Yes, you are right, I wonder why it did that. I do know the felt dropped the SPL by 5dB (the measurements showed this).
I will real post it....good catch 😀
Hello jzagaja,
Thanks for the measurements! I'll check out the IR files.
Where I can find mechanical specifications for STX-D1300-Ti and STX D800-Ti
I see stx.pl but didn't find??
- Elias
Thanks for the measurements! I'll check out the IR files.
Where I can find mechanical specifications for STX-D1300-Ti and STX D800-Ti
I see stx.pl but didn't find??
- Elias
The link there seems to be broken - there's a mini-paper available here:
http://www.diyaudio.com/forums/multi-way/103872-geddes-waveguides-74.html#post2018399
http://www.diyaudio.com/forums/multi-way/103872-geddes-waveguides-74.html#post2018399
Catapult
the excess phase plots might be a good tool to give a quick first check - would have to run several IR files to get a feeling.
...in the end we need some more precise tools I'm afraid...
Michael
the excess phase plots might be a good tool to give a quick first check - would have to run several IR files to get a feeling.
...in the end we need some more precise tools I'm afraid...
Michael
Hi,
AES paper on wavelets, also talking about horns:
http://www.audiomatica.it/download/audaesny2007.pdf
"Compression Driver on a Constant Directiv-
ity Diffraction Horn
...
The sound energy that reaches the diffraction slot is
not radiated completely into the conical section. A
portion of it is reflected back to the driver throat be-
cause of the abrupt change in the surface shape that
takes place at the diffraction point. This generates
reflected waves. The Wavelet Analysis applied to
the impulse response of such an horn can show how
much energy is reflected back and forward inside of
the horn and which are the frequency bands essen-
tially affected by the phenomenon for that specific
horn."
= Horn Honk 😀
- Elias
AES paper on wavelets, also talking about horns:
http://www.audiomatica.it/download/audaesny2007.pdf
"Compression Driver on a Constant Directiv-
ity Diffraction Horn
...
The sound energy that reaches the diffraction slot is
not radiated completely into the conical section. A
portion of it is reflected back to the driver throat be-
cause of the abrupt change in the surface shape that
takes place at the diffraction point. This generates
reflected waves. The Wavelet Analysis applied to
the impulse response of such an horn can show how
much energy is reflected back and forward inside of
the horn and which are the frequency bands essen-
tially affected by the phenomenon for that specific
horn."
= Horn Honk 😀
- Elias
Not always. There are some that don't honk but also use a difraction slot. It also has to do with how abrupt the transition is in the slot. First generation horns used abrupt transitions with sharp angles. The newer ones don't do that but use a gradual transition leading up to and through the slot.
Even more significant is some of the worst honkers out there, like some of the older exponentials a JBL 2307 is a fine example can honk terribly depending on level and crossover points. All of these older honkers don't even have difraction slots as they are used in the more modern CD type horns.
It's not the slot alone that is causing the problem
Rob🙂
Attachments
Last edited:
I thought I'd post some ETC simulation comparisons possibly associated with honk.
BTW, diffraction slots don't cause honk, but the distance between diaphragm and diffraction slot might, as will the flare rate and it's association with termination at the lip.
BTW, diffraction slots don't cause honk, but the distance between diaphragm and diffraction slot might, as will the flare rate and it's association with termination at the lip.
Attachments
Last edited:
D1300 was made for me only. D800 (price $33):
D-800-8-Ti - STX Polski producent - kino domowe g?o?niki akcesoria
I've cut the thread and drill M4 mounting threads.
Hello Zilchlab,
Let's try to see how to interpret the wavelet graph you extracted from Elias' document.
First, this graph should be analysed only over 450Hz which is the frequency at which appears the maximum of the group delay curve corresponding here to the frequency below which the acoustical impedance will be purely reactive.
(for the measurement of the relative delays, I'll take t=2.5 as the origin of time and give the delays from that origin).
In the interval of frequency between 450 and 800Hz:
- the main phenomenon is the group delay rise (+ 1,5ms) from 800 to 450Hz. It is due probably to the increase in the ratio between acoustic reactance and acoustic resistance. For a complex tone, if a fundamental has a frequency below 800Hz, it will arrive later than its harmonics with a change in the waveform and a more or less audible alteration of the timber.
In the interval of frequency between 700 and 6000Hz:
- the wavelet graph (which is limited in amplitude to 25dB which is a bit too small IMHO) shows some reflections around 4ms between 700Hz to 800Hz
Generally if such Impulse Response measurement is done at home in a living room (e.g.) then we could interpret such large delay reflection as due to the proximity to a piece of furniture or to the floor or a wall, but if the measurement has been done inside an anechoic room then this reflection is related to the horn (if the horn is small it could be a multiple reflection).
