• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Effect of Loudspeaker Directivity Compared with In-room Measurements

OP
Kvalsvoll

Kvalsvoll

Addicted to Fun and Learning
Audio Company
Joined
Apr 25, 2019
Messages
878
Likes
1,643
Location
Norway
Would that also correct reflections arrival time?
Not really, so it will not be exactly the same. Even with time delay correction.
 
OP
Kvalsvoll

Kvalsvoll

Addicted to Fun and Learning
Audio Company
Joined
Apr 25, 2019
Messages
878
Likes
1,643
Location
Norway
Two people might be listening.

Rick “choosy about who sits in his lap” Denney
For demo sessions, there are often more than 2 listening. And I get the worst place sitting by the left wall.

The sweet-spot is quite large with the new speakers, but it will fall short trying to cover 2 seats in my small room - you get 2 slightly compromised locations, vs one where everything is right, and all off-center reasonably good.

What happens off-center is not so much that objects are no longer in their correct location, it is more that you loose focus, everything is slightly blurred. And this can not be fixed, at least not with current technology.

What happens off-center can be very different, for different speakers. Some collapse everything into the nearest speaker, while others can still present a soundstage of sort, with instruments not sounding like they come from 2 small speakers.
 

thorvat

Senior Member
Joined
Aug 9, 2021
Messages
323
Likes
386

Yes, I did. Did you understand what I'm trying to say? It will only work for relatively small movements from optimal LP position, meaning that in a small room you will still not be able to cover 2-seats LP.

If you want to compensate moving LP further away toward one speaker you would need to compensate not only for amplitude change but for timing change as well if you want to retain proper stereo imaging. That would of course imply that other listening positions will be negatively impacted. If you want to have multi-seated LP you simply need larger room and relatively large lsitening distance so that differences between seats are small relative to lsitening distance.
 

markus

Addicted to Fun and Learning
Joined
Sep 8, 2019
Messages
646
Likes
653
Yes, I did. Did you understand what I'm trying to say? It will only work for relatively small movements from optimal LP position, meaning that in a small room you will still not be able to cover 2-seats LP.

If you want to compensate moving LP further away toward one speaker you would need to compensate not only for amplitude change but for timing change as well if you want to retain proper stereo imaging. That would of course imply that other listening positions will be negatively impacted. If you want to have multi-seated LP you simply need larger room and relatively large lsitening distance so that differences between seats are small relative to lsitening distance.
Well, the Rodenas paper shows that it works. The necessary level differences are rather large though. As you said you can't compensate the time differences as that would destroy imaging for other locations. What compensates the time difference is level. That's what time-intensity trading is all about. An image that was shifted by interaural time difference can be shifted to another location by interaural level adjustments alone, keeping the time difference. See Blauert "Spatial Hearing".
 

thorvat

Senior Member
Joined
Aug 9, 2021
Messages
323
Likes
386
Well, the Rodenas paper shows that it works. The necessary level differences are rather large though. As you said you can't compensate the time differences as that would destroy imaging for other locations. What compensates the time difference is level. That's what time-intensity trading is all about. An image that was shifted by interaural time difference can be shifted to another location by interaural level adjustments alone, keeping the time difference. See Blauert "Spatial Hearing".

You seem to fail to understand that the Rodenas paper shows that it works only for timing less than 1ms, hence for small distances relative to listening distance.
 
OP
Kvalsvoll

Kvalsvoll

Addicted to Fun and Learning
Audio Company
Joined
Apr 25, 2019
Messages
878
Likes
1,643
Location
Norway
I claim this problem with good-sound-off-center is solved good enough, with the radiation pattern that also gives the best sound in-center. Sound is just fine sitting off-center, even when laying down on the sofa. When you want to do more critical listening, you sit in the chair in-center, to experience the best sound. Getting the best focus and rendering and depth requires sitting in-center, but this sweet-spot is large enough to allow for moving your head around a little, in all directions.

While similar sound across the width of the room may actually be possible to solve, it requires some effort, and the speaker installation will be more complex, requiring some sort of array of speakers up-front. I am not so sure sound enthusiasts would care enough about getting a more similar sound across the room, cost-benefit just does not match up.

How this can work with 2 speakers and directivity pattern optimized for this purpose, can be tested as described in my post above with the pictures:
  • Play something with a solid center image
  • Move the image by moving yourself to one side, or delay one speaker
  • Try to compensate and move the image back to center by adjusting gain on one or both L-R channels (Typical 3dB diff for smaller movement, up to 8-10dB for full left or right)
Then you realize it is possible to move the image back, but it will not be focused, it is not the same. If your system did not have a focused, solid rendering of the center image to begin with, this experiment may seem more successful, as the low frequency part never was focused.
 

markus

Addicted to Fun and Learning
Joined
Sep 8, 2019
Messages
646
Likes
653
You seem to fail to understand that the Rodenas paper shows that it works only for timing less than 1ms, hence for small distances relative to listening distance.
Where do you read that in the paper? Those numbers come from dichotic listening tests. You can't move a phantom image any further than from left to right ear. It works just within a certain interaural time delay.
Time-intensity trading over loudspeakers works even for more distant off-center listening locations. It's far from perfect and also depends on signal type, signal level and likely also shows significant inter-individual variations. Better than nothing though and as long as it doesn't compromise the center listening position well worth a speaker design consideration in my opinion.
 
Last edited:

youngho

Senior Member
Joined
Apr 21, 2019
Messages
486
Likes
799
Cardioid is first and foremost an engineering challenge. Whether the design is working or not can be objectively demonstrated with measurements. What problems it actually solves hasn't been evaluated in an objective way though. I've made a couple of suggestions how that could be done but it was rejected as "demanding an unreasonable level of proof".

If someone would present a "rotated cardioid pattern" I would be very interested by the way as it tries to solve a fundamental objective problem of stereo, the small sweet spot. See http://www.aes.org/e-lib/browse.cfm?elib=5177

View attachment 174679

Source: https://pure.tue.nl/ws/files/2873329/Metis235355.pdf
@markus I skimmed both papers, and I have two questions:
1. For the Davis paper, could one achieve a relatively similar result by using two cardioid speakers pointed straight at each other? That basically seems to be the radiation pattern shown. "The loudest axis points not out into the room but at the other loudspeaker..." The contrast between the loudest and softest axes seems to depend on the distance between the loudspeakers.
2. The the PI-stereo paper, the apparent positioning of the loudspeakers relative to the wall behind it was straight ahead, as shown in Figure 1. Figure 12 again seems to demonstrate a radiation pattern primarily directed 90 degrees from "out into the room." The ideal radiation pattern seems to have an almost warped bipolar quality to it with cancellation at about 200 degrees and some dampening (~5 dB) of the backwave. Could one achieve a relatively similar result by using two curved panel loudspeakers (Martin Logan?) pointed straight at each other, with an absorption element like a TubeTrap or acoustic panel positioned directly behind (relative to each panel) each speaker?
 

markus

Addicted to Fun and Learning
Joined
Sep 8, 2019
Messages
646
Likes
653
@markus I skimmed both papers, and I have two questions:
1. For the Davis paper, could one achieve a relatively similar result by using two cardioid speakers pointed straight at each other? That basically seems to be the radiation pattern shown. "The loudest axis points not out into the room but at the other loudspeaker..." The contrast between the loudest and softest axes seems to depend on the distance between the loudspeakers.
2. The the PI-stereo paper, the apparent positioning of the loudspeakers relative to the wall behind it was straight ahead, as shown in Figure 1. Figure 12 again seems to demonstrate a radiation pattern primarily directed 90 degrees from "out into the room." The ideal radiation pattern seems to have an almost warped bipolar quality to it with cancellation at about 200 degrees and some dampening (~5 dB) of the backwave. Could one achieve a relatively similar result by using two curved panel loudspeakers (Martin Logan?) pointed straight at each other, with an absorption element like a TubeTrap or acoustic panel positioned directly behind (relative to each panel) each speaker?
The effect can be achieved with different designs. Extreme toe-in is probably just good for first experiments. Too much unusable energy is pumped into the room which then needs to be absorbed. Shaping the response with some kind of on/in-wall horn design is probably best.

Attached 2 graphs from Kates, "Optimum Loudspeaker Directional Patterns".
 

Attachments

  • Screenshot 2021-12-29 at 17.14.26.png
    Screenshot 2021-12-29 at 17.14.26.png
    77.3 KB · Views: 60
  • Screenshot 2021-12-29 at 17.14.46.png
    Screenshot 2021-12-29 at 17.14.46.png
    104.6 KB · Views: 60
Last edited:

youngho

Senior Member
Joined
Apr 21, 2019
Messages
486
Likes
799
The effect can be achieved with different designs. Extreme toe-in is probably just good for first experiments. Too much unusable energy is pumped into the room which then needs to be absorbed. Shaping the response with some kind of on/in-wall horn design is probably best.

Attached 2 graphs from Kates, "Optimum Loudspeaker Directional Patterns".
It just seems strange to have cited the previous two papers that have such unusual directivity/setup recommendations.

With regards to the Kates paper, which I downloaded here (http://decoy.iki.fi/dsound/ambisonic/motherlode/source/Optimum loudspeaker directional patterns_Kates_1980.pdf), it appears to me that the D&D 8C and Kii Audio Three come closest to meeting their optimal directional pattern at both low (which is not clearly defined) and high frequencies for an equilateral triangle setup (figure 7, table 1).
 

markus

Addicted to Fun and Learning
Joined
Sep 8, 2019
Messages
646
Likes
653
It just seems strange to have cited the previous two papers that have such unusual directivity/setup recommendations.

With regards to the Kates paper, which I downloaded here (http://decoy.iki.fi/dsound/ambisonic/motherlode/source/Optimum loudspeaker directional patterns_Kates_1980.pdf), it appears to me that the D&D 8C and Kii Audio Three come closest to meeting their optimal directional pattern at both low (which is not clearly defined) and high frequencies for an equilateral triangle setup (figure 7, table 1).
Per the Kates table the dispersion of both the D&D and Kii seems to be too wide when used in a standard stereo setup (equilateral triangle). -3dB point for both around +/-40° and more with the Kii having a bit wider dispersion.
 

