased, or ever noticed. But as pleasant as it might sound, this distortion is not on the recordings and I personally don't care for it.
(*) composite adapter is 1.4" to 2" conical followed by 2" to rectangular Athos stock adapter.
Bottom line
Over the last few days I have been listening to flare-rate-fitted Prototype 3 adapters in both speakers, and what strikes me is how the speakers now disappear. They draw very little attention to themselves. I notice other small improvements in transparency, soundstage focus, and air. The sum of these many small improvements mean that the Prototype 3 adapters are here to stay.
More generally, minimizing stored energy has been a worthwhile objective throughout my speaker quest: Be diaphragm driver, milled wood horn, sealed midwoofer enclosure, and now flare-rate-fitted adapters, all contribute to a more natural and non-fatiguing sound from top to bottom.
In the frequency domain
As observed earlier, the adapter impacts the frequency response on and off axis in measurable and audible ways. The result can be positive (e.g. smoothing the response) or negative. There are too many acoustic mechanisms to get into here (standing waves, dispersion, etc.). But generally I find that this impact in the frequency domain is not too concerning to us, fearless DIYers, as we can tweek our filter and voice another piece of our system to compensate.
Soundstage
Commenting briefly on the TH4001 horn, I appreciate how it positions the drivers closer together than the round AH425, and as a result, how musical instruments sound better integrated, e.g. notes don't appear to climb up or down in space according to their frequency. Tonality remains acceptable even if I move my head somewhat close to the speakers' plane (a hard test). The image shifts left or right within the sweet spot (unavoidable), yet the tonality remains mostly unchanged. The fins of the TH4001 rule here, and the adapters don't seem to make much difference.
In the time domain
My most educative session happened when listening to the composite adapter(*) in one speaker and Prototype 2 in the other. My initial impression was that something was wrong in Prototype 2. This speaker sounded lifeless compared to the one fitted with the composite adapter. Then, after a few tracks, I caught myself wondering how strange it is that mixing engineers always put the "reverb" in only one and the same channel (the composite adapter one). As it turned out, I could hear what I believe is "unintended reverberation" caused by flare mismatch in the composite adapter. Perhaps this arises due to standing waves in the presence frequency range. In all fairness, the resulting effect is not unpleasant to the ear. In front of me stood a "wet" and a "dry" speaker. Had the two speakers been "wet" (composite adapters) I don't think I would have been displeased, or ever noticed. But as pleasant as it might sound, this distortion is not on the recordings and I personally don't care for it.
(*) composite adapter is 1.4" to 2" conical followed by 2" to rectangular Athos stock adapter.
Bottom line
Over the last few days I have been listening to flare-rate-fitted Prototype 3 adapters in both speakers, and what strikes me is how the speakers now disappear. They draw very little attention to themselves. I notice other small improvements in transparency, soundstage focus, and air. The sum of these many small improvements mean that the Prototype 3 adapters are here to stay.
More generally, minimizing stored energy has been a worthwhile objective throughout my speaker quest: Be diaphragm driver, milled wood horn, sealed midwoofer enclosure, and now flare-rate-fitted adapters, all contribute to a more natural and non-fatiguing sound from top to bottom.
Thanks Allen. When one views the adapter as part of the horn, then I find it not so surprising that it must be a certain length.
The length of Prototype 3 is entirely determined by the surface areas of the driver, horn entrance, and my choice of linear increase in flare rate from 8.2 to 10 (horn flare rate). There is a little bit of leeway, if one were to choose a constant flare rate, or a non-linear increase in flare rate for example, but these choices have little impact on the resulting length of the adapter for this particular driver-horn pair. This length lands somewhere between 70 and 75 mm. For another horn having higher flare rate, the adapter could be shorter of course.
The length of Prototype 3 is entirely determined by the surface areas of the driver, horn entrance, and my choice of linear increase in flare rate from 8.2 to 10 (horn flare rate). There is a little bit of leeway, if one were to choose a constant flare rate, or a non-linear increase in flare rate for example, but these choices have little impact on the resulting length of the adapter for this particular driver-horn pair. This length lands somewhere between 70 and 75 mm. For another horn having higher flare rate, the adapter could be shorter of course.
Adapters
@secanbj - Yes the surface area increase from TAD 4001 driver to Yuchi horn entrance is large enough for my adapter profile. As a matter of fact, I have a pair of adapters in front of me that were generated specifically for this driver and horn combination. These are already sold however. If you are interested please contact me by PM and I can have a pair manufactured for you.
Thank you - Pierre
@secanbj - Yes the surface area increase from TAD 4001 driver to Yuchi horn entrance is large enough for my adapter profile. As a matter of fact, I have a pair of adapters in front of me that were generated specifically for this driver and horn combination. These are already sold however. If you are interested please contact me by PM and I can have a pair manufactured for you.
Thank you - Pierre