Over 800Hz:
- we notice around 1300Hz what seems to be a resonance eventually related to a reflection having a delay around 2.5ms.
- between 2kHz and 6kHz or more a delay of 1ms probably due to some reflection inside the horn (what Michael calls "quarter wave honk")
- above 6kHz many stripes having corresponding to small delay (less than 1ms). Their origin can be reflections from throat to diaphragm due to abrupt profile or expansion variation at the throat or HOMs inside the horn.
If the origin of those small delay reflections is inside the compression driver then we can notice that the delay of the main stripe is smaller at 20kHz that at 8kHz which can eventually be interpreted as due to a bad design of the couple dipahragm + phase plug and slits.
IMHO one of the main advantages of continuous wavelets over multiresolution /CSD graphs is the ability to analyse such short delay reflections.
Best regards from Paris, France
Le Cléac'h Jean-Michel
Let's try to see how to interpret the wavelet graph you extracted from Elias' document.
First, this graph should be analysed only over 450Hz which is the frequency at which appears the maximum of the group delay curve corresponding here to the frequency below which the acoustical impedance will be purely reactive.
(for the measurement of the relative delays, I'll take t=2.5 as the origin of time and give the delays from that origin).
In the interval of frequency between 450 and 800Hz:
- the main phenomenon is the group delay rise (+ 1,5ms) from 800 to 450Hz. It is due probably to the increase in the ratio between acoustic reactance and acoustic resistance. For a complex tone, if a fundamental has a frequency below 800Hz, it will arrive later than its harmonics with a change in the waveform and a more or less audible alteration of the timber.
In the interval of frequency between 700 and 6000Hz:
- the wavelet graph (which is limited in amplitude to 25dB which is a bit too small IMHO) shows some reflections around 4ms between 700Hz to 800Hz
Generally if such Impulse Response measurement is done at home in a living room (e.g.) then we could interpret such large delay reflection as due to the proximity to a piece of furniture or to the floor or a wall, but if the measurement has been done inside an anechoic room then this reflection is related to the horn (if the horn is small it could be a multiple reflection).
Over 800Hz:
- we notice around 1300Hz what seems to be a resonance eventually related to a reflection having a delay around 2.5ms.
- between 2kHz and 6kHz or more a delay of 1ms probably due to some reflection inside the horn (what Michael calls "quarter wave honk")
- above 6kHz many stripes having corresponding to small delay (less than 1ms). Their origin can be reflections from throat to diaphragm due to abrupt profile or expansion variation at the throat or HOMs inside the horn.
If the origin of those small delay reflections is inside the compression driver then we can notice that the delay of the main stripe is smaller at 20kHz that at 8kHz which can eventually be interpreted as due to a bad design of the couple dipahragm + phase plug and slits.
IMHO one of the main advantages of continuous wavelets over multiresolution /CSD graphs is the ability to analyse such short delay reflections.
Best regards from Paris, France
Le Cléac'h Jean-Michel
Where, in Fig. 34:
Last edited:
The level range of the wavelets displayed mostly seem to show only 25db, is there any specific reason for it? Has anyone done wavelets on direct radiating drivers of comparible bandwidths?
If I crank up the gain, it's noise down at -50 dB, looks like:
Here's the full range, crossover's at ~1.5 kHz, MLS gated at 200 Hz:
Here's the impulse:
And the FR:
The waterfall:
The normalized CSD:
SO, how's it SOUND...? 😀
Attachments
Last edited:
Ah, so some little holes start to show up.If I crank up the gain, it's noise down at -50 dB, looks like:
...
SO, how's it SOUND...? 😀
Obviously where will be some notes that will be a bit pronounced. The thing is, if people like it, then that's fine.
Is it possible to show the equal level lines?
Last edited:
Nice work Zilch.
What loudspeaker do we actually look at here?
May I recommend to visualize your measurement in 40dB wavelet range too, as thisis the range most wavelet plots have been published here so far ?
Michael
What loudspeaker do we actually look at here?
May I recommend to visualize your measurement in 40dB wavelet range too, as thisis the range most wavelet plots have been published here so far ?
Michael
Not a lot of work involved -- I'm just clickin' buttons here. 😉
I hope you are sitting down:
I can do +10 to -40, but I lose the color resolution:
Here are the UN-normalized waterfall and CSD:
The wavelets are not normalized; I can do that, but all it does is bump up the falloff at 20 kHz.
There's stuff happening up at 40 - 50 kHz, but that's gotta be aliasing, because the data was acquired at 48 kHz, no?
Attachments
State your hypothesis as to where it is appearing in the wavelet and we'll test that out.
[There's a bigger sandbag, tho.... 😉 ]
Last edited:
- Status
- Not open for further replies.