Martin Poulin
Best horn Profile, is there such thing?
​
This is a summary of a 128 page Sunday morning read from the master himself, Jean Michel LeCleach' (RIP)
http://www.rintelen.ch/download/JMMLC_horns_lecture_etf10.pdf
​
Below is a measurement of 16 horns profile by JMLC again.
JMLC and Kugelwellen horns are obviously the best profiles while on axis. At 40°, not exactly as good, both those horns do "beam" a lot. The document fails to mention this. Obviously given who wrote it, the inventor of the horn bearing his name may forget to discuss his design flaws... Let's forget him.
http://mariobon.com/Articoli_storici/Horns_measurements_ETF2010d.pdf
​
I own a few horn profiles in 1" format. Either the Eminence N151M or CDX1-1425 will be used.
​
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
​
2000hz flare
on my JMLC
0-15-30-45-60degree
2000hz flare
My friend finished two of four JMLC custom horns.
​
Those are for my newly acquired BMS 4526
The horn has a 2000 Hz profile and will be used at 3 kHz and up.
​
For testing purposes, he eliminated the lip on the contour of the JMLC horn, left it square and measured the result.
​
He then compared it to the finished JMLC with the contour lip.
​
Big difference indeed. The extra ripple is significant thru all of the band.
​
I've read in several documents, people add a lip to terminate the edge of their horn. This mod obtains great results.
This test tends to confirm it.
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Here are 3 horns (all with 1" throat)
​
ME10, B&C speaker, has a Hyperbolic Cosine profile, 90° x 60° pattern, 1500 Hz cut-off (this horn ended up in the garbage bin)
​
STH100, Faital Pro, has tractrix profile, 80° x 70° pattern, 1400 Hz cut-off
​
Jmlc-2000, Auto-Tech, has a round JMLC profile, a round pattern, 2000 Hz cut-off
​
Those 3 different horns will affect four factors (applied 3kHz - 20 kHz)
-
Frequency response
-
Exhibits different diffraction (seen on spectrogram)
-
Distortion (as they load the diaphragm differently)
-
Time coherence (from different air loading, seen on step response)
​
​
​
For the following tests below, I've used the Eminence N151M ring radiator, which is basically a cheap copy of the more expensive BMS drivers. (The graph above is provided in eminence datasheet)
​
We can see that there's a boost @ 3 kHz - 4 kHz, a flat area @ 4 kHz -10 kHz and a sharp drop afterwards. (this requires equalization to flatten out)
​
**The tests below are without equalization**
ME10 horn with the N151M on axis at 12"
Tractrix horn with the N151M on axis at 12"
JMLC horn with the N151M on axis at 12"
jmlc=orange tractrix=green me10=purple
roughly the same
tractrix and ME10 are about same.
jmlc=orange tractrix=green me10=purple
Grouped results for better appreciation of differences.
-Frequency response does change a bit from the horn profile, up to 4 dB difference between profiles (EQ can fix this).​
​​
-Spectogram is cleaner (above 4 kKz) on the JMLC than the other two horns
​
-The distortion profile, from the same driver, varies quite a bit, especially the 3rd harmonic being the lowest on JMLC and 2nd harmonic being lowest on the tractrix. (EQ or DSP can't do anything for correcting distortion)
​
-Energy storage is very different. The JMLC recovers faster with less overshoot. (EQ or DSP is useless for correcting this)
cp22=pink N151m=orange 1425=blue
1425 has least distortion.
cp22=pink N151m=orange 1425=blue
Comparing the eminence N151M to the Celestion CDX1-1425 and the excellent Beyma CP22 bullet tweeter for reference.
​
On the same JMLC Autotech horn, except the CP22 which is a 40° conical horn.
​
-
Cp22 has a tiny breakup at 19500 Hz, but it recovers very quickly
-
CDX1-1425 (metal dome) has a stronger break-up at 16600 Hz and takes much longer to stop ringing as most metal domes do. (mods can fix that)
-
N151M has no break-up at all (spectrogram)
​
-
1425 exhibits significantly less distortion than the other two ring radiators. No surprise here, as it has more radiating surface area. (distortion graph)
-
1425 has the most storage energy, even in the JMLC horn, the CP22 stops the quickest here. (see step response)