One would hope the listener doesn't sit in a null.
It's our job as speaker designers to make a speaker with wide enough coherent radiation pattern to make this possible. It is the job of the speaker owner to setup the speakers so they don't position them where nulls will be a problem. But that requires they know where the nulls are.
If the null angle is narrow, it's easy for a listener to get outside the pattern, just by moving from seated to standing. If the pattern is very narrow, then close listeners may not even have to move very much to get outside the pattern. It could be the difference between sitting on the floor or sitting in a chair, or perhaps between different chair heights.
What are the null angles on your speakers? I don't see that listed anywhere, either by description or by published response curves. I would hope they would be at least +/- 20 degrees above and below the forward axis, but I think the distance between drivers will make this impossible if you crossover at a frequency to match the horizontal pattern. Would it be possible to publish vertical off-axis charts, at least up to the null angle?
It's our job as speaker designers to make a speaker with wide enough coherent radiation pattern to make this possible. It is the job of the speaker owner to setup the speakers so they don't position them where nulls will be a problem. But that requires they know where the nulls are.
If the null angle is narrow, it's easy for a listener to get outside the pattern, just by moving from seated to standing. If the pattern is very narrow, then close listeners may not even have to move very much to get outside the pattern. It could be the difference between sitting on the floor or sitting in a chair, or perhaps between different chair heights.
What are the null angles on your speakers? I don't see that listed anywhere, either by description or by published response curves. I would hope they would be at least +/- 20 degrees above and below the forward axis, but I think the distance between drivers will make this impossible if you crossover at a frequency to match the horizontal pattern. Would it be possible to publish vertical off-axis charts, at least up to the null angle?
gedlee said:
A few years back at CES I had lunch with Tony Grimani, who formerly worked at both Dobly and THX. I had asked his thoughts on surround speakers, and his response pointed to the dipole as more of an available solution rather than a preferred device. He noted that the target set was actually to achieve a -3dB power response (direct vs. total).
Given the factors which dominate subjective imaging, this makes more sense to me. We also should realize that part of the problem they were trying to solve was to get similar sound levels to multiple listeners, and using a large reverberant/ambient field is one way to do it, albeit in a rather crude/non-specific manner. Some more creative consideration of directivity and aiming of devices can achieve better results while still maintaining a direct sound.
While most of the attention in the original surround sound systems was given to level matching (it's also easier), I have found delay settings to have much greater impact on the perceived sound field & surround imaging. This should be of no surprise, but if we consider the loudspeaker problem further, then this would imply that the relative phase response/group delay of the speakers should matter equal or more so than the frequency response for those cases where the two aren't directly related (as in different driver arrangements).
Anyone who has ever tried to measure a dipole in real rooms has found that the only way to look at the response is with significant smoothing on an RTA, else small movements make for big changes in frequency response. Hardly the ideal we hold for our main speakers.
ttan98 said:
The center of a null is marked by a very sharp notch at a specific frequency surrounded by a wider region of reduced amplitude through the crossover overlap range.
As you pass through the null, moving out of the coherent pattern, what you hear is a sort of phasey sound. If you sit right in it, you'll notice a bit of reduced output in the crossover region and a general discontinuity of the sound.
The amount of these effects depends on the speaker, but it is definitely a large anomaly that overshadows most of the more subtle acoustic problems discussed in this thread. Nulls are formed by incoherent summing and if you're listening to that, diffraction and high order modes are not your biggest acoustical problem.
Wayne Parham said:
They are available for the 15" version here:
http://gedlee.com/downloads/Yamaha vs. ESP15.pdf
The last 4 curves are ESP15 and not the Yamaha. From what I can tell, the first nulls at the crosspoint are around 30 degrees upward and 10 degrees downward.
Mark Seaton said:
Tony has his opinions, but in any case the desire "to achieve a -3 dB power response" is not an adequitely defined spec to support or not support dipoles. The dipole may only be "an available solution", but if there isn't a better one "available" then its kind of optimal isn't it?
Very good. Print these and lay them out to look at.
http://gedlee.com/downloads/Yamaha vs. ESP15.pdf
As I understand it, pages 10-14 are mislabeled and are actually the Geddes flagship ESP15 speaker with 15" horn and woofer.
I am assuming each curve 7.5 degrees further off-axis from the one above it.
First, the Geddes on-axis and horizontal curves are very nice. Good job!
The Yamaha isn't very smooth on-axis. Also, the horizontals don't track the on-axis chart very well. Usually I see better tracking at least up to about 45 degrees. A real clue as to what is going on with this speaker is the vertical nulls appear almost immediately, at relatively narrow angles. Between the roughness of response on-axis and the nearly immediate onset of nulls, I'd say summing isn't coherent on-axis. Response actually cleans up off-axis, so that tells me the crossover could use some work. All these things considered, the Yamaha may not be the best speaker to compare with, but it does show some features worth looking at.
Notice the Yamaha's vertical charts. It appears to use an HF horn with a narrow vertical pattern. At the nulls, output dips through the crossover region, as we would expect. At high frequencies, output is also reduced at large vertical angles. Outside about 30 degrees, the HF output is reduced considerably, approximately equal to the amplitude of the nulls in the crossover region.
When you move outside the null angle of a speaker like this, it sounds like you've moved outside the pattern. The treble falls off. It doesn't sound phasey so much as it just sounds like you've moved outside the pattern.
Now lets look at Geddes speaker, with the large round horn. The crossover is phased properly for coherent summing on-axis, as shown by the on-axis and horizontal charts. Very nice.
Assuming the charts are all 7.5 degrees apart, the nulls first appear at approximately 15 degrees upward and 7.5 degrees downward. This narrow arc is because of the distance between the sound sources. And notice that there's a large dip around 1kHz followed by increased response above that. Above the null dip, it comes back up and is nice and smooth from 2kHz on. This dip is very noticeable off-axis because it is so pronounced. With the HF output still high at large vertical angles, the null dip is very pronounced. Movement outside the coherent pattern doesn't sound so much like you're out of the pattern, because most of the audio band is still present there. But the anomaly in the crossover region produces a 10dB+ dip that sounds phasey as you pass through it.
This illustrates what I've been talking about. The Yamaha speaker isn't the best example, but it still shows the HF reduction at large vertical angles. If you make a loudspeaker with the same care that Dr. Geddes has put into his Summas, but use a horn that allows closer spacing and also has limited vertical directivity at HF, you've really got the best of both worlds. That's the point I've been trying to make.
You know, I'm not discounting HOM or any of the other work Dr. Geddes has made in that regard. I'm not saying it's inaudible or dismissing it in any way. But I am saying that nulls from incoherent summing are much larger acoustical anomalies. Reflections and diffraction effects can be measured in the time domain and the frequency domain. So can cancellation notches from incoherent summing, and their effects are much larger. This isn't something that should be discounted.
http://gedlee.com/downloads/Yamaha vs. ESP15.pdf
As I understand it, pages 10-14 are mislabeled and are actually the Geddes flagship ESP15 speaker with 15" horn and woofer.
I am assuming each curve 7.5 degrees further off-axis from the one above it.
First, the Geddes on-axis and horizontal curves are very nice. Good job!
The Yamaha isn't very smooth on-axis. Also, the horizontals don't track the on-axis chart very well. Usually I see better tracking at least up to about 45 degrees. A real clue as to what is going on with this speaker is the vertical nulls appear almost immediately, at relatively narrow angles. Between the roughness of response on-axis and the nearly immediate onset of nulls, I'd say summing isn't coherent on-axis. Response actually cleans up off-axis, so that tells me the crossover could use some work. All these things considered, the Yamaha may not be the best speaker to compare with, but it does show some features worth looking at.
Notice the Yamaha's vertical charts. It appears to use an HF horn with a narrow vertical pattern. At the nulls, output dips through the crossover region, as we would expect. At high frequencies, output is also reduced at large vertical angles. Outside about 30 degrees, the HF output is reduced considerably, approximately equal to the amplitude of the nulls in the crossover region.
When you move outside the null angle of a speaker like this, it sounds like you've moved outside the pattern. The treble falls off. It doesn't sound phasey so much as it just sounds like you've moved outside the pattern.
Now lets look at Geddes speaker, with the large round horn. The crossover is phased properly for coherent summing on-axis, as shown by the on-axis and horizontal charts. Very nice.
Assuming the charts are all 7.5 degrees apart, the nulls first appear at approximately 15 degrees upward and 7.5 degrees downward. This narrow arc is because of the distance between the sound sources. And notice that there's a large dip around 1kHz followed by increased response above that. Above the null dip, it comes back up and is nice and smooth from 2kHz on. This dip is very noticeable off-axis because it is so pronounced. With the HF output still high at large vertical angles, the null dip is very pronounced. Movement outside the coherent pattern doesn't sound so much like you're out of the pattern, because most of the audio band is still present there. But the anomaly in the crossover region produces a 10dB+ dip that sounds phasey as you pass through it.
This illustrates what I've been talking about. The Yamaha speaker isn't the best example, but it still shows the HF reduction at large vertical angles. If you make a loudspeaker with the same care that Dr. Geddes has put into his Summas, but use a horn that allows closer spacing and also has limited vertical directivity at HF, you've really got the best of both worlds. That's the point I've been trying to make.
You know, I'm not discounting HOM or any of the other work Dr. Geddes has made in that regard. I'm not saying it's inaudible or dismissing it in any way. But I am saying that nulls from incoherent summing are much larger acoustical anomalies. Reflections and diffraction effects can be measured in the time domain and the frequency domain. So can cancellation notches from incoherent summing, and their effects are much larger. This isn't something that should be discounted.
