Strangly enough, I would actually agree with Jean-Michel's analysis for the most part. I did not respond to Soongsc because I had defined CD before, he just didn't listen.
The range of CD is correct as Jean-Michel says, but another way to look at it is that the frequency responses at off-axis positions have to be parallel. there is no problem if the levels fall slightly, in fact this is desirable from my perspective, but they must be parallel. Once the level is down by 6 to 10 dB then there is no longer any significant coverage and I would consider that point to define the Coverage Angle, 45 degress as pointed out. The low frequency point on the graph is determined by the size of the waveguide and will move lower for a larger device. The angular dependence is a function of the waveguide wall angles and can be set at will. Theoretically there is no HF limitation for the waveguide, but the driver always imposes one.
The range of CD is correct as Jean-Michel says, but another way to look at it is that the frequency responses at off-axis positions have to be parallel. there is no problem if the levels fall slightly, in fact this is desirable from my perspective, but they must be parallel. Once the level is down by 6 to 10 dB then there is no longer any significant coverage and I would consider that point to define the Coverage Angle, 45 degress as pointed out. The low frequency point on the graph is determined by the size of the waveguide and will move lower for a larger device. The angular dependence is a function of the waveguide wall angles and can be set at will. Theoretically there is no HF limitation for the waveguide, but the driver always imposes one.
Hello Earl,
Here it is:
http://www.diyaudio.com/forums/attachment.php?s=&postid=1518917&stamp=1211617551
obtained by Jzagaja from a pulse measurement you posted yourself on DIYaudio.
see post:
http://www.diyaudio.com/forums/showthread.php?postid=1518917#post1518917
Best regards from Paris, France
Jean-Michel Le Cléac'h
Here it is:
http://www.diyaudio.com/forums/attachment.php?s=&postid=1518917&stamp=1211617551
obtained by Jzagaja from a pulse measurement you posted yourself on DIYaudio.
see post:
http://www.diyaudio.com/forums/showthread.php?postid=1518917#post1518917
Best regards from Paris, France
Jean-Michel Le Cléac'h
gedlee said:
You will notice that you are once again taking the post out of context. Those results were contaminated by noise as shown by the data itself. They ARE NOT valid CSD. I don't pay much attention to the noise floor in general because I don't routinely look at CSD and for what I do the noise floor is not an issue. The top one is reasonably representative, and quite good, the bottom one is all noise after the intial decay. This is an UnEQ'd response and the only significant "tail" in the top trace is the drivers resonance. It's not due to the waveguide. An EQ'd waveguide would not have any sigificant "tails".
I can never seem to figure out if you get things all messed up on purpose, or is it just your nature.
I can never seem to figure out if you get things all messed up on purpose, or is it just your nature.
Hello Earl,
In the previously mentionned thread you gave 3 different sets of data from your own made measurements.
In a recent post
http://www.diyaudio.com/forums/showthread.php?postid=1777712#post1777712
you told us:
"I don't think that anyone is questioning my ability to perform the tests."
Now you are saying that your own measurements were noised (so why did you gave them...) .
Well, this is a bit contradictory...
Best regards from Paris, France
an have a good week end everyone!
Jean-Michel Le Cléac'h
In the previously mentionned thread you gave 3 different sets of data from your own made measurements.
In a recent post
http://www.diyaudio.com/forums/showthread.php?postid=1777712#post1777712
you told us:
"I don't think that anyone is questioning my ability to perform the tests."
Now you are saying that your own measurements were noised (so why did you gave them...) .
Well, this is a bit contradictory...
Best regards from Paris, France
an have a good week end everyone!
Jean-Michel Le Cléac'h
gedlee said:
What I'm asking for is an industry wide definition in a formal document, not some personal interpretation. For example, if one uses the term "Kona coffee" it means all of the coffee bean must be from Kona, otherwise it can atmost be labelled as "Kona Blend", etc. This is a clear line of definition. Obviously this sort of definition is non-existent when we talk about CD.gedlee said:
Let's assume that we understand the definition as you explained here, is this attached graph considered CD when equalized? Are the lines parallel enough? Can any of the lines intersect?
BTW, the throat size limits the high frequency directivity as well.
Attachments
Soongsc
I would consider that CD.
The lines can cross if they are close to the axis where the responses are quite close, but they should not cross at wider angles where the differences are larger. You are correct, there is no "standardized by the NBS" deffinition of CD, but that is beside the point. Its easy to identify and define as has been done here. I use the deffintion, a little different than Jean-Michel's above that the response needs to be +-2 dB within the window near the listening axis and that the coverage angle and frequency bandwidth is at the -6 dB points. Using the peak values in the "passband" is not as effective I don't think.
Jean-Michel
Because I don't worry about noise does not mean that I can't make noise free measurements. One only does what is necessary. I explained this at the time, but I guess you missed that part.
I don't come to your thread and harass you, why do you feel that you need to do that here?
I would consider that CD.
The lines can cross if they are close to the axis where the responses are quite close, but they should not cross at wider angles where the differences are larger. You are correct, there is no "standardized by the NBS" deffinition of CD, but that is beside the point. Its easy to identify and define as has been done here. I use the deffintion, a little different than Jean-Michel's above that the response needs to be +-2 dB within the window near the listening axis and that the coverage angle and frequency bandwidth is at the -6 dB points. Using the peak values in the "passband" is not as effective I don't think.
Jean-Michel
Because I don't worry about noise does not mean that I can't make noise free measurements. One only does what is necessary. I explained this at the time, but I guess you missed that part.
I don't come to your thread and harass you, why do you feel that you need to do that here?
I see. Because I do not consider that CD, and would only refer to the +/-7.5 deg range CD.gedlee said:
In this case, if a LeCleach expansion were combined with and OS throat, it is still possible for it to become a CD wave guide. And since Jean-Michel really refers to his method as LeCleach expansion method. I really see no conflict.
Now I might just call this the "LeCleach OS wave guide".
I would like to point out that it is the very large radius of Jean-Michels horn mouth that results in such smooth response. In theory all infinite horns and waveguides have perfectly smooth responses, there is nothing to cause otherwise. Its the mouth diffraction that is the biggest cuprite of non-ideal behavior. So if you want an improved OS waveguide, then adapt Jean-Michel's mouth profile to the OS. I use what could only be considered a minimu mouth radi simply because I have to be practical. I'm making reasonable cost and size systems and very large mouth radi are not feasible, nor is a free standing waveguide.
So, its a no-brainer to just adapt the large mouth radi, with the gradual radi change that Jean-Michel uses. I would admit that this is a better approach than the small radi that I use. The problem is that in my total systems designs I can't use mouths that large, they don't fit.
So, its a no-brainer to just adapt the large mouth radi, with the gradual radi change that Jean-Michel uses. I would admit that this is a better approach than the small radi that I use. The problem is that in my total systems designs I can't use mouths that large, they don't fit.
"Eevryone looking at the pressure field map delivered by the OS waveguide may easily be convinced that the response modifies with distance even at 2meters and more ."
It's presumptuous to speak for everyone.
In any case, what plots are you referring to, as polars don't have distance as a parameter.
It's presumptuous to speak for everyone.
In any case, what plots are you referring to, as polars don't have distance as a parameter.
From my measurements and sims, it seems the throat and driver match is the most important factor for smooth polar response. The angle right after throat diffraction effects high frequency smoothness, transition of guide wall to baffle is the third aspect. The reason why LeCleach Horns have a smooth polar response is because for the frequency range it's used, the initial angle is small, and the transition to the lip end is gradual. I think this is already evident in the pics I had previously posted.gedlee said:
This is true, which is also why simulation and analysis tools are very usefull to look at changes. I've been tweaking different parts to see how it changes the response, and I think I've got one that is worth testing, but I might tweak a bit more till the sim looks better. What I think might work is to find a trend between sim and actual measurements. And very interestingly, the shape is getting closer to an OS type expansion, although not exactly the opening as mentioned earlier in this thread. Tweaking is narrowed down to about 1~2mm difference in cross section radius.tinitus said: