Horn Honk $$ WANTED $$

Status
Not open for further replies.
Hello,

...
Inversely to the common knowledge waveguide don't have less HOMs than horns. What an OSwaveguide do better than horns is that it uses the HOMs in order to reflect them in various directions in order to obtain a more evenly distributed directivity.

...
Best regards from Paris, France

Jean-Michel Le Cléac'h
Quite interestingly, the simulations I have performed do not show a firm trend in this regard. The important thing just seems related with how to control the expansion process to minimize reflections.
 
Hello Rob,

Here are the relevant messages :

Earl's measurement on his OS waveguide
Waveguide_00 (it appeared that Fs = 32kHz and there was some noise)

http://www.diyaudio.com/forums/multi-way/123426-horn-vs-waveguide-10.html#post1516725

Earl's measurement on a JBL waveguide (which reference?)

http://www.diyaudio.com/forums/multi-way/123426-horn-vs-waveguide-10.html#post1516728


Earl's new measurement on the same OS waveguide
http://www.diyaudio.com/forums/multi-way/123426-horn-vs-waveguide-13.html#post1517241

validation by Earl Geddes of that new measurement:
http://www.diyaudio.com/forums/multi-way/123426-horn-vs-waveguide-17.html#post1518269

"the waveguide low noise is definately what I expect."

My measurements on one of my Le Cléac'h horns:

Le Cléac'h horn with curent driven TD2001

http://www.diyaudio.com/forums/multi-way/123426-horn-vs-waveguide-52.html#post1527765

Le Cléac'h horn with voltage driven TD2001
http://www.diyaudio.com/forums/multi-way/123426-horn-vs-waveguide-53.html#post1528568

Best regards from Paris, France

Jean-Michel Le Cléac'h


I didn't know there was a second measurement and I stand corrected.
 
Hello Panomaniac,

You can find in the attached file 2 graphs obtained using my Matlab quasi-wavelets routine on the impulse response of the Pioneer horn you gave. One is with the normalized response, the other one for the not-normalized response.

Which Pioneer horn is this (TH4001?)

Best regards from Paris, France

Jean-Michel Le Cléac'h


I'd like to see a wavelet or good CSD of the Pioneer horn, but I"m having a devil of a time. Can anyone help? Impulse is attached below.

And here is the FR graph. Remarkable for a horn. 2nd harmonic rises sharply below 750Hz (not shown here).

If someone can run a graph on this and tell me what your setting were, that would be great. Thnx
 

Attachments

  • Pioneer_horn.gif
    Pioneer_horn.gif
    73.8 KB · Views: 245
Last edited:
What JBL waveguide is that??
Context will remind you -- Earl was comparing EconoWave to his product. The "JBL horn" was the $9.90 90° x 50° thread-on progressive transition waveguide 338800-001 driven with a $40 Selenium D220Ti compression driver.

As to the issue of noise, Earl provided a relatively noise-free impulse he had taken on his waveguide elsewhere. It should be easy enough to compare wavelets from the two OS impulse samples he attached in that thread.

You can see me request that the noise be removed from the JBL impulse for comparison, but that did not happen.

CLIO wavelet upgrade is here, but not yet installed. It'll take me some time to crawl up the learning curve, probably; see Chapter 18 of the CLIO 8.5 User Manual for the capabilities:

http://www.audiomatica.com/clio10/download10.htm
 
Last edited:
Hello Soongsc,

Don't misinterpret what I wrote, I did'nt say that the most part of the control of the directivity of a waveguide is due to HOMs.

But surely they play their part in filling the holes in the radiation pattern....

Best regards from Paris, France

Jean-Michel Le Cléac'h



The important thing just seems related with how to control the expansion process to minimize reflections.
 
Hi,

Here's the post for JBL and OS comparisons:
post #323
http://www.diyaudio.com/forums/multi-way/161627-horn-honk-wanted-7.html#post2124412

Yes there is a large amount of reflections/diffractions before 1ms in OS seen here also in this plot 4kHz-20kHz:
An externally hosted image should be here but it was not working when we last tested it.