Attachments

  • Kii THREE Horizontal Contour Plot (Normalized).png
    Kii THREE Horizontal Contour Plot (Normalized).png
    333.2 KB · Views: 62
  • Dutch & Dutch 8c Horizontal Contour Plot (normalized).png
    Dutch & Dutch 8c Horizontal Contour Plot (normalized).png
    304 KB · Views: 46

youngho

Senior Member
Joined
Apr 21, 2019
Messages
486
Likes
799
Per the Kates table the dispersion of both the D&D and Kii seems to be too wide when used in a standard stereo setup (equilateral triangle).
Actually, the text and the table seem to be at odds, which I didn't realize before.

I had been looking at the unnormalized horizontal contour plots at https://www.erinsaudiocorner.com/loudspeakers/dutch_dutch_8c/, where the 99-102 range seems to roughly go from +/- 0-25 degrees for much of the frequency range from several hundred Hz on up, then the 96-99 range is pretty close to +/- 50 for most of this range, as well. The Kii Three is certainly wider, but again, it also seems to come closer at both and high frequencies (and go lower) than other speakers. I didn't say that they matched it, but the D&D 8C seems to come closest, at least to me.
 

markus

Addicted to Fun and Learning
Joined
Sep 8, 2019
Messages
646
Likes
653
Actually, the text and the table seem to be at odds, which I didn't realize before.

I had been looking at the unnormalized horizontal contour plots at https://www.erinsaudiocorner.com/loudspeakers/dutch_dutch_8c/, where the 99-102 range seems to roughly go from +/- 0-25 degrees for much of the frequency range from several hundred Hz on up, then the 96-99 range is pretty close to +/- 50 for most of this range, as well. The Kii Three is certainly wider, but again, it also seems to come closer at both and high frequencies (and go lower) than other speakers. I didn't say that they matched it, but the D&D 8C seems to come closest, at least to me.
Best to look at the normalized plots as they make it obvious what the beamwidth really is. For the D&D and Kii the -3dB beamwidth is around 80°-100°.
 

youngho

Senior Member
Joined
Apr 21, 2019
Messages
486
Likes
799

markus

Addicted to Fun and Learning
Joined
Sep 8, 2019
Messages
646
Likes
653
Erin's normalized plots seem to compress oddly. Compare the Genelec 8361A review on ASR (https://www.audiosciencereview.com/...s/genelec-8361a-review-powered-monitor.28039/), especially relative to the directivity indices of the 8361A and D&D 8C. This does not seem compatible to me.
What do you mean by "compress"? Amir's plots are in 1dB increments. I find them harder to read than Erin's 3dB steps. Broader steps are more useful for initial examination whereas smaller steps are better for detailed analysis.
 

youngho

Senior Member
Joined
Apr 21, 2019
Messages
486
Likes
799
Ah, my mistake. It feels strange for me to look at the Kef Wireless II (https://www.erinsaudiocorner.com/loudspeakers/kef_wireless_ii/), which has a lower directivity index for most of its response compared with the D&D 8C (https://www.erinsaudiocorner.com/loudspeakers/dutch_dutch_8c/), compare their unnormalized horizontal contour plots, but then the LS50 has a +/- 50 degree beamwidth (https://www.audiosciencereview.com/forum/index.php?threads/kef-ls50-meta-review-speaker.25574/ and https://www.erinsaudiocorner.com/loudspeakers/kef_wireless_ii/), while the D&D 8C has a +/- 80-100 degree beamwidth.

It's hard to imagine a speaker, then, that could achieve the suggested directivity in the Kates paper. Even the Beolab 90 seems to be too wide (https://www.tonmeister.ca/wordpress/2015/06/30/bo-tech-what-is-beam-width-control/)
 
OP
Kvalsvoll

Kvalsvoll

Addicted to Fun and Learning
Audio Company
Joined
Apr 25, 2019
Messages
878
Likes
1,643
Location
Norway
Ah, my mistake. It feels strange for me to look at the Kef Wireless II (https://www.erinsaudiocorner.com/loudspeakers/kef_wireless_ii/), which has a lower directivity index for most of its response compared with the D&D 8C (https://www.erinsaudiocorner.com/loudspeakers/dutch_dutch_8c/), compare their unnormalized horizontal contour plots, but then the LS50 has a +/- 50 degree beamwidth (https://www.audiosciencereview.com/forum/index.php?threads/kef-ls50-meta-review-speaker.25574/ and https://www.erinsaudiocorner.com/loudspeakers/kef_wireless_ii/), while the D&D 8C has a +/- 80-100 degree beamwidth.

It's hard to imagine a speaker, then, that could achieve the suggested directivity in the Kates paper. Even the Beolab 90 seems to be too wide (https://www.tonmeister.ca/wordpress/2015/06/30/bo-tech-what-is-beam-width-control/)
None of those speakers have a pattern that can work well for this. The pattern is too wide at mid-high freq, and it does not fall off sharp enough.

A radiation pattern can not be described well just using a radiation angle number, you need to look at how the pattern is across both angle and frequency.
 
Top Bottom