JBL differs from your take on farfield monitors.
Now here is a real fact about JBL speakers, and even the studio monitors through the years "Mama always said JBL was like a box a chocolates..........."
Your guessing here because there is no data to support this position. If I were to guess, as you just have, I would "guess" the opposite.
Why can't we get "good" data and settle this issue and stop just "guessing".
Fair enough. I guess "we" (as in not me (novice)) would have to find a way to correlate measurable and audible to make the exercise quantitatively useful.
Sheldon
Scott,
Similarly the diy Synergies I have built are also the best sounding I've heard- and therange does include all that I designed myself. But then again, I've never heard real Danley Synergies, Unitys, or any of Earl's speakers, so a grain of salt is called for, of course.
Care to share any details you've found about shapes? I have found the magic of using "frustrums", but still don't know if it's better to have round holes (more area away from the corners) or oval holes (covers more of the horn length). I did find that you can significantly reduce the effect of the ports on higher frequencies by covering the holes with thin (say, T-Shirt) cotton. It of course reduces the sensitivity of the drivers feeding out through the ports, too, but there is output SPL to spare in a living room with these, anyway. The cotton also seems to smooth the response shape from the drivers feeding through it.
The effects I see from the ports (on frequencies sourced from closer to the horn apex) is most significant when measured at the "0 degree" axis. At other angles they tend to smooth out pretty well. Fortunately, as recommended with Earl's speakers, these are arranged well toed-in so the 0degree axis isn't so important.
Bill,
I still don't know if it's better to do round or oval holes as well. I tried a bunch of stuff with the following midrange drivers:
Dayton PA130
Fountek FE85
Dayton DS90
And the little 2" buyout full-range-cone tweeters.
Of those only the PA130 would probably have been a good candidate for a midrange oval port. My concern would be that it would have moved the effective entry point in the speaker forward of where I wanted it. For the smaller drivers it's just not practical to try and cut an oval slot.
For the PA130 I found that I needed less volume under the cone, so I was turning custom plugs on my lathe. That takes a while. The issue with the PA130 is it's tough to get a back enclosure on the cheap. 6" grey PVC can be turned down on a lathe to make it work...but that stuff chews through even carbide tipped turning tools.
I next tried the Fountek FE85 because of it's smoother published curves (if they can be trusted) My first try was like Danley mentions in Fig. 10 of his patent...https://docs.google.com/a/google.co...ges.storage.googleapis.com/pdfs/US8284976.pdf, except I used .5" holes which ended up being too small. It would have been an easy build though since I could use two forstner bits to drill the holes...the larger hole just wider than the surround diameter. This what doomed that approach...it set up a nasty resonance at wavelengths centered around the diameter of the hole....storing and releasing energy over the course of ~5mS or more. There were a couple of other troublesome spots I could only attribute to the thin metal cone....so I moved to the DS90's....
These work very well...went with larger holes, .875" and an aggressive asymmetrical frustrum, since the hole was off-center of the driver. I also painstakingly carved a relief for the surround using a router on a circle jig and a Dremel for the 12 midranges (and 12 woofers)
For home use the PA130's would be *COMPLETE* overkill...the DS90's are padded down pretty heavily and can still drive you out of a room with sheer output. Never strained...just crazy high output. They (and the Fountek) have the advantage that the back enclosure can be a 3" PVC end cap with a small amount of shaping on the inside to miss the frame. They are then easily hot-glued to the horn so if you ever need to get them out...you can. I have 3 access panels in each speaker to aid in assembly.
Scott
Question regarding the throat transition in the Danley SH-50
I have been wanting to hear the Danley SH-50's for quite some time now, but unfortunately I happen to live on the wrong side of the Atlantic🙁
I have question regarding the Danley SH-50's that I hope someone who have seen these very interesting speakers close by, would answer for me:
I would very much like to know whether the transition from the circular exit of the BMS 4550 compression driver to the rectangular shaped conical horn has been smoothed out?
Thanks a lot!
Best regards
Peter
I have been wanting to hear the Danley SH-50's for quite some time now, but unfortunately I happen to live on the wrong side of the Atlantic🙁
I have question regarding the Danley SH-50's that I hope someone who have seen these very interesting speakers close by, would answer for me:
I would very much like to know whether the transition from the circular exit of the BMS 4550 compression driver to the rectangular shaped conical horn has been smoothed out?
Thanks a lot!
Best regards
Peter
Hi Scott, Earl, guys
This polar discussion has opened a can of worms and complicated worms at that.
I would offer a couple thoughts.
First, our polar data isn’t taken as a marketing aid, rather, the full spherical data is taken to make an EASE model which is used un system design allowing one to find the best locations and aiming angles for various loudspeaker possibilities and also to make an Aural model where one can listen to the model if you’re sitting at location X in the room.
As a result it takes an accurate model to get an accurate prediction.
This model is based on measurements over an entire sphere, every 5 degrees and these must be done at a distance large enough to avoid the interference from the size of the system, so that the pattern shape has stopped changing. This required minimum measuring distance and the reason for following this approach is outlined here;
Far-field Criteria for Loudspeaker Balloon Data Synergetic Audio Concepts
The polar map I posted here was what our programmer came up with plugging in the CLF file which was also made from the spherical measurements. In that case, the horizontal polar which was shown was the “front half” of the radiation and was comprised of 36 of the 72 impulse response measurements that make up the horizontal plane.
Resolution, if one had more than 36 data points for the forward half, you have more spatial resolution, if you have fewer, you have less and the plotted data between the actual data points interpolated.
For the amplitude response, the amplitude is smoothed either to 1/3 or 1/6 octave bandwidth (1/6 approximating your ears Bark or critical bandwidth in the mid band resolution) I forget which is used for the clf file.
Bark scale - Wikipedia, the free encyclopedia
Earl wrote his own routine for plotting his polars but the more common type is like that shown here;
3D3A Lab at Princeton University
Shown there are measurements for a number of loudspeakers and would suggest you examine them so see what real loudspeakers done this way look like (although probably too close to the mic).
The Manger is an interesting example of having stellar performance in one domain (time) and a problem in it’s radiation pattern, up high, you can see it has some pretty strong concentric conical side lobes with collapsing angle with increasing frequency. The Manger is the best loudspeaker in the tijme domain I have ever seen, it can reproduce a square wave over a broad band BUT has high harmonic distortion as the frequency falls and off pattern radiates an interference pattern. Note the Sanders too, note how smoothly the pattern shifts from a dipole with directivity and smoothly collapses to a very narrow projection with increasing frequency.
While a couple side lobes are present up high, they are still around -12dB relative to on axis and would not be a problem (I don’t imagine) but the very narrow hf might be depending how wide your chair / couch is.
As one can see, the real loudspeakers they measured are not perfect, most have variable directivity in horizontal and most have crossovers plainly visible in the vertical because the sources are not adding into one new source.
It is the format above I asked our programmer to replicate.
The plot I included in this thread is a wider bandwidth than these (the low corner is 30Hz) and I forgot to ask him to justify the frequency plot to iso numbers but they can be read as is.
Just like these, it is normalized and each color graduation represents a 3dB change in level so one color also represents a 3dB span. Polar plots are normally polarized because the intent is to show the change in level according to angular projection and the frequency response plot shows the variation in amplitude vs frequency. . I do not have and didn’t ask for the Vertical map for the SH-50 but as it is symmetrical it is essentially the same as the H.
The reason I had asked to have that map, was to send it to the folks who supplied this set of plots.
The more common look at directivity is more to grasp how much sound goes where both on and off axis even behind. These guys had concluded high directivity made their crosstalk cancellation work more robustly (because more of direct sound reaches you compared to reflection interference).
I sent the plot to them to look at and to see if they wanted to measure a pair of SH-50’s but have not heard back.
To do a comparison closer to apples to apples compare the SH-50 map to these. While the SH-50 amplitude is smoothed to either 1/3 or 1/6 octave (I don’t recall what the standard for the CLF file is) and they don’t state the frequency resolution here but you can still get some idea how it stacks up so far as being constant directivity in H and V.
A last thought, the response curve on the data sheet is an 8 year old measurement taken in the warehouse (with the etc not quite on) which needs to be updated, I posted one here taken with ARTA from a meter in my living room which is more current.
Wow, someone from the old bass list, now that is going back a ways. That was back when 4MB of memory was a full length plug in card full of chips haha. Good your still a speaker guy!
Hey Bill, i haven't given up.
Take care all, time to clean the cat boxes ugh.
Tom
This polar discussion has opened a can of worms and complicated worms at that.
I would offer a couple thoughts.
First, our polar data isn’t taken as a marketing aid, rather, the full spherical data is taken to make an EASE model which is used un system design allowing one to find the best locations and aiming angles for various loudspeaker possibilities and also to make an Aural model where one can listen to the model if you’re sitting at location X in the room.
As a result it takes an accurate model to get an accurate prediction.
This model is based on measurements over an entire sphere, every 5 degrees and these must be done at a distance large enough to avoid the interference from the size of the system, so that the pattern shape has stopped changing. This required minimum measuring distance and the reason for following this approach is outlined here;
Far-field Criteria for Loudspeaker Balloon Data Synergetic Audio Concepts
The polar map I posted here was what our programmer came up with plugging in the CLF file which was also made from the spherical measurements. In that case, the horizontal polar which was shown was the “front half” of the radiation and was comprised of 36 of the 72 impulse response measurements that make up the horizontal plane.
Resolution, if one had more than 36 data points for the forward half, you have more spatial resolution, if you have fewer, you have less and the plotted data between the actual data points interpolated.
For the amplitude response, the amplitude is smoothed either to 1/3 or 1/6 octave bandwidth (1/6 approximating your ears Bark or critical bandwidth in the mid band resolution) I forget which is used for the clf file.
Bark scale - Wikipedia, the free encyclopedia
Earl wrote his own routine for plotting his polars but the more common type is like that shown here;
3D3A Lab at Princeton University
Shown there are measurements for a number of loudspeakers and would suggest you examine them so see what real loudspeakers done this way look like (although probably too close to the mic).
The Manger is an interesting example of having stellar performance in one domain (time) and a problem in it’s radiation pattern, up high, you can see it has some pretty strong concentric conical side lobes with collapsing angle with increasing frequency. The Manger is the best loudspeaker in the tijme domain I have ever seen, it can reproduce a square wave over a broad band BUT has high harmonic distortion as the frequency falls and off pattern radiates an interference pattern. Note the Sanders too, note how smoothly the pattern shifts from a dipole with directivity and smoothly collapses to a very narrow projection with increasing frequency.
While a couple side lobes are present up high, they are still around -12dB relative to on axis and would not be a problem (I don’t imagine) but the very narrow hf might be depending how wide your chair / couch is.
As one can see, the real loudspeakers they measured are not perfect, most have variable directivity in horizontal and most have crossovers plainly visible in the vertical because the sources are not adding into one new source.
It is the format above I asked our programmer to replicate.
The plot I included in this thread is a wider bandwidth than these (the low corner is 30Hz) and I forgot to ask him to justify the frequency plot to iso numbers but they can be read as is.
Just like these, it is normalized and each color graduation represents a 3dB change in level so one color also represents a 3dB span. Polar plots are normally polarized because the intent is to show the change in level according to angular projection and the frequency response plot shows the variation in amplitude vs frequency. . I do not have and didn’t ask for the Vertical map for the SH-50 but as it is symmetrical it is essentially the same as the H.
The reason I had asked to have that map, was to send it to the folks who supplied this set of plots.
The more common look at directivity is more to grasp how much sound goes where both on and off axis even behind. These guys had concluded high directivity made their crosstalk cancellation work more robustly (because more of direct sound reaches you compared to reflection interference).
I sent the plot to them to look at and to see if they wanted to measure a pair of SH-50’s but have not heard back.
To do a comparison closer to apples to apples compare the SH-50 map to these. While the SH-50 amplitude is smoothed to either 1/3 or 1/6 octave (I don’t recall what the standard for the CLF file is) and they don’t state the frequency resolution here but you can still get some idea how it stacks up so far as being constant directivity in H and V.
A last thought, the response curve on the data sheet is an 8 year old measurement taken in the warehouse (with the etc not quite on) which needs to be updated, I posted one here taken with ARTA from a meter in my living room which is more current.
Wow, someone from the old bass list, now that is going back a ways. That was back when 4MB of memory was a full length plug in card full of chips haha. Good your still a speaker guy!
Hey Bill, i haven't given up.
Take care all, time to clean the cat boxes ugh.
