Wow, I thank you!!!
That PDF of the meaurements is simply the best document I have seen in a long time wrt horn measurements.
I wish some of the "American" horns could have been tested, QSC HPR-152i, OSWG 12" and 15"
Soongsc,
A fine analysis of the reflection around 3ms will show that the arrival time differs from horn to horn (depite the fact that the mouth and the microphone were always at fixed positions). A better correlation for the arrival time of theat reflection seen on the wavelets graph is with the double of the distance from throat to mouth.
In order to reduce at minimum the effect of the floor refelection,for all the measurement in the report I used to cover the floor with an acoustic foam which is very efficient over 1000Hz (nearly 100% absorption at 1000Hz, something around 60% at 500Hz).
First reflection I could identify as coming from the room was over 7,5milliseconds and this is outside the window analysis..
Best regards from Paris, France
Jean-Michel Le Cléac'h
A fine analysis of the reflection around 3ms will show that the arrival time differs from horn to horn (depite the fact that the mouth and the microphone were always at fixed positions). A better correlation for the arrival time of theat reflection seen on the wavelets graph is with the double of the distance from throat to mouth.
In order to reduce at minimum the effect of the floor refelection,for all the measurement in the report I used to cover the floor with an acoustic foam which is very efficient over 1000Hz (nearly 100% absorption at 1000Hz, something around 60% at 500Hz).
First reflection I could identify as coming from the room was over 7,5milliseconds and this is outside the window analysis..
Best regards from Paris, France
Jean-Michel Le Cléac'h
Last edited:
Hello Doug,
Thanks for the appreciation about my report!
You wrote:
Me too, but I was very surprised nobody brought such waveguides... (I have measurements of OS Waveguide done apart of ETF2010, but that's another story...).
May be the rason why OSwaveguides were not presented to ETF2010's "shoot out" is because the organizing comitee (of which I am not member) fixed the lower limit for the effiency to 100dB/1W/1m. For sure the needed equalization of the OSWaveguide is a bit difficult to perfom passively but I think that it was possible to obtain both a good passive equalization and the required minimum efficiency.
(as an example : the efficiency of the couple TAD TD2001 + Musique concrête J321 Le Cléac'h horn was downgraded to 100dB/1W/1m by the use of a serial resistor...)
Best regards from Paris, France
Jean-Michel Le Cléac'h
Thanks for the appreciation about my report!
You wrote:
Me too, but I was very surprised nobody brought such waveguides... (I have measurements of OS Waveguide done apart of ETF2010, but that's another story...).
May be the rason why OSwaveguides were not presented to ETF2010's "shoot out" is because the organizing comitee (of which I am not member) fixed the lower limit for the effiency to 100dB/1W/1m. For sure the needed equalization of the OSWaveguide is a bit difficult to perfom passively but I think that it was possible to obtain both a good passive equalization and the required minimum efficiency.
(as an example : the efficiency of the couple TAD TD2001 + Musique concrête J321 Le Cléac'h horn was downgraded to 100dB/1W/1m by the use of a serial resistor...)
Best regards from Paris, France
Jean-Michel Le Cléac'h
I have stopped using foam to absorb high frequency data because it does not allow me to see the reflection, yet the low frequency is still effected, which was not so good for my measurements. Even like this, I can get usefull data above 400Hz when the driver is positioned 1.2M above the floor and mic position 1M from the diaphragm.
The path difference of floor reflection is based on the acoustic center, where is this acoustic center for a compression driver on a horn, I do not know. But for a direct radiating driver with a wave guide, my data show it's at the diaphragm. Since I do not know the exact dimensions of the driver and horn for your case, my assumption would be the floor. However, it would be interesting if you could use your software to analyze one of my wave guide impulse to see if it correctly identifies the depth of the device.
Hello,
I am very happy with the measurement as performed with the acoustic foam on the floor.
From the practical point of view they fulfill my requirements for good absorption of the reflected wave above 500Hz and easy placement of the horns (when you have 16 to manipulate, that's a task!)
Please remember that the 16 horns are only analyzed in the frequency range between 200Hz and 24kHz and the cut-off ferquency of the majority of them is above 400Hz. Things are different if you have to measure a bass enclosure as in this case the acoustic foam has nearly no efficiency in absorbing such low frequencies.
About what you said about the delay between a direct wave and a wave having been reflected on the floor . The analyzed horns have a non negligeible length thus there cannot be any direct reflection on floor from the acoustic center. To be possibly reflected from the floor, the wave has first to travel to the mouth. This is simple geometry. See attached graph.
As the mouth position is fixed and such is the microphone position there is virtually no possibility for the arrival time of the wave reflected by the floor to be dependant on the length of the horn. The arrival time should not differ from one horn to the other. This is not the case as we can see that the arrival time on the spectrogram varies from horn to horn inside an interval 2.5ms - 3.5ms. This varying arrival time has to be interpreted as due to the wave reflected from the mouth to the throat and interferring with the direct wave. There is more evidence of that when we can see that several horns have a high energy level for that reflected wave at high frequency (frequency above 2kHz, e.g. N°5) for which the absorption by the acoustic foam is very efficient.
For reflected waves from mouth to throat here is the equivalence between the arrival time and the distance from diaphragm to the discontinuity creating the reflection (generally a truncated mouth).
____t_________d
___2ms______34,4cm
___2,5ms____43cm
___3ms______51,6cm
___3,5ms____60,2cm
Best regards from Paris, France.
Jean-Michel Le Cléac'h
I am very happy with the measurement as performed with the acoustic foam on the floor.
From the practical point of view they fulfill my requirements for good absorption of the reflected wave above 500Hz and easy placement of the horns (when you have 16 to manipulate, that's a task!)
Please remember that the 16 horns are only analyzed in the frequency range between 200Hz and 24kHz and the cut-off ferquency of the majority of them is above 400Hz. Things are different if you have to measure a bass enclosure as in this case the acoustic foam has nearly no efficiency in absorbing such low frequencies.
About what you said about the delay between a direct wave and a wave having been reflected on the floor . The analyzed horns have a non negligeible length thus there cannot be any direct reflection on floor from the acoustic center. To be possibly reflected from the floor, the wave has first to travel to the mouth. This is simple geometry. See attached graph.
As the mouth position is fixed and such is the microphone position there is virtually no possibility for the arrival time of the wave reflected by the floor to be dependant on the length of the horn. The arrival time should not differ from one horn to the other. This is not the case as we can see that the arrival time on the spectrogram varies from horn to horn inside an interval 2.5ms - 3.5ms. This varying arrival time has to be interpreted as due to the wave reflected from the mouth to the throat and interferring with the direct wave. There is more evidence of that when we can see that several horns have a high energy level for that reflected wave at high frequency (frequency above 2kHz, e.g. N°5) for which the absorption by the acoustic foam is very efficient.
For reflected waves from mouth to throat here is the equivalence between the arrival time and the distance from diaphragm to the discontinuity creating the reflection (generally a truncated mouth).
____t_________d
___2ms______34,4cm
___2,5ms____43cm
___3ms______51,6cm
___3,5ms____60,2cm
Best regards from Paris, France.
Jean-Michel Le Cléac'h
Attachments
Last edited:
hello Soongsc,
The table on top of which was placed the measured horn is a kind of classroom table. Several of those are visible in the background of the picture.
http://www.melaudia.net/zfoto/etf10/concours/IMG_4338-800x533.jpg
The height of those tables is 68 centimeters. Depending on the diameter of the horns the axis of the horn was at a height comprised in the interval from 75 to 85 centimeters. The horizontal distance from mouth to the capsule of the microphone was fixed at 1.40meter (the SPLmeter was at 1.5 meter during the setting of the reference acoustic level on filtered noise)
This means that, the difference of pahlength between the direct wave and a wave reflected by the floor is comprised between 62centimeters and 80centimeters. This means also that the delay of the wave reflected by the floor versus the direct wave is comprised between: 1.81millisecond and 2.33millisecond. This doesn't fit with the mean value of 3 milliseconds you can see for what I considered as reflected waves from mouth to throat on the spectrograms. The difference of path between such reflected wave by the mouth to the throat and the direct wave being the double of the length of the horn (whatever length means in that case).
Best regards from Paris, France
Jean-Michel Le Cléac'h
The table on top of which was placed the measured horn is a kind of classroom table. Several of those are visible in the background of the picture.
http://www.melaudia.net/zfoto/etf10/concours/IMG_4338-800x533.jpg
The height of those tables is 68 centimeters. Depending on the diameter of the horns the axis of the horn was at a height comprised in the interval from 75 to 85 centimeters. The horizontal distance from mouth to the capsule of the microphone was fixed at 1.40meter (the SPLmeter was at 1.5 meter during the setting of the reference acoustic level on filtered noise)
This means that, the difference of pahlength between the direct wave and a wave reflected by the floor is comprised between 62centimeters and 80centimeters. This means also that the delay of the wave reflected by the floor versus the direct wave is comprised between: 1.81millisecond and 2.33millisecond. This doesn't fit with the mean value of 3 milliseconds you can see for what I considered as reflected waves from mouth to throat on the spectrograms. The difference of path between such reflected wave by the mouth to the throat and the direct wave being the double of the length of the horn (whatever length means in that case).
Best regards from Paris, France
Jean-Michel Le Cléac'h
Last edited:
You're Welcome!
..it's also a fairly easy area of experimentation with normal dynamic drivers (cones). Note though that some drivers are more immune to this than others.. but generally when you start dropping mms (moving mass) for a given sd (surface diameter) - then it tends to make a greater difference.
Scan Speak has experimented with this for quite some time now. In particular their tweeter range (which extended into the Vifa & Peerless line), had a broad range of *very* similar tweeters. As the tweeters got more expensive, *less* damping was used on most diaphragms, (though with a greater emphasis on specific "spot" damping), & rear chamber "flow" is no longer a simple chamber stuffed full of absorption material. (..look to the CSD plots on Zaphs "Tweeter mishmash", specifically the Scan Speak 6600 and the Vifa DX25.. also see dlr's look at the construction of the Scan Speak.)
I'd like to see this "carry-over" into compression drivers, but haven't yet. The major differences with most traditional compression drivers relates to the phase-plug and exit angle, and the surround. The surround choices are almost exclusively chosen for extending the upper freq. response, not for improved low level detail (..and the same is true for the phase plug).
This is primarily the reason that I prefer larger format compression drivers when compared to their smaller 1" exit variety. Basically the larger the diaphragm, the less damping. Sure, it ends up with a lower freq. "break-up" region as a result, but freq.s *below" this region (usually an octave plus below), tend to sound more "alive" with better low level detail (depending on the diaphragm material and surround, and if ferofluid is used or not).
We seem to be partly on the same path - partly different.
IMO there is no correlation with mms per se.
There might be a correlation as when throwing together a chassis and mms got changed possibly also the spider gets changed.
As I lined out on several occasions there is a difference in the kind of dampening that actually is involved. Mechanic friction being a bad thing here regarding low level accuracy.
In short it pushes up higher order distortion the lower SPL (excursion) gets.
From that underlaying principle I would not agree that the "resistance" in CD phase plugs has any correlation to those ill dampening characteristic as its more or less purely velocity dependent.
Michael
Now I see. Thank you.
I did not think reflection from the mouth could be so bad because I have not seen this in my measurements. However, in some analysis, I do see things like this in larger horns/waveguides, and scratching my head trying to figure out what causes this. It does not seem to be the lip only, but rather a combination of curvature and lip design.
I do believe the Scanspeak and Vifa drivers could be improved more. But it seems understanding of where the energy is stored is not completely understood. The back chamber is one place, however, looking at Zaph's site shows they are not the most critical place.You're Welcome!
..it's also a fairly easy area of experimentation with normal dynamic drivers (cones). Note though that some drivers are more immune to this than others.. but generally when you start dropping mms (moving mass) for a given sd (surface diameter) - then it tends to make a greater difference.
Scan Speak has experimented with this for quite some time now. In particular their tweeter range (which extended into the Vifa & Peerless line), had a broad range of *very* similar tweeters. As the tweeters got more expensive, *less* damping was used on most diaphragms, (though with a greater emphasis on specific "spot" damping), & rear chamber "flow" is no longer a simple chamber stuffed full of absorption material. (..look to the CSD plots on Zaphs "Tweeter mishmash", specifically the Scan Speak 6600 and the Vifa DX25.. also see dlr's look at the construction of the Scan Speak.)
I'd like to see this "carry-over" into compression drivers, but haven't yet. The major differences with most traditional compression drivers relates to the phase-plug and exit angle, and the surround. The surround choices are almost exclusively chosen for extending the upper freq. response, not for improved low level detail (..and the same is true for the phase plug).
This is primarily the reason that I prefer larger format compression drivers when compared to their smaller 1" exit variety. Basically the larger the diaphragm, the less damping. Sure, it ends up with a lower freq. "break-up" region as a result, but freq.s *below" this region (usually an octave plus below), tend to sound more "alive" with better low level detail (depending on the diaphragm material and surround, and if ferofluid is used or not).
See 3" speaker
An externally hosted image should be here but it was not working when we last tested it.
and ribbon tweeter
An externally hosted image should be here but it was not working when we last tested it.
These are by no means the best, but they do stand out in CSD.
All drivers store and release energy.
If you look at the CSD plots of these drivers (in excellent horns/waveguides), then the linear decay can be pretty well controlled.
The problem is that "well controlled" and linear on a standard test signal may not necessarily be accurate (or more accurate) for a complex transient signal.
I've actually heard what you are referring to. At low freq.s hall-sound becomes less well defined with a substantial increase in non-linear distortion.
Not so much of a problem at higher freq.s however (or those in the fundamental range). There it's more of a coloration or "imprint" on the sound, and can make the presentation "louder" - having an "expander-like" quality.
On the other hand I've experienced many times where lowering mass tends to allow for better low-level detail. The "flip-side" however is that again, these designs tend to be a bit more "messy" with their decay structure, and often shift modes at higher spl.s because of their lack of damping - and when doing so "crap-out" on their previously good low-level detail. (..many "fullrange" drivers have precisely this problem.)
Linear decay over the used spectrum produces a balanced sound. That's all the benefit I can see, and it also explains why different drivers with very similar SPL can sound very different.All drivers store and release energy.
If you look at the CSD plots of these drivers (in excellent horns/waveguides), then the linear decay can be pretty well controlled.
The problem is that "well controlled" and linear on a standard test signal may not necessarily be accurate (or more accurate) for a complex transient signal.
The less energy stored, the better.
Hello,
Nice evaluation.
My criticism is about the selection of the reference (page 6 in the document), which appears to be Hornresp prediction of J321 horn:
Shouldn't the reference be ideal impulse instead?
General comment can be made all the included horns suck in terms of time-frequency energy spread, except the J321 with TAD TD2001 which was already found to perform well in this thread earlier.
- Elias
Nice evaluation.
My criticism is about the selection of the reference (page 6 in the document), which appears to be Hornresp prediction of J321 horn:
An externally hosted image should be here but it was not working when we last tested it.
Shouldn't the reference be ideal impulse instead?
General comment can be made all the included horns suck in terms of time-frequency energy spread, except the J321 with TAD TD2001 which was already found to perform well in this thread earlier.
- Elias
Hello Elias,
As the increase of group delay near the cut-off frequency is a common feature to all the horns measured at ETF2010 and, as this feature is not displayed on the spectrogram of a Dirac, I prefered to choose as a reference, the spectrogram of the impulse of a theorical horn.
Best regards from Paris, France
Jean-Michel Le Cléac'h
As the increase of group delay near the cut-off frequency is a common feature to all the horns measured at ETF2010 and, as this feature is not displayed on the spectrogram of a Dirac, I prefered to choose as a reference, the spectrogram of the impulse of a theorical horn.
Best regards from Paris, France
Jean-Michel Le Cléac'h
Hello,
Nice evaluation.
My criticism is about the selection of the reference (page 6 in the document), which appears to be Hornresp prediction of J321 horn:
Shouldn't the reference be ideal impulse instead?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.