Open Baffle musings

[fdieck]

Re: Open Baffle musings

I agree that constant directivity (where practical) is a great goal and the only way to be truly pure dipole.

On the other hand, Charlie might be right and those lobes might not measure as bad as the pics show, if that is the case then obviously the lobing issue above the first dipole peak is not nearly as pressing.

I'm all for keeping a constant radiation pattern but at what cost? Even John's tweeter isn't narrow enough to cover the high frequencies so there's no perfect option. Where do we draw the line? Is it wrong for hobbyists to try a wide MJK style OB with a fullrange driver? It's not pure dipole but I can tell you from experience that it's not that bad, although I didn't do any comprehensive measurements.

http://www.quadesl.com/articles/quad_review.html

Looking at the first full range dipole the Qual ESL, it is very interesting that crossover frequencies of 1kHz and 7kHz and effective mid and high frequency radiator (driver) widths were about 6 inches and 1 inch. These are comparable what we are talking about for 2 of the crossover points and drivers of a four way dipole. I believe that there was some absorbing material behind the high frequency panel as well. It is often useful to examine first principles and Peter Walker was a genius. No speaker has passed the test of time like the orignal quad ESL. The curved line source HF radiator might even be considered a precursor of DB Keele's Constant Beamwidth Transducer.

[charlielaub]

Re: Open Baffle musings

Not sure what you want to hear . . . on the face of it the multiple lobes are bad. Since their location (angle) varies with frequency the result (apart from power response variations) is frequency response variation (both direct and reflected) at any particular listening position.


Yes. Physics makes the single-driver full-range impractical, the added constraint of constant (dipole) directivity makes it impossible. In general the most we can get from a single driver/baffle is three octaves . . . other constraints limit even that. Since we're looking for 9 octaves in a "full range" speaker it's obvious why 4-way is the end point of most evolved dipole designs.

I don't see the point of rehashing why constant directivity is a good idea . . . if that's not understood and "part of the plan" then there's little point in dealing with all the dipole issues in the first place. A loudspeaker does not become "a dipole" just by taking the box away, but at the same time taking the box away does not free dipoles of all other design considerations. If anything it makes them more pronounced, since excursion and equalization demands become greater and some of the "fudge factors" that come with a box (like "make the messy back wave go away") are lost as well.

If you want to make a full range dipole there are just things you have to do (and other things that you can't). If you disregard those things you won't get the constant directivity and radiation pattern (and resultant benefits) that are the reason for undertaking a dipole design in the first place, so why bother.

Dipoles are bi-directional sources of radiation (acoustic radiation in this case). It doesn't matter in what part of the frequency regime one is operating the bi-directional source, it's still called a dipole. I think that one needs to be careful to say that a dipole is only a dipole below the first peak in the response. That's rubbish. And this isn't a "linear region", it's the region in which the front and back radiation "see each other" more and more, and thus there is increasing amounts of cancellation. That's what is leading to the linear portion of the classic dipole frequency response plot. Again I stress that the classic dipole pattern (see SL's webs site that I linked to above) is for two point sources isolated in space. A real world driver has a non-point-source radiating surface, which at higher frequencies will help to mitigate the lobing in the dippole pattern. Varying pathlengths to the baffle edge is another way to help mitigate the dipole pattern lobing at higher frequencies.

With that said, I now see the inevitability that people are talking about. Deward mentions constant directivity in the quote above. You get that only when you are well below the first dipole peak in frequency. As you approach the peak, the pattern is narrowing into front and rear lobes on the listening axis and so the constant directivity is breaking down. But I'm guessing that power response is probably about the same, because there are peaks and nulls occurring at different frequencies and different angles as you increase in frequency which probably cancel out for the most part. I'm sure that JohnK has a nice model for this that proves me wrong! But I sense that this is like the Butterworth 3rd order crossover power response (for a closed box) where peaks and nulls in the response as a function of position and frequency around the crossover point end up nicely cancelling out and the power response is flat.

As diyspeakerguy mentioned, where do we draw the line on constant directivity and the dipole pattern itself? As JohnK mentioned, and that I have now convinced myself with some sims, you can't make the baffle narrow enough (because of the finite size of the driver itself) to make it possible to have the nice constant directivity dipole pattern that you get below the first peak, because the first peak just can't be pushed much higher than 2-3 kHz. So you are going to have to lose the good fight somewhere, so you might as well figure out how to do it gracefully, meaning how can we ameliorate the inevitable lobing that will occur at some frequency, and above what frequency do we need to no longer worry that we have constant directivity as a result of being below the first peak in the dipole response? In that context, my proposed loudspeaker is not attempting to push the dipole peak as high as possibly in frequency (by making the baffle as narrow as possible) but instead I am trying to use a combination of a large(r) midrange driver and careful positioning on the baffle so that the dipole frequency response pattern is largely washed out, but the sytem is still operating like a dipole, and the back radiation is free to escape however it wants to. Since I will operate the midrange through lots of different dipole pattern regimes before crossing over to the HF driver (e.g. tweeter) I need to make it work the best I can. It's a different approach than SL and JohnK, but it seems that it's one that hasn't been received too badly if done well.

-Charlie

[Deward Hastings]

Re: Open Baffle musings

Is it wrong for hobbyists to try a wide MJK style OB with a fullrange driver? It's not pure dipole

Nothing "wrong" with it at all. Nothing wrong with building a Geddes Summa or one of Bill's DR280s either. They're not dipoles, not constant directivity, and for their applications it doesn't matter. What "gets me" is when someone says they've heard a "MJK style OB with a fullrange driver" and "don't see why all the fuss about dipoles".

Well Duh . . .

I've heard them too, and I've heard dipoles, and . . .

If the intention is just to do some "OB" speakers on a plank and neither constant directivity nor matching on-axis and power response are concerns . . . ie. just "box" speakers without a box . . . then none of my comments above apply, and to the opening comment

with all the talk about SL's LX521 system, I'm itching to try some kind of open baffle system.

I should just have said "no connection . . . completely different beasts . . . not the same thing at all" . . . and "have fun" . . .

Perhaps I misunderstood the desire and intent . . .

[beaujewel]

Re: Open Baffle musings



My solution to the dipole tweeter conundrum. Hopefully will move the dipole peak way up past 7k at least. No rear structure to interfere as well.
Prettier model to come.

[Deward Hastings]

Re: Open Baffle musings

I'm all for keeping a constant radiation pattern but at what cost? Even John's tweeter isn't narrow enough to cover the high frequencies so there's no perfect option. Where do we draw the line?

It's a hard question, like so many of the "compromises" in speaker design. The "improvement" (as much as there is) of LX521 compared to ORION presumably comes almost entirely from correcting ORION's pattern irregularity from about 1.5kHz up to about 4kHz. Above and below those frequencies the polar behavior of the two speakers is essentially the same, and I didn't notice on-axis frequency response differences overall of any notable import. I did notice, however, that the "acoustic scene" thing is substantially improved with LX521, even in that not-particularly-favorable room . . . enough of a difference that I'm seriously contemplating tweeter/baffle mods on my ORION. I got the sense that Charlie was favorably impressed by what he heard from LX521, but I think he is misunderstanding the most important characteristic of the speaker that might have produced that impression.

No point beating a dead horse . . . I've made my points, and any more that I might say would be repetition (I've already probably said too much).

Anyway, a box of drivers arrived today, I've got my eye on a couple of planks of White Oak at the local yard (for my ORION cabinet mods) that I should grab before someone else gets them, and I'm sure we'll be comparing notes on our projects at BurningAmp next year, if not before.

[diy speaker guy]

Re: Open Baffle musings

My solution to the dipole tweeter conundrum. Hopefully will move the dipole peak way up past 7k at least. No rear structure to interfere as well.
Prettier model to come.

Details please.

[johnk...]

Re: Open Baffle musings

Dipoles are bi-directional sources of radiation (acoustic radiation in this case). It doesn't matter in what part of the frequency regime one is operating the bi-directional source, it's still called a dipole. I think that one needs to be careful to say that a dipole is only a dipole below the first peak in the response. That's rubbish. And this isn't a "linear region", it's the region in which the front and back radiation "see each other" more and more, and thus there is increasing amounts of cancellation. That's what is leading to the linear portion of the classic dipole frequency response plot. Again I stress that the classic dipole pattern (see SL's webs site that I linked to above) is for two point sources isolated in space. A real world driver has a non-point-source radiating surface, which at higher frequencies will help to mitigate the lobing in the dippole pattern. Varying pathlengths to the baffle edge is another way to help mitigate the dipole pattern lobing at higher frequencies.

With that said, I now see the inevitability that people are talking about. Deward mentions constant directivity in the quote above. You get that only when you are well below the first dipole peak in frequency. As you approach the peak, the pattern is narrowing into front and rear lobes on the listening axis and so the constant directivity is breaking down. But I'm guessing that power response is probably about the same, because there are peaks and nulls occurring at different frequencies and different angles as you increase in frequency which probably cancel out for the most part. I'm sure that JohnK has a nice model for this that proves me wrong! But I sense that this is like the Butterworth 3rd order crossover power response (for a closed box) where peaks and nulls in the response as a function of position and frequency around the crossover point end up nicely cancelling out and the power response is flat.