Tom
Earl,
Sometimes the DIY "amateurs" making the 🙄 claims are jaded former hifi professionals with 20 years of technical experience, multiple patents, skill and ears. Sometimes there are other people who are competent too.
If you're ever in Austin, let me know, come on over...we'll drink some wine, you can have a listen.
Scott
The point is that we all claim our speakers sound best, so what is the point of constantly saying it. Post some data and then we can make rational judgments'. I am far more interested in seeing some detailed polar maps of a synergy horn than I am of hearing one. If it measures really great then I'd like to hear. If it has serious flaws then I'm not so interested. But somehow seeing "good" data is like pulling teeth around here.
Scott - what is you like about the older plots (using a software package that is now obsolete) and the newer ones (using my own software)? Is it the one dB line? The colors? I am not sure that I understand. Since I write the graphics that I use now I can make them like the older ones if I wanted to, but:
1) one dB steps for the lines got very dense in many situations and I found 3 dB step to be more reasonable.
2) the colors were chosen after a complaint I got about the colors! I read an article about the use of color in graphs and a strong point was made about how the colors chosen should correlate with the values presented in a visual sense - high sound intensity should correlate with high visual intensity - bright colors mean high SPL, dark is low SPL.
Last edited:
Earl wrote his own routine for plotting his polars but the more common type is like that shown here;
Hi Tom
I wrote my own routines to solve the problems that your links clearly represent.
First, I do not do any interpolation in either frequency or space. Interpolation in space is problematic because a straight-line interpolation is not correct. I fix this by finding the radiation modes and then reconstructing the polar plots with an arbitrary angular resolution. Hence no interpolation.
When one normalizes the polar map data to the central axis one creates a couple of problems. First who is to say that the central axis is "golden"? Why this axis rather than any other axis? Maybe, like my designs, the listening axis is NOT the central axis - then things get all messed up just like the data shown on the Princeton site for the Nathan - there is no widening of the polar response at about 5 kHz, there is a small hole in the on axis response, hence the visual impression is totally wrong.
Another problem with on-axis normalization is that the plots no longer represent the power response as they do when there is no normalization. Again the Nathan shown there is a classic example. The small hole on axis at 5 kHz has almost no effect on the power response because it is so narrow. But when normalized it appears that there is a huge increase in the power response at 5 kHz when this is not the case. Un-normalized a visual inspection shows the power response and the Directivity Index (DI) right off. Normalized it is impossible to tell what the true power response and DI are without more data plots. I just do not like normalized data plots.
I also use a smoothing that is accurate across the hearing bandwidth based on critical bands. It is wide at LFs (about 1/3 - 1/2 octave) and very narrow, about 1/20th octave, at HFs. Read Prof. Farina's discussion of this as he is right on the money that this is really the only truly ideal way to do frequency smoothing. 1/3, 1/6 or 1/20th fixed across the band is not correct.
Fair enough. I guess "we" (as in not me (novice)) would have to find a way to correlate measurable and audible to make the exercise quantitatively useful.
Sheldon
If you believe in Toole and Olive and the JBL guys (and I completely agree for the most part), this has been done. We could argue about some of the details, but it is not as if we don't already have a high degree of correlation between measurements and subjective impressions.
The point about the synergy technique that seems to get missed IMO is this:
When the HF compression driver sends a wave down the horn, what happens when it encounters the midranges? It will diffract off of them a create HOMs and some rearward reflections. This is never going to be a good thing although it can be minimized. But since it does not exist in a singe driver waveguide it does not have to be minimized.
Also the midranges send sound back down the device towards the compression driver and reflect off of it. This too is never going to be a good thing, but could be minimized.
There is also the mouth treatment and potential diffraction from it.
As with all designs the devil is in the details and only a high resolution polar map would show if these problems are significant in any individual design or not. I have yet to see such measurements. I would love to, or better yet, I would love to do them myself, but that has not happened either. Everyone here is arguing from a "it should work great" point if view, but without real data how things actually work is just a guess.
I agree with you on those points about HOMs
In what situation do you see a frequency-constant 360° phase shift happen?
In the situation of testing to show a square wave.
Hi Scott,
This polar discussion has opened a can of worms and complicated worms at that.
Tom
Hi Tom,
Nah.. not really problems. (..or "worms". 😀 )
Just "kicking around" very particular aspects of any design that might be an issue when compared to the some of the best loudspeakers in existence on *just* that issue.
It's an unreasonably point of view in two respects:
1. You aren't making an unreasonable claim to be disputed (..or at least I don't think that's what we are talking about here), and
2. It's an extreme and "myopic" perspective that doesn't factor-in the design as a whole. It's like looking at a super model and saying: "looked at that jacked-up pinky toe on her, she's not perfect!". 😛
..BTW, the directivity plots in the Princeton website are useful for their purposes, but don't have the resolution necessary for ours.
Last edited:
Myhrrhlein
You might be able to hear a .1dB change in level under the right conditions, but you can’t hear a .1dB change somewhere in a loudspeakers frequency response.
If you are in Synaudcon, or were, write Pat about that and also ask him about how the Synergy horns work. He has measured them, listened to them and installed them.
I suppose your right, they do not preserve time as the upper and lower limits have the phase response of each section.
What they don’t have is the phase shift / group delay between the frequency ranges like one would have with normal crossovers. In other words, the acoustic phase looks like a single broad band driver.
Best,
Tom
You are right Tom, under almost all circumstances .1dB is inaudible.
However, I thought the OP was about whether your horns could be improved. I think they can. They are close to perfect, but not as perfect as they will ever be.
p.s. My synaudcon training was in 1980.
Scott - what is you like about the older plots (using a software package that is now obsolete) and the newer ones (using my own software)? Is it the one dB line? The colors? I am not sure that I understand. Since I write the graphics that I use now I can make them like the older ones if I wanted to, but:
1) one dB steps for the lines got very dense in many situations and I found 3 dB step to be more reasonable.
2) the colors were chosen after a complaint I got about the colors! I read an article about the use of color in graphs and a strong point was made about how the colors chosen should correlate with the values presented in a visual sense - high sound intensity should correlate with high visual intensity - bright colors mean high SPL, dark is low SPL.
Different Scott. 😉
I'll think about it this weekend ..looking over both carefully, and then try and make a detailed response.
When I've got it done I'll make the post in your portion of the forum (Manufacturer's section).
Last edited:
Then you have a very dead room and it will sound that way. High directivity allows you to have a live - good acoustics - room with no early reflections. Directivity does matter.
My room is fairly live, not totally dead.
The dead end eliminates early reflections.
The live end provides the diffuse sound field.
The live end is also electro-acoustically enhanced.
What I meant was, when do you have a real speaker that jumps 360° neatly enough to preserve a square wave? You seemed to be saying that the Synergy design does that. I am under the impression that the crossover shifts the phase in a usual fashion, but then the drivers are offset in space to cancel that shift and bring things back to 0°, a luxury afforded by the concept of the shared horn. I may be mistaken. I asked about it in another thread, wasn't sure I understood the answer and decided I wouldn't pester Tom about it unless I was actually going to buy a pair.In the situation of testing to show a square wave.
Hi Dumptruck,
I am not talking about the particular speakers, only testing.
I am merely pointing out that a square wave test will test for phase alignment and not time alignment.
I am not talking about the particular speakers, only testing.
I am merely pointing out that a square wave test will test for phase alignment and not time alignment.
My room is fairly live, not totally dead.
The dead end eliminates early reflections.
The live end provides the diffuse sound field.
The live end is also electro-acoustically enhanced.
For you to eliminate the early reflections from the sidewalls (I seriously doubt they can be "eliminated") the room would have to have some serious absorption very far into the room and not just a "dead end", making the room highly absorptive. Or are you forgetting about those sidewall reflections? Those are the most serious ones and they will be very difficult to eliminate without high directivity in the speakers.
Different Scott. 😉
Yeah...this thread is going to get confusing.
ScottH.
Hi Scott, Earl, guys
Wow, someone from the old bass list, now that is going back a ways. That was back when 4MB of memory was a full length plug in card full of chips haha. Good your still a speaker guy!
Take care all, time to clean the cat boxes ugh.
Tom
Tom, I must re-iterate...thank you again for providing as much information as you do on your designs.
I had some of those full length cards...and a LAUD/CALSOD setup I thought was the bomb.
And thank you for reminding me about the cat boxes before my wife does. (double ugh)
Scott Hinson
- Status
- Not open for further replies.
- Home
- Loudspeakers
- Multi-Way
- Synergy Horns. No drawbacks, no issues?