Acoustic Horn Design – The Easy Way (Ath4)

An other data point. NB its not an original mabat 520.. Same conditions as Naturlyd above.

BF52_15deg_80cm_4ms_30cm.jpg


80cm above floor, 15 degres, 30cm, 4ms, level cal. SB ROSSO. 300Hz 4th.

Edit: 500ms was uploaded, now 4.

//
 
Last edited:
Hi, attached is curvesImport python script that you can use with Fusion360. Load the profiles CSV file with it to get the splines into a sketch.

For some reason the original script managed to add the spline points in perpendicular to sketch plane and scaled in cm rather than mm. Did some fixes with ChatGPT, fixed script is attached. It isn't perfect still as it creates multiple sketches one of which contains the curve and the rest are empty, but was lazy to fix that 😀 Have fun!
 

Attachments

  • Like
Reactions: poiy
How do you interpret my impulse response?
Looks like a typical high pass response to me.
It occurred to me that I may have misinterpreted your question. If you mean the burst of high-frequency energy at 0.8-1.2ms, I agree that it's likely diaphragm resonances (break up). The period seems to be about 70µs or so, which is ~14kHz. When you have multiple closely-spaced peaks and/or dips, you can get beats and/or apparent "bursts" in the impulse response. Same idea as two or more closely-spaced simultaneous tones beating as they alternate between constructive and destructive interference.
 
  • Thank You
Reactions: Naturlyd
I usually insert a L-pad with fixed resistors on the compression driver. Takes care of that little bit of noise in the playback chain with 105-115dB sensivity. When Mabat wrote about measuring an external 1Khz noise source in his impedance measurement I thought that I should test how sensitive my system was for extarnal noise and also that I should try out pseudo Current drive using serial resistance instead of my L-pads.

Here is a recording of the impedance of the JBL 2452SL driver in the A520 horn with and without 24.6 Ohms of series resistance. Measuring with DATS the overall level of the impedance varied from measurement to measurement on the bare driver. With series resistance it was very consistent. You can clearly se the various tones I used to exite the impedance measurement. I used the bass/mid driver in the cabinet underneath the horn as a source for the test tones while doing the impedance measurements.

A520 - JBL 2452SL w series resistance DATS.jpg




To check how the series resistance changed the frequency response I did some sweeps with 8.2, 16.4, 24.6 Ohms compared with the bare driver in the A520 horn. Using DSP its no problem to compensate.


A520 - JBL 2452SL w series resistance.jpg



How would one go about to establish if the horn/driver system is more immune to external influence with current drive? Will the distortion measurement in REW be accurate?
 
  • Like
Reactions: tmuikku
That's a crazy difference for a non reactive component. What kind of resistor - wire wound? These wide band differences are easily detected by ear. I did this test also and measured distortion after introducing a resistor... I found out that one need to re EQ for every change before distortion measurement as things change so much.

//
 
Anyway, to make at least a partial conclusion, I think it's clear that the 520 doesn't need any surface warping, as @Naturlyd showed in his measurements, which seem to be very clean. What's still not clear is the mechanism how the midrange "non-cleanesness" originate in the smaller horns. I'd be surprised if it were just my specimens.

Now I'm going to measure the acoustic response inside the same 4554 adapter with all three horns. The assumption is that the 520 should get the least amount of reflected sound back into the throat.
 
Last edited:
  • Like
Reactions: Naturlyd
How would one go about to establish if the horn/driver system is more immune to external influence with current drive? Will the distortion measurement in REW be accurate?
Maybe best to conduct AB test listening if at all possible. If you want to measure I'd say a multitone distortion test, or FSAF thing where you can extract the residual to a sound file and listen the difference, one thread about it here https://www.diyaudio.com/community/...sing-music-michael-tsiroulnikovs-fsaf.418843/

Some tones are missing on your first image resistor measurements, is this intentional?

The second image showing difference in frequency response kinda indicates that indeed electrical damping works for both the main resonance (lowest impedance peak in impedance plot) and also at the loading ( high acoustic impedance ) bandwidth around 1-2kHz. At the loading bandwidth diaphragm excursion is kinda low so perhaps this doesn't matter at all other than what the effect is on frequency response, but any extra resonances at the main resonance might have effect on sound as that is relatively high excursion movement. Since the diaphragm seems to be prone to external excitation (microphonics) then it could make the whole bandwidth to distort a bit if diaphragm main resonance is not damped well. For this reason I think electrical damping is of benefit if it's possible to arrange.

On active system, driver connected directly to regular voltage amplifier electrical damping is already fine, but if there is resistor pad to reduce noise then it's likely benefitical to connect inductor in parallel with the driver to maintain low impedance = electrical damping at main resonance.

Regarding microphonics on the loading bandwidth:
I tried to get my head around on the loading, what it actually is, and I think currently that basically the diaphragm and air column in the throat become very well connected, one could imagine the diaphragm and air at the throat essentially become one. Efficiency goes up but also it's a form of resonance, hence hump in electrical impedance and boost in acoustic output. Microphonics is enhanced here as well, sound entering the throat directly translates into movement of the diaphragm, but since efficiency is high and sound pressure is relatively small the movement of diaphragm due to microphonics on this bandwidth is small even though it looks big from the electrical impedance measurement due to efficiency. Thus, I think microphonics due to high loading is no issue, and electrical damping at this high acoustic impedance bandwidth (loading) is not necessary due to small displacement, thus it's likely fine to have high impedance here in order to suppress any distortion products from high excursion low frequency output.
 
Largest possible reflection free zone is really the key here. Not only the room size itself but also the surroundings of the waveguide. Anyway, I keep measuring the systematic results of my three waveguides (520/460/400) in exactly the same setup, so there must be something going on, at least in my case.

I wound need ideally a few weeks of free time to concentrate on this... 🙂
 
Recently there was a discussion about a driver picking up noise from the environment during the impedance measurements. I tested something similar with clapping and hissing close to the Faital HF108 (8 ohms) driver I use in my speakers.

In these measurements the driver is connected directly to a low noise amplifier (Topping B100, set to 0dB gain) and a microphone input of the Audioprobe Spartan A is connected in parallel to the faston connectors of the driver. The graphs are not calibrated, but the 0dBFS is about 1Vrms.

hf108-mic-ampoff.png


hf108-mic-ampon.png


Old measurement of an HF108 in the Sandhorn (the driver was not well aligned at that time):
hf108 imp.png
 
  • Like
Reactions: Naturlyd
It took some time but finally I could measure the DCX 464 with modified internal geometry on a G2 460.

IMG_20250105_215827.jpg
IMG_20250106_142624.jpg


The new insert features Marcels BMS 4554 geometry starting right at the entrance of the midrange channels; a few mm below the HF diaphragm. Thank of the help from alex who overworked my 3D scans the first adapter was already 90% perfect. Some minor adjustments made the second attempt fit completely. The outside adapter has at its end the usual Gen 2 interface.

IMG_20250110_203552.jpg
IMG_20250110_203601.jpg


Black filament was out of stock so....something different 🙂

How does it perform?
I attached the REW file for whoever wants to check the results more in detail.
I measured 0 / 10 / 20 / 30°
No smoothing applied.
12ms windowed with additional FDW 12 cycles
Accidently level at mic pre was changed between mid and high measurement. Please ignore absolute levels. All was at ~90db

  • In the image below I tried to generate additional curves in relative values to the 0° measurement for better comparison.
  • Waterfall, Decay, Spectrogram are all without issues
  • This HF drop @5,4kHz is not visible in any way in distortion. I tried to measure 5,4kHz and raised loudness. There was no issues in distortion also on higher levels. This seems not to be a resonance and can in my opinion be EQed without issues
  • I also measured at 105 and 117 db. There the measurement mic started distorting. I need to check why (maybe the preamp or levels).


dcx464.jpg



So.....any comments on this?
 

Attachments