Nulls are not "incoherent summing" if it was incoherent it couldn't sum to near zero. And your measurement of where the nulls are at is not the same as mine. I see the upper null at about 25° and the downward one is not a "null" since its only -6 dB and its at about 15° - the real "null" is far above that. No one could be 15° below the axis, nor 25 ° above it.
As I said horizontal is everything, vertical is only somewhat important.
I expect your systems to look like the Yamaha. Show me otherwise.
As I said horizontal is everything, vertical is only somewhat important.
I expect your systems to look like the Yamaha. Show me otherwise.
If you're talking about the four Pi measurements, you've seen them. The response is nice and smooth in both vertical and horizontal planes through the 90x40 pattern.
As for your charts, isn't each one 7.5 degrees further off-axis from the one above it? Third-octave smoothing, I believe, yes? And the charts on pages 10-14 are of your ESP15, correct?
Once those clarifications are made, I think your charts speak for themselves.
As for your charts, isn't each one 7.5 degrees further off-axis from the one above it? Third-octave smoothing, I believe, yes? And the charts on pages 10-14 are of your ESP15, correct?
Once those clarifications are made, I think your charts speak for themselves.
gedlee said:
I dissagree with many of Tony's oppinions and approaches, which is separate topic from this mention. This was just an anecdote from someone previously involved with Dolby & THX and how some priorities might have contributed to the common useage of dipoles.
My point was that dipoles were more an option which seem to work at first glance, but the need was not specifically for dipoles. A dipole happened to be closer to what they wanted than compact, more conventional options of the time.
I would not say there isn't a better one available, but rather there wasn't one readily obvious or one which they thought practical at the time. We understand quite a bit more and have much more readily available modeling and measurement capability today than years past.
I would liken the choice of dipoles to past determinations that waveguides can't work well, so dome tweeters must be better... The past examples observed might have suggested that casual conclusion, but the understanding or interest wasn't there to get around the problems of the older designs. In my oppinion, there has to be better solutions than most dipole implementations. That's not to say a dipole can't be part of the solution, but I haven't seen many behave in a way that looks in any way optimal.
As a simple test of the possibility, I would suggest at some point you set up a system with Nathan's or similar for the surrounds, set up analogous to the main speaker aiming at the listening area. The setup may not be suitable for a full room, but the differences should be apparent at a single listening location or over some listening area.
The Nathans are just too big to seriously consider that in my room. So for me thats not an option. And I thought that what I offered up WAS a good solution to the problem. If you read that post I said that I was not completely bought into dipoles, maybe bipoles, but the drivers at the ends of a tube still has some attraction for me in this highly constrained situation.
Q for Mark Seaton.
So Mark, what would you suggest for sides/rears in an HT, where size of the speakers is less of an issue? I've got some of Earl's WG's arriving soon, and am decided on the design of the mains yet am usure of the sides rears because of conflicting suggestions re dipole/bipole etc.
I started a thread on it while back here,
http://www.diyaudio.com/forums/showthread.php?s=&threadid=125915
but got little response. Please feel free to post there if you don't want to take this thread further away from the original intent.
So Mark, what would you suggest for sides/rears in an HT, where size of the speakers is less of an issue? I've got some of Earl's WG's arriving soon, and am decided on the design of the mains yet am usure of the sides rears because of conflicting suggestions re dipole/bipole etc.
I started a thread on it while back here,
http://www.diyaudio.com/forums/showthread.php?s=&threadid=125915
but got little response. Please feel free to post there if you don't want to take this thread further away from the original intent.
I believe that it would work in reverse to yield a directional microphone at HF.
But there would be no point in that. Microphones don't have to be very efficient - they are not at all - so they can be made small and inexpensive and directivity can be far better accomplished with array techniques.
This too doesn't work so well in reverse, although it has become quite popular - using arrays for directivity control. The reason here is that for playback the transducer does need to be efficient and this causes size restrictions and bandwidth considerations that don't exist for the mic case. I have heard a lot of these array systems using small drivers to point the different channels in different directions. The sound locations are indeed impressive, but the sound quality is always lacking. People these days don't seem to care much about sound quality. As long as there is sound coming from all over the place, thats enough. Kind of like Bose 901's.
But there would be no point in that. Microphones don't have to be very efficient - they are not at all - so they can be made small and inexpensive and directivity can be far better accomplished with array techniques.
This too doesn't work so well in reverse, although it has become quite popular - using arrays for directivity control. The reason here is that for playback the transducer does need to be efficient and this causes size restrictions and bandwidth considerations that don't exist for the mic case. I have heard a lot of these array systems using small drivers to point the different channels in different directions. The sound locations are indeed impressive, but the sound quality is always lacking. People these days don't seem to care much about sound quality. As long as there is sound coming from all over the place, thats enough. Kind of like Bose 901's.
But nature recordist would be indeed interested 
Question arises from http://www.simpsonmicrophones.com/
Question arises from http://www.simpsonmicrophones.com/