In general, if the used frequency range has a similar decay rate throughout the frequency, the sound will be more balanced. Even if the SPL is relatively flat, the decay rates at different frequencies play a major role in tone balance.
Thanks Doug. No offense man but you gotta relax on the strong professing you have a tendency to do. It comes off as disrespectful. I don't want this to turn into another cable thread. This one is actually useful. If your correct or at least have strong evidence to support a claim, I don't mind confidence. I have evidence to support what I say and I'm not big on repeating it many times in just a couple posts--it's sort of holding things back. The issues go all the way out. From 0 - 90 degrees. I do think 90 may be less audible, but more audible than you think. You'd be surprised how large the difference is between:
and:
The second graph sounds better by a large margin. By your criteria it would be hard to say, but the top one would look better overall. It is the WHOLE thing. Maybe even further than that--the way it loads the room. I'm surprised you worried about things 30 dB down, but not 50 degrees off axis. All the while talking about missing importance stuff d/t data overload.😉
I'm trying to simplify by going with what's been demonstrated to be most audible. Later evidence may show flaws in my way of thinking to some degree, but going beyond what has been demonstrated is guessing. I'm just not a huge fan of guessing. There are enough issues to contend without guessing beyond those which are known. After I get those right, I'll start guessing. Soongsc has some interesting ideas on that wrt CSD.
That notch filter was just built out of spare parts, but was still an attempt to get every axis as correct as I possibly could. My biggest concern was actually 56.25-90 degrees off axis as that where the biggest issues were.😱 It makes a lot of difference subjectively as I'm sure every axis on the frontal hemi has.
Nah, you are again professing too strongly what you do not know. You can take that as being worth what you paid for it--that's how I take your claim b/c I've done the experiments(45-90). A decent response to 45 degrees isn't very difficult and I can understand the desire to make thing as simple as possible. You'd be wise to listen-- take that both to me in this case and your speakers/graphs out to 90 degrees. 180 isn't a bad idea, but probably not as important as 90. I am thinking of going a bit further just to have a look around as I think further would be better. My rig is set up for 360 degrees.
Yea, crossing lower is a better idea. Know any good, cheap CDWGs that will get down to 1kHz? That's a dream of mine to find.
My glaring crossover issues are now gone--graphically and audibly. Remember my first graph posted earlier in the thread. Dr. Geddes seems to have very little compromise in his crossovers--less than even my latest. That's my goal or better. My latest is great, but still not a Geddes. I need a woofer tester or at least a rig I'm looking to build for REW. I think I can trick it into getting all the measurements I absolutely need to improve what I'm doing. I dislike Microsoft products that much even though I have family that write it.
I see cone break up as the biggest concern for CD speaker building and CD speakers the most desirable type. Gedlee WGs on the open market would be nice. I wish he would go to Taiwan and get these things built cheaply for all to enjoy. It would solve a lot of my concerns and allow more give in woofer selection. It would probably be very beneficial for B&C--more so that for Dr. Geddes. Dr. Geddes, you've done some research with them, why not try and propose a deal for us poor people?
Dan
CSD with different window length shown side by side with respective SPL of the same window length.
An externally hosted image should be here but it was not working when we last tested it.
Nothing written up, just accomulation experiements and listening to lectures related with noise control. I think that is an area where more research in audibility and masking effects are conducted.
Again, you profess too strongly.
I certainly was talking about woofers--the break up region.
For the second half, take a look at my graphs--any of them.
Dan
BTW, I don't disagree with your take on CSD. Just on the magnitude of its importance. All things equal, I'd take the one with the better CSD. Look at Zaph's tidbits on CSD:
Zaph|Audio
interesting. Thoughts?
I can not spot the same stored energy in a FR plot. I believe what you are saying but the CSD to a very average guy like myself gives us the differences in the comparisons that we would have to be blind to not see.
The same goes for wavelets, its the first time I have ever seen tangible differences in many waveguide/CD measurements. We didnt get that with the polar plot and impulse measuremnts. If you can, congrats!
CSDs and Wavelets make it easier for people like myself.
Im not sure what you mean by "too strongly" (maybe its just the way I am with my convictions, wrong or right). Im just posting I am that I have never read anything complaints from any experts about not having 90degree measurements (especially on woofers). When Augerpro did his measurements (several of those are my drivers/waveguides), there was never a discussion over "Man I wish we had 90 degrees". I just believe you are going over board with the need for 90 degrees and if you really want it then you should start buying drivers and getting them measured to 90 degrees. Augerpro measured to what he feels is enough to chose a great speaker. You think its not a enough then by all means show us how 90 degrees allows for a better speaker selection.
I tend to spend more money on drivers so that I have little worry about being close to the breakup point (See TD12M measurements, Worth the $$$ if you can not find that cheap Waveguide/CD that plays really nice below 1K).
I know all about Zaph, Geddes and many others on what they think of CSDs. I have my own usage and so far the best sounding drivers have correlated back to my CSD comparisons. That is all I have, Im not the expert but there is not a single other measurement that has as much data in it in telling me which driver I will like. I have tried to use on and off axis FR plots, distortion, Impulse but nothing jumps out at me like CSDs do.
From my limited experience in building speakers (and reading builds from experts probably over 2000 hours a year, I tend to waste a lot of time reading and posting) I have learned one thing. There are many, many ways to build a great sounding speaker. We have to find out what fits our styles. If you choose 90 deg as a necessity that is cool, if Geddes chooses never uses a CSD or wavelet...more power too him.
Last edited:
Thanks SONGSC for the side by side comparison. Its a great example of why I like CSDs.
I couldnt pick anything out from those FR plots (first 2, the 3rd is obvious) but the CSD gives me a choice.
I couldnt pick anything out from those FR plots (first 2, the 3rd is obvious) but the CSD gives me a choice.
Last edited:
Thanks Doug. No offense man but you gotta relax on the strong professing you have a tendency to do. It comes off as disrespectful. I don't want this to turn into another cable thread. This one is actually useful. If your correct or at least have strong evidence to support a claim, I don't mind confidence. I have evidence to support what I say and I'm not big on repeating it many times in just a couple posts--it's sort of holding things back. The issues go all the way out. From 0 - 90 degrees. I do think 90 may be less audible, but more audible than you think. You'd be surprised how large the difference is between:
and:
The second graph sounds better by a large margin. By your criteria it would be hard to say, but the top one would look better overall. It is the WHOLE thing. Maybe even further than that--the way it loads the room. I'm surprised you worried about things 30 dB down, but not 50 degrees off axis. All the while talking about missing importance stuff d/t data overload.😉
I'm trying to simplify by going with what's been demonstrated to be most audible. Later evidence may show flaws in my way of thinking to some degree, but going beyond what has been demonstrated is guessing. I'm just not a huge fan of guessing. There are enough issues to contend without guessing beyond those which are known. After I get those right, I'll start guessing. Soongsc has some interesting ideas on that wrt CSD.
That notch filter was just built out of spare parts, but was still an attempt to get every axis as correct as I possibly could. My biggest concern was actually 56.25-90 degrees off axis as that where the biggest issues were.😱 It makes a lot of difference subjectively as I'm sure every axis on the frontal hemi has.
Sorry Im actually relaxed and Im not trying to be disrepectful at all. I like to get to the meat of any opinion. Im cool with being wrong but Im just a guy that runs his own business and deals with engineers all the time who have big egos (conference calls these days since Im always 1000s of miles away now). I have to be tough and strong willed to be successful and get anything done. Sorry I rub you the wrong way, it was never my intent.
Now back to the discussion.....Your example of the 90deg problem looked more like a > 40 degree problem. Meaning if you dealt with the issues at > 40 degrees then you automatically deal with the 90 degree issues. I just can not see many instances where a drive has issues at 90degrees and not at 45 degrees. Again, I have not see all drivers.
You will have to define cheap?
QSC HPR15i waveguide and the DE250 can play down pretty low.
Other waveguides like the 18sound versions with 1.5" throats play below 1KHz. They cost a little more money. See The raptor build over on Htguide.com
Last edited:
The Toole paper that Zilch posted earlier "Loudspeaker measurements and their relationship to listener preferences Part 2" (but link no longer working) does do measurements in the full 360 degrees around the loudspeaker. The comment about the 60 to 75 degree off axis measurements is enlighteing (well I thought so).
He says
Now our brains are used to dealing with reflected sounds, but what if those reflected sounds are different to the direct sounds, how does our brain interpret that??... this may give a clue to why Dan finds these measurements so important.
Another thing in the article talks about the difference between resonances and interference effects. This leads me to believe that what you are seeing Dan in your off axis measurements is more related to interference (maybe baffle diffration?) than due to inherent resonances in the woofers (if I'm interpreting what Toole is saying correctly).
Toole also uses spacial averaging where various measurements taken on an off axis are averaged together to end up with a "smoothed" curve. Any irregularities that remains are almost certainly indications of "problems".
Tony.
He says
Now our brains are used to dealing with reflected sounds, but what if those reflected sounds are different to the direct sounds, how does our brain interpret that??... this may give a clue to why Dan finds these measurements so important.
Another thing in the article talks about the difference between resonances and interference effects. This leads me to believe that what you are seeing Dan in your off axis measurements is more related to interference (maybe baffle diffration?) than due to inherent resonances in the woofers (if I'm interpreting what Toole is saying correctly).
Toole also uses spacial averaging where various measurements taken on an off axis are averaged together to end up with a "smoothed" curve. Any irregularities that remains are almost certainly indications of "problems".
Tony.
Confused- someone please explain?
Man I am confused...looking at post #262 and thought these waveguides were supposed to have constant directivity? (waveguides right? If that is wrong please skip the rest of this post and define the transducer system type.) These graphs are not constant directivity past the top three curves. It is nice the transducer has similar frequency response at the different angles but that is not constant directivity. Normally beamwidth is defined as the point on the polar radiation curve where the signal is -6dB from the on axis signal and maintaining a single phase. In the top graph of post #262 that includes about the top three ( I am assuming single phases but that has not been shown) curves at 4kHz and most all the curves at 400Hz. My questions- are these curves supposed to represent some constant directivity transducer system? If this is the "famous waveguide" is this a typical angular radiation pattern of such a waveguide? If this is the angular radiation pattern of the "famous waveguide" then I am clearly in the wrong place as fitting the definition of constant directivity only includes the top three curves of the graph which are what 0,10,20 degrees referenced to the axis at 0 degrees? Is that right, constant directivity out to 20 degrees off axis? Fits constant directivity to 20 degrees off axis (polarity is assumed to remain constant but not shown) in the broadest sense of -6dB, so what? Or is that wrong, are the steps in angle bigger than 10 degrees? Was the point of the angular graphs supposed to show there is no side lobes generated by this device? The graphs do not prove that at all so if that was the intent these graphs do nothing to indicate a no lobe condition. Is the intent to show at the measured angles the frequency response is similar? This is somewhat true.
The tweeter I use at 0, 10, 20, with 30 degrees being -3dB has the same graphs with the curves, for all practical purposes, laying right on top of each other and within the total included angle of 60 degrees being -3dB at the edge. The tweeter has constant directivity and constant phase (I know others discount phase but phase is still number one here and a requirement to meet the definition of constant directivity. My antenna designer background is showing!) within the included solid conical angle of 60 degrees. Radiation beyond 30 degrees off axis was not deemed important so has not been measured therefore cannot speak to those angle greater than 30 degrees off axis. We expected minor lobe boundaries to appear somewhere past 30 degrees and because most of the energy was in the main lobe it was a case of "who cares past 30 degrees." Also this tweeter is much smoother in its frequency response and stays in the same phase all the way to above 20kHz. It requires +4dB at 20kHz to be a flat frequency response as EQ is required by any constant directivity device resulting in +/-2dB from 2000-above 20kHz so is much "flatter" than the graphs of post #262. A much better transducer for my list of important items and easily fitting within the definition of constant directivity to 30 degrees off axis.
Have I fooled myself again..... thinking that I was following some meaningful discussion about transducers and then only to find this is just another discussion which has turned into a discussion of a joke high frequency device as almost all (99.9%) "tweeters" (1000Hz-20kHz) are? Ouch, that hurts to have wasted that time fooling myself with a joke device. Or I hope that I am just confused and all this represents something else that is worth learning about.
Would someone please take the short time to clear this up.😕 That would be so kind and greatly appreciated.🙂🙂
=SUM
Man I am confused...looking at post #262 and thought these waveguides were supposed to have constant directivity? (waveguides right? If that is wrong please skip the rest of this post and define the transducer system type.) These graphs are not constant directivity past the top three curves. It is nice the transducer has similar frequency response at the different angles but that is not constant directivity. Normally beamwidth is defined as the point on the polar radiation curve where the signal is -6dB from the on axis signal and maintaining a single phase. In the top graph of post #262 that includes about the top three ( I am assuming single phases but that has not been shown) curves at 4kHz and most all the curves at 400Hz. My questions- are these curves supposed to represent some constant directivity transducer system? If this is the "famous waveguide" is this a typical angular radiation pattern of such a waveguide? If this is the angular radiation pattern of the "famous waveguide" then I am clearly in the wrong place as fitting the definition of constant directivity only includes the top three curves of the graph which are what 0,10,20 degrees referenced to the axis at 0 degrees? Is that right, constant directivity out to 20 degrees off axis? Fits constant directivity to 20 degrees off axis (polarity is assumed to remain constant but not shown) in the broadest sense of -6dB, so what? Or is that wrong, are the steps in angle bigger than 10 degrees? Was the point of the angular graphs supposed to show there is no side lobes generated by this device? The graphs do not prove that at all so if that was the intent these graphs do nothing to indicate a no lobe condition. Is the intent to show at the measured angles the frequency response is similar? This is somewhat true.
The tweeter I use at 0, 10, 20, with 30 degrees being -3dB has the same graphs with the curves, for all practical purposes, laying right on top of each other and within the total included angle of 60 degrees being -3dB at the edge. The tweeter has constant directivity and constant phase (I know others discount phase but phase is still number one here and a requirement to meet the definition of constant directivity. My antenna designer background is showing!) within the included solid conical angle of 60 degrees. Radiation beyond 30 degrees off axis was not deemed important so has not been measured therefore cannot speak to those angle greater than 30 degrees off axis. We expected minor lobe boundaries to appear somewhere past 30 degrees and because most of the energy was in the main lobe it was a case of "who cares past 30 degrees." Also this tweeter is much smoother in its frequency response and stays in the same phase all the way to above 20kHz. It requires +4dB at 20kHz to be a flat frequency response as EQ is required by any constant directivity device resulting in +/-2dB from 2000-above 20kHz so is much "flatter" than the graphs of post #262. A much better transducer for my list of important items and easily fitting within the definition of constant directivity to 30 degrees off axis.
Have I fooled myself again..... thinking that I was following some meaningful discussion about transducers and then only to find this is just another discussion which has turned into a discussion of a joke high frequency device as almost all (99.9%) "tweeters" (1000Hz-20kHz) are? Ouch, that hurts to have wasted that time fooling myself with a joke device. Or I hope that I am just confused and all this represents something else that is worth learning about.
Would someone please take the short time to clear this up.😕 That would be so kind and greatly appreciated.🙂🙂
=SUM
Last edited:
As always great stuff from Toole.
I wonder what some of his assumptions are, listening distance, speaker placement angles, etc. Also I wonder if his discussion revolves around the reflection of higher frequencies ( > 3K Hz). 20ms is a clue to their relative size. This is where we all know constant directivity plays a key role in a better in room response.
I guess 90deg on geddes speakers is going to matter since the recommended speaker placement probably creates a early reflection of the 90deg response.
I wonder what some of his assumptions are, listening distance, speaker placement angles, etc. Also I wonder if his discussion revolves around the reflection of higher frequencies ( > 3K Hz). 20ms is a clue to their relative size. This is where we all know constant directivity plays a key role in a better in room response.
I guess 90deg on geddes speakers is going to matter since the recommended speaker placement probably creates a early reflection of the 90deg response.
I find that funny because not knowing the Tweeter I would automatically say they are constant directivity because they are completely uniform from 0 to past 60degree off axis and they look like every other measurement from CD waveguides I have found over the past two years.
If they were not CD then they would have a severe high frequency roll off 30 deg and beyond.
What do you think Constant directivity should look like....look at the spacing between each measurement. Its uniform!
All the way up past 15KHz? Can you post this tweeter and its measurements up to 60degrees? I know the RAAL ribbons have incredible FR plots on and off axis ( and their CSDs rule all others! I definitely want them at some point).
Last edited:
FWIW, based on my experience (but alas, no controlled testing), I'm with Earl on CSDs. I haven't really found any correlation between CSD and sound quality. I'm sure at some level there would be (because it's of course just another way of viewing impulse response).
But the last speakers I made as a test (an omnidirectional design using a reflector) have absolutely horrid looking CSD (and rather odd impulse response, too), but sound much better and more lifelike to me than the nice linear-phase 3-Ways I made years ago that have admirably clean CSD. I've by now written five different commercial software implementations of generation and display of CSDs and have even promoted their value in the past, but have come to the conclusion that they are way over-rated. Real, but maybe not so important. Heck, to even get a decent-looking CSD you have to either be in an anechoic chamber or have to window out reflections after milliseconds of the start of the impulse response -- reflections TOTALLY destroy CSD appearance! I'm pretty sure that ears don't do this kind of windowing, so close in to the IR, in a usual home listening room.
I'd say that radiation pattern seems to me at the moment to be the most important (given that general response flatness can be dealt with by an equalizer if the off-axis isn't too far from the on-axis shape). Output level capability matters a lot, too, I think. Everything else being equal, sure I'd go with a cleaner CSD, but never is everything else equal. Probably the mood of the listener is way more important than anything, actually!
But the last speakers I made as a test (an omnidirectional design using a reflector) have absolutely horrid looking CSD (and rather odd impulse response, too), but sound much better and more lifelike to me than the nice linear-phase 3-Ways I made years ago that have admirably clean CSD. I've by now written five different commercial software implementations of generation and display of CSDs and have even promoted their value in the past, but have come to the conclusion that they are way over-rated. Real, but maybe not so important. Heck, to even get a decent-looking CSD you have to either be in an anechoic chamber or have to window out reflections after milliseconds of the start of the impulse response -- reflections TOTALLY destroy CSD appearance! I'm pretty sure that ears don't do this kind of windowing, so close in to the IR, in a usual home listening room.
I'd say that radiation pattern seems to me at the moment to be the most important (given that general response flatness can be dealt with by an equalizer if the off-axis isn't too far from the on-axis shape). Output level capability matters a lot, too, I think. Everything else being equal, sure I'd go with a cleaner CSD, but never is everything else equal. Probably the mood of the listener is way more important than anything, actually!
I haven't finished reading the whole paper yet but here is what he says about the listening room. There is a diagram, but I think posting a few excepts of text is ok within the confines of copyright, but posting reproductions of the graphics might be pushing it.
I think they pretty much concentrated on the 300 - 3000Hz range when doing the correlations between listener preferences and objective measurements.
I think that the distance was around 3M (depending on the listeners position) note also that the tests were mono. ie only one speaker, hence imaging was not a factor.
Note that the listeners all had backgrounds of serious critical listening, and only 28 out of the 42 that showed showed very low variability in their judgments were used to make the correlations, as such they do not represent the average person. It would be interesting to see how a group of average joes rated the same speakers!
Tony.
No doubt, there is no way to see a CSD and say that sounds horrible or sounds great and we all know we can also EQ raw drivers removing many of the issues too.
Using Zaph|Audio
and the "toggle link", I clicked on the SB29 vs the Peerless_810921. Both have a good reputation and both sound good to different people.
Now compare them vs the Hiquphon_OW4 which seems to be they top sounding choice from some build discussions.
Toggling the SB29 vs OW4 might have someone thinking the OW4 is a superior tweeter and it might be for someone but it does not mean the SB29 will sound bad.
There is definitely something to different decay signatures.
What you guys are saying is that all of these comparisons are meaningless because the comparisons can be seen in other measurements?
I would like to add that this is all just raw driver talk. In the end people can take rather simple, low cost drivers that does not have the "subjectively best" measurements and make an incredible sounding speaker design. Others can buy the best drivers possible and still have a crap sounding speaker. This is where the true speaker building skills reside.
- Status
- Not open for further replies.