Acoustic Horn Design – The Easy Way (Ath4)

Assuming the side mounted low (midrange..) cone drivers output output is time aligned with delay to the high frequency compression driver, the on axis distance being more than 1/4 wavelength apart is not an issue.
What about the reflection of the low/midrange sound wave off the throat? Shouldn't it be as close as possible to the throat also for this reason? The sound emitted by the midrange ports doesn't propagate only towards the mouth.
 
Last edited:
Is there any easy attempt to check horizontal Dispersion match vor 8/10/12" LF driver? Not much beaming in that region anyway...
I think that something similar in size to the waveguide, or a little smaller, will always turn out as best. That's a proven rule of thumb. Even the smallest of the Gen horns still has more than 6 dBDI around 600 Hz, the bigger ones are like 7.5 dB. You need something of a comparable size to keep up.
 
Last edited:
I printed the A520G2 and T520-25-STD-1 and here are the results with BMS 5530ND and B&C DE360 (4 ms gate, no smoothing):
5530ND on A520G2 and STD-1.png

DE360 on A520G2 and STD-1.png


My conclusion is that the 5530ND is smoother and has better output around 1 kHz. Distortion is very similar.

@mabat: Could your DE360 be damaged somehow?

One note: I noticed during assembly that the petals had lifted off of the build plate at the end closest to the throat. This has probably affected the final shape a bit and I'll try printing a new set.
 
I can redo my measurement of the DE360, yours is quite different indeed. What surprises me are all the small wiggles in the HF response of the 5530ND. I never saw this in any of my (gated) data. What would be your explanation?

The on-axis dip around 2.2k in the response of the DE360 is also strange - this shouldn't depend on a driver, IMO. Would be interesing to trace this back.

- Otherwise the 5530 looks pretty similar to my measurements of the 4554 in A520G2 / T520-36-STD-1.
 
  • Like
Reactions: keyser
Comparable size of LF to the horn itself would dictate a 15". I am not happy with that for a XO of 600Hz.

What about a smaller one in a waveguide for best match? For such low frequency quite easy. I remember there is a tool (it was named ATH or so...) that makes it very comfortable 🤣
In that case it would be cool to match it to the HF horn. Like theese waveguides...just with horns. You get the idea?

1734683965952.jpeg
 
I can redo my measurement of the DE360, yours is quite different indeed. What surprises me are all the small wiggles in the HF response of the 5530ND. I never saw this in any of my (gated) data. What would be your explanation?

The on-axis dip around 2.2k in the response of the DE360 is also strange - this shouldn't depend on a driver, IMO. Would be interesing to trace this back.

- Otherwise the 5530 looks pretty similar to my measurements of the 4554 in A520G2 / T520-36-STD-1.
I'm guessing that it's the warped print causing a bad transition between the base and the petals. It could also be the table the waveguide was resting on I guess. I'll try a better measurement setup next time.
 
In my mind the only difference is the objective when designing it. A horn is to amplify sound, while a waveguide is to control directivity.

mabat: Here are some measurements that I hope help you design a 5530ND version of the extended adapter. Photos for reference, not actual measurement.
The end of the throat connected to the diaphragm is around 19.5 mm and at the start (where the channels come in from the sides) it's 16.3 mm.
1734689805693.png


The hole in the magnet assembly closest to the phase plug is 20 mm:
1734690063641.png


The opening on the other side, where it meets the waveguide is 24.8 mm (not 25.4):
1734690192119.png


The thickness of the entire magnet assembly is 20 mm:
1734690255710.png


There is also a 0.7 mm step where the foam gasket (on top of bug screen) was placed.

A photo down the throat after removing bug screen:
1734689637688.png

This photo also shows the 0.7 mm step for the foam gasket.
 
  • Like
Reactions: Naturlyd
I would also need to know the depths of all the different sections that have different wall slopes.

- Could you perhaps try to make something like this?

1734692374641.png

I had to do this with the 4554 - to print a few iterations of such "template" and only after it fitted perfectly I could continue, as the first try based on the measured numbers was still not accurate enough.
 
Last edited:
Would you say there is an angle change between the conical section in the diaphragm assembly and the conical section through the magnet for the 4554? Looking at the 5530ND I can't see any of that. This is what I would say the cross-section looks like:
1734699707890.png

I believe you could live with the 0.1-0.2 mm gap above the red part and thereby not make the geometry more complex by avoiding the step:
1734700959461.png