[D OB G]

Here are some thoughts and notes about an experiment in “merging” directivity.
This should come with pictures and graphs, but the setup has already been dismantled for the next attempt.
If a driver is used below the dipole peak, the directivity will be the figure of eight dipole pattern, if the rear radiation is the same as the front.
Controlled directivity waveguides (CD) are available with varying directivities, such as 60, 75, 80, 90 degrees.  This is the included angle i.e half of each of those figures is to each side of the on line axis.
The quoted angle is the –6dB point.
The minus 6dB point for the dipole pattern is… someone please correct this… an included angle of 120 degrees.
CD waveguides lose pattern control below a certain frequency based on the diameter of the mouth. i.e  the directivity flares out to be greater than the specified angle, and as the frequency is lowered, to approach 180 degrees.
Typically CD waveguides are only used where the pattern is controlled.
An attempt was made to merge these directivities.
The mid was a Boston Acoustics 4.4LF with an outside diameter of 100mm.
It was essentially free mounted, but up against the waveguide which was a Parts Express 6” X 6” elliptical with a horizontal angle of 80 degrees.
Edge gives a dipole peