Importance of vertical polar response

[dantheman]

I'm trying to design a 2 way with a waveguided tweeter, but I'm a bit anxious about the crossover. My first concern is the center to center distance of the woofer and tweeter. Everything I'm looking at building will require this distance to be greater than the wavelength of the crossover frequency in order to get the horizontal directivity at the crossover frequency to match. Best case scenario I'm looking at is about 110% of the wavelength of the crossover frequency. I wonder how this will effect what I hear at 2-3 meters with the middle of the space between the drivers at me listening ears' height? Is there and sims or real world measurements of speakers with varying c to c spacing anywhere on the web? Or of a vertical polar response of a mid/wg or horn combo where c to c spacing is less than the wavelength of the crossover frequency and another where the space is greater?

I have the ability to measure speakers myself, but I can't seem to come up with a clever jig for measuring vertical polar response that will do an accurate job. Has anyone done this or know of an example?

Maybe Pi Speakers will have some info on this as Wayne seems to be the only designers who bothers with vertical polars--which is what begs me to question what effect they will have in room.

In my head it seems a narrow c to c would produce a broader response lobe, but is that good or bad? Seems more broad might create issues from early reflections, too narrow and you'll get response anomalies when you stand up. Thing is I generally don't watch movies standing up.

In any case I'd like to get a neutral intelligent discussion on the stated subject of the thread if there is any interest.

Dan

[thadman]

Originally Posted by dantheman

I'm trying to design a 2 way with a waveguided tweeter, but I'm a bit anxious about the crossover. My first concern is the center to center distance of the woofer and tweeter. Everything I'm looking at building will require this distance to be greater than the wavelength of the crossover frequency in order to get the horizontal directivity at the crossover frequency to match. Best case scenario I'm looking at is about 110% of the wavelength of the crossover frequency. I wonder how this will effect what I hear at 2-3 meters with the middle of the space between the drivers at me listening ears' height? Is there and sims or real world measurements of speakers with varying c to c spacing anywhere on the web? Or of a vertical polar response of a mid/wg or horn combo where c to c spacing is less than the wavelength of the crossover frequency and another where the space is greater?

I have the ability to measure speakers myself, but I can't seem to come up with a clever jig for measuring vertical polar response that will do an accurate job. Has anyone done this or know of an example?

Maybe Pi Speakers will have some info on this as Wayne seems to be the only designers who bothers with vertical polars--which is what begs me to question what effect they will have in room.

In my head it seems a narrow c to c would produce a broader response lobe, but is that good or bad? Seems more broad might create issues from early reflections, too narrow and you'll get response anomalies when you stand up. Thing is I generally don't watch movies standing up.

In any case I'd like to get a neutral intelligent discussion on the stated subject of the thread if there is any interest.

Dan

If I was designing a loudspeaker around a waveguide, I would consider applying FIR filters. A FIR filter should minimize overlap (>96dB/octave) without contributing any phase or amplitude errors.

[dantheman]

Originally Posted by thadman

If I was designing a loudspeaker around a waveguide, I would consider applying FIR filters. A FIR filter should minimize overlap (>96dB/octave) without contributing any phase or amplitude errors.

Seems a bit too complicated for me to even begin contemplating. Interesting idea though. I think we already know why others haven't taken this approach. $$$

Dan

[gedlee]

Many comments. SPEAK will model this for you. Its free now. Just go to my web site.

First, the horizontal polars are critical, while the verticals are just "important". Never compromise on the horizontal to solve an "imagined" problem in the vertical.

The spacing and crossover frequency sets the angle between the nulls. The lower the frequency the wider the nulls. There is NOTHING that you can do electrically to alleviate this and very sharp filters have very high phase changes and a whole other set of problems of there own. Basically the nulls are there no matter how sharp the filter is, it doesn't fix the most significant problem, it only creates other ones - like huge cost implications.

Of course closer spacing is always desirable as long as NOTHING is done to degrade the horizontal polars. This is why I don't approve of coax (except in space critical apps.) - no vertical lobes, but lousy horizontal pattern control. Give me a couple of vertical lobes over that any day.

Now, there will be a pair of nulls, up and down, and its best to place them symmetrical about the listening axis. This is not always easy to do, it takes some real effort in the crossover to get the vertical lobes symmetric AND a good match in the horizontal plane. Just getting a good axial response is trivial by comparison.

Thats about the best that you can do, and personally, I find that this situation works just fine.

As a practical example, the Summa has a C-C spacing of about 18 inches, and a crossover at about 800 Hz. The nulls are up and down at about 15-20 degrees. At 3-4 m back this places them at about 1 m up and down from ear level - not a problem. You said 2-3 m so worst case would be .5 m, more likely .75 m, still not a problem. There is an ever so slight dip in the power response at the crossover as a result of this, but nothing significant.

[JR-W]

Originally Posted by gedlee

Many comments. SPEAK will model this for you. Its free now. Just go to my web site.

I looked your site over, and must have missed this SPEAK of which you speak. On what page is it located, if I may?

[dantheman]

Originally Posted by gedlee

Many comments. SPEAK will model this for you. Its free now. Just go to my web site.

First, the horizontal polars are critical, while the verticals are just "important". Never compromise on the horizontal to solve an "imagined" problem in the vertical.

The spacing and crossover frequency sets the angle between the nulls. The lower the frequency the wider the nulls. There is NOTHING that you can do electrically to alleviate this and very sharp filters have very high phase changes and a whole other set of problems of there own. Basically the nulls are there no matter how sharp the filter is, it doesn't fix the most significant problem, it only creates other ones - like huge cost implications.

Of course closer spacing is always desirable as long as NOTHING is done to degrade the horizontal polars. This is why I don't approve of coax (except in space critical apps.) - no vertical lobes, but lousy horizontal pattern control. Give me a couple of vertical lobes over that any day.

Now, there will be a pair of nulls, up and down, and its best to place them symmetrical about the listening axis. This is not always easy to do, it takes some real effort in the crossover to get the vertical lobes symmetric AND a good match in the horizontal plane. Just getting a good axial response is trivial by comparison.

Thats about the best that you can do, and personally, I find that this situation works just fine.

As a practical example, the Summa has a C-C spacing of about 18 inches, and a crossover at about 800 Hz. The nulls are up and down at about 15-20 degrees. At 3-4 m back this places them at about 1 m up and down from ear level - not a problem. You said 2-3 m so worst case would be .5 m, more likely .75 m, still not a problem. There is an ever so slight dip in the power response at the crossover as a result of this, but nothing significant.

Thank you Dr. Geddes! So the wavelength of 800Hz is about 17inches or about 106% of the crossover frequency wavelength and you're just a touch beyond that with nulls at 15-20 degrees. The best I can do with my design seems to be about 122% of the wavelength at the crossover frequency. Anyone know how to calculate my nulls if I were to build such a thing? I'll model it when I get my windows rig up and running again if it can't be calculated. Seems to me it should be fairly straight forward though.

Right now I'm listening to a speaker that is fairly CD in the horizontal plane, but with a c to c spacing greater than 2x the crossover frequency and I still hear an advantage to CD compared to my direct radiators.

Dan

[twest820]

Speak is here. Also check out John Kreskovsky's article on power response and crossovers. Won't give you any answers but does provide a lot of tools.

My current system is just about two wavelengths at its LR4 2kHz cross. It's perfectly liveable but is definitely happier with a 2.5+m listening distance than at ~2m or lower. The design I'm currently looking into replacing it with would be about 1.3 wavelengths with an LR6 2kHz cross; should try putting that in Speak and seeing what I get.

[dantheman]

Thanks twest! Looks like I'll be doing some building soon.

[Don Hills]

Originally Posted by dantheman

I have the ability to measure speakers myself, but I can't seem to come up with a clever jig for measuring vertical polar response that will do an accurate job. Has anyone done this or know of an example?

Use the same setup as you would for horizontal polar response, except mount the speaker on its side instead of upright.

[dantheman]

Originally Posted by Don Hills

Use the same setup as you would for horizontal polar response, except mount the speaker on its side instead of upright.

That difficult d/t reflections off the device. I guess there not that critical in this situation b/c all you're looking for is the nulls and precise FR is less critical for performance evaluation. I'll see how it works.

Thanks Don!

Dan