Mouth reflections are different phenomena than the above but they do exist in OS
163324d1269038638-horn-honk-wanted-geddes_waveguide_wavelet1_reflections.png



- Elias


Hello,

You'll see in attached file a comparison I have done using my Matlab routine (quasiwavelets) between a Le Cléac'h horn (Fc = 320Hz) a JBL waveguide and an OSwaveguide. I modified somewhat my routine in order that the shown graphs are normalized in amplitude.

Elias yet gave some multiresolution wavelets graphs for those horns and waveguides in the message:

http://www.diyaudio.com/forums/multi-way/161627-horn-honk-wanted-33.html


The vertical stripes having an yellow-orange color you can see on the graphs for the JBL and the OS_waveguide appear within the interval of time between 0 and 1ms. This means that there are due to reflexions having a short delay( by that I mean that the correspodning delay is smaller than the delay due to the reflection between mouth and throat.)

IMHO a large part of those short delay reflections are HOMs.

Inversely to the common knowledge waveguide don't have less HOMs than horns. What an OSwaveguide do better than horns is that it uses the HOMs in order to reflect them in various directions in order to obtain a more evenly distributed directivity.

To confirm or not those yellow-orange stripes are du to HOMs we should analyse with the same method the impulse response at different off-axis angles.

Best regards from Paris, France

Jean-Michel Le Cléac'h

165825d1270549429-horn-honk-wanted-3horns_.gif
 
Hello,

I agree with Jean-Michel, there is no noise issues with those IR files that would prevent wavelet analysis. Inherently built in feature of wavelet analysis is its noise rejection ability due to its band pass filtering nature.


See post #323 in this thread to see the noise level of JBL and OS files. See before the beginning of the wavelet above 1kHz, and you'll see the noise is below -35...-40dB in the normalised plots (right hand side). And then compare the plots showing the temporal reflections/diffractions and you'll see they are much higher in amplitude thus having nothing to do with noise.

Even if the noise would be at higher level, one could still see the temporal sequence of reflections because they do not correlate with random noise.


- Elias


There was a noise issue when Earl made those measurements. That could easily be a reason for all the clutter in the first milisecond.


There is no noise issue there.


As to the issue of noise, Earl provided a relatively noise-free impulse he had taken on his waveguide elsewhere. It should be easy enough to compare wavelets from the two OS impulse samples he attached in that thread.

You can see me request that the noise be removed from the JBL impulse for comparison, but that did not happen.
 
Hello,

This is a very interesting statement Jean-Michel!

If some of the pre 1ms reflections are due to HOM, how the reflection pattern should behave versus direction angle?

Any volunteers to do a set of IR measurements versus angle of some horns?

I can measure 18sound small horn later for this purpose, thats all that I have at the moment. I do have IR files at different angles already but didn't analyse them this wise.


- Elias


IMHO a large part of those short delay reflections are HOMs.

Inversely to the common knowledge waveguide don't have less HOMs than horns. What an OSwaveguide do better than horns is that it uses the HOMs in order to reflect them in various directions in order to obtain a more evenly distributed directivity.

To confirm or not those yellow-orange stripes are du to HOMs we should analyse with the same method the impulse response at different off-axis angles.
 
Thanks Jean Michel! I was having a hard time with that.
It looks better in FR and distortion than it does in the wavelet.

Which Pioneer horn is this (TH4001?)

The only markings are on the drivers. PM-40-1
Horn is bent 90 degs about 70mm from the driver. Small throat, only 16mm
Horn mouth is 31.5mm wide x 14.5 tall.
Driver has the female thread, not the horn.
 
Michael,

It seems that you haven't gated ghost impulse response. Ghosts could happen with MLS stimuli.


...might be – but leave those "time machine ghost pattern" aside and have a look at my worst piece of honker (0 deg horn with no round over = quarter wave tube of 20cm length) there:

http://www.diyaudio.com/forums/multi-way/161627-horn-honk-wanted-8.html#post2128416


163776d1269333291-horn-honk-wanted-wave_jmlc.gif


There should be no horn performing any worse as above !?



The OS wavelets here

http://www.diyaudio.com/forums/multi-way/161627-horn-honk-wanted-7.html#post2125170

and here

http://www.diyaudio.com/forums/multi-way/161627-horn-honk-wanted-7.html#post2125178

do not show such extended reflections, especially in the upper frequency range - as seen in your measurements - either....


Michael
 
Last edited:
Status
Not open for further replies.