As diyspeakerguy mentioned, where do we draw the line on constant directivity and the dipole pattern itself? As JohnK mentioned, and that I have now convinced myself with some sims, you can't make the baffle narrow enough (because of the finite size of the driver itself) to make it possible to have the nice constant directivity dipole pattern that you get below the first peak, because the first peak just can't be pushed much higher than 2-3 kHz. So you are going to have to lose the good fight somewhere, so you might as well figure out how to do it gracefully, meaning how can we ameliorate the inevitable lobing that will occur at some frequency, and above what frequency do we need to no longer worry that we have constant directivity as a result of being below the first peak in the dipole response? In that context, my proposed loudspeaker is not attempting to push the dipole peak as high as possibly in frequency (by making the baffle as narrow as possible) but instead I am trying to use a combination of a large(r) midrange driver and careful positioning on the baffle so that the dipole frequency response pattern is largely washed out, but the sytem is still operating like a dipole, and the back radiation is free to escape however it wants to. Since I will operate the midrange through lots of different dipole pattern regimes before crossing over to the HF driver (e.g. tweeter) I need to make it work the best I can. It's a different approach than SL and JohnK, but it seems that it's one that hasn't been received too badly if done well.

-Charlie

Hi Charlie,

A point source dipole power response in the 6dB region is -4.8 compared to a monopole with the same axial intensity. As the dipole peak is approached it rises to, as I recall, +6dB. The it wiggles around finally settling at +3dB. In the +3dB region the two sources just behave as two uncorrelated sources which is why you get +3dB compared to a single point source.

But talking about power response takes me to a different point. Up in the midrange the dipole is radiating into 4Pi space. The woofers, near the floor, are of course radiating into 2Pi space. The woofers are actually radiating 3dB less power than the mids. It would seem that a monopole woofer would yield a better power match; 3dB less than a free space monopole while the dipole mids would be 4.8 less that a free space monopole, so a monopole woofer would radiate 1.8dB more power than the dipole mid where as a dipole woofer radiates 3dB less than the dipole mids.

Then there is another issue that we could discuss. That is woofers in general. I find these days that I have come around to Gedde's way of thinking. Actually, I was always there, but never realized it. That is, the idea of a dipole, cardioid, or monopole woofer really has no meaning in a small acoustic space. Dipole and cardioid radiation patterns are free field phenomena. There is no dipole or cardioid radiation is a small acoustic space at low frequency. All there is are multiple monopole sources distributed around the room with different amplitude and phase and perhaps delay. it's all room modes. When you do a room simulation for a dipole or cardioid how do you set it up? You decide the orientation and place two monopole sources at the correct distance apart and with correct phase. For the cardioid you add delay to one source. They are all just multiple monopole source situations. Then all that matters is the near field response of each source, the source positions, the listening position, and the room transfer function between each source and the listening position.

Now, you may think that for a dipole or cardioid the front and rear sources are only about 12 to 24 inches apart, and how much difference can that make between the room transfer function for each source. You can see that very easily by invoking reciprocity. Reciprocity says that if you place a source at position A and take a measurement at position B you will get the same result as if the source is at position B and the measurement is at position A. So, place a woofer at you listening position and make a measurement at some point in your room where you might have a woofer. The move the Mic 18 inches to the left or right, up or down, forward or back. I think you may be surprised at the difference. So obviously, for a dipole, the two sources will produce different SPL at the listening position and the sum will not just be the sound of two out of phase sources with the same amplitude. This is in part why the argument that dipoles don't excite certain modes is not generally correct. It is not that dipole don't excite the modes. It is that they MAY excite the modes with inverted phase. But that is only 1/2 the story because to cancel out the amplitude of the excited mode must be the same from both sources. In a perfectly rectangular room with perfectly rigid walls and the correct placement, this can happen. But is a typical room, not so much. The net result is that dipoles and cardioid are just multiple monopole with unique amplitude and phase and position relations as far as a room is concerned. So dipoles and cardioids sounding better than monopole probably has more to do with the observation that they represent distributed stimulation of the room than anything else. I.E. good bass comes from multiple sources placed around the room with different phase and delays. That's pretty much what Geddes has been saying for years.

[johnk...]

Re: Open Baffle musings

My solution to the dipole tweeter conundrum. Hopefully will move the dipole peak way up past 7k at least. No rear structure to interfere as well.
Prettier model to come.

It looks about 1" wide which wouldput the peak at about 7K. But above that I would expect to see some axial nulls, not to mention the over all length will likely create problems with the vertical response.

[diy speaker guy]

Re: Open Baffle musings

(I've already probably said too much).

Nonsense. Other than John, around these parts you probably have the most practical experience with cutting edge open baffle design. I've never even heard an active dipole, much less a no holds barred design so none of my opinions count for much.

[diy speaker guy]

Re: Open Baffle musings

It looks about 1" wide which wouldput the peak at about 7K. But above that I would expect to see some axial nulls, not to mention the over all length will likely create problems with the vertical response.

Yeah but it looks diy and I want to know how to do it. The details of the design choices I'm not so interested in.

[johnk...]

Re: Open Baffle musings

Yes. That was/is the problem with the back-to-back tweeters in ORION . . . although they made the treble "bi-directional" (in some ways an improvement) they are "dipole" over exactly no part of their range. Until they start beaming there is nothing to control the side lobe(s), and that really compromises the "constant directionality" of the design. The tweeters in LX521 (also not aligned to be or constituting a dipole) are larger (diameter) domes than they otherwise need to be presumably to insure that they are already beaming at the crossover frequency chosen.

I think the problem with back to back in the Orion and the NaO II is more just tweeter bloom. The baffles are wide enough to keep the tweeter radiating into 2Pi for the most part. I know the NaO II on axis response is barily effected by turning off the rear tweeter. Could be more of an issue with the orion since the crossover point is lower.

[Deward Hastings]

Re: Open Baffle musings

I think the problem with back to back in the Orion and the NaO II is more just tweeter bloom.

Well yes, in some sense that's all it is. If there were some way to configure them as dipoles, however (which would of necessity include a smaller baffle as well) the dipole cancelation would take care of the null, just as it does on a properly baffled (single) cone driver. In any case it doesn't happen (just as it doesn't for those "full-range" drivers in the middle of a 36" wide baffle, regardless whether someone calls them "dipoles" or not). In any case it wouldn't matter whether they are "dipole" or not, if they didn't "bloom". It's not "dipole for dipole's sake", it's the pattern and the associated power response and reflections that matter. I don't find myself particularly bothered when the tweeters "beam" (as long as it's not too extreme), especially since that transition is relatively gradual. It's the abrupt discontinuity at the M/T crossover that is . . . bothersom. If I stand in (what should be) the null I can hear it. Notable about the LX521 was that I couldn't. Maybe that's why they sound different . . .

Totally arguable, of course, is whether it really matters . . . and that on a number of levels. With "studio" and most "mixed" recordings it doesn't, regardless . . . at least not usually. Whatever "position information" is in the recording is synthetic anyway . . . a gimmick, a "sound effect". And while there's a real immersive and satisfying feel with some well-done orchestral recordings that I've never heard from anything but "full range" dipoles (Magnaplanars can do it sometimes, too) it's simply never necessary for me to be "transported" to "D-21 Orchestra" to enjoy the music. Given the choice I'd generally take front row balcony anyway.

Don't know why I'm in this mood . . .if I weren't on a diet I'd have a drink (or two) . . .

[johnk...]

Re: Open Baffle musings

If I stand in (what should be) the null I can hear it. Notable about the LX521 was that I couldn't. Maybe that's why they sound different . . .

... it's simply never necessary for me to be "transported" to "D-21 Orchestra" to enjoy the music. Given the choice I'd generally take front row balcony anyway.

. .

Now that is what I was trying to get across at DIY about something missing from the Note compared to the NaO II. All that side radiation above the crossover. In the NaO II I do get some cancellation from the tweeters, but the tweeter radiation is still much broader than it is in the Note. Now I think you are experiencing exactly what I experience between the NaO II and the Note.

I have to say I have always preferred front row, first balcony. Probably because when I was young that is where I sat when my father took me to concerts. Very different than orchestra seating.

[Pallas]

Re: Open Baffle musings

I mean mount two units of a small tweeter (like the Vifa OX20) literally back to back and then mount those in the baffle at the top (there is no "box").

Anyway, I could get HF dipole operation if I use the Tang-Band W2-800SL:


It's a full range, but the radiating surface is about 1.5" in diameter IIRC. I've used it as a tweeter in a recent project. Mounted on top of the proposed baffle, I would get a dipole pattern. Add a rear chamber and it's monopole.

-Charlie

Not really. You won't get a dipole pattern (especially at high frequencies, but even in the mids) due to the fact that there's a big motor and basket on one side of the cone, right in the middle where the high frequencies usually emanate from. ***

The Aura Whisper would be a better choice for that kind of thing, because the motor structure and basket is all around the diaphragm. The bug screen is easy to remove, as long as you don't push it in and dent the diaphrgam.



Experimenting with stuffing inside the magnet ring/"basket" might be interesting, too.

[Pete Schumacher]

Re: Open Baffle musings

If the tweeter response is the issue with these dipole designs, why not put the tweeter into a waveguide and control the directivity that way since the flat baffle isn't helping?

Heck, put a dipole tweeter into a symmetrical waveguide . . . )-( . . . where "-" is the tweeter diaphragm. The Aurum Cantus Aerostriction tweeters with their felt pads removed from the back would be ideal candidates for such an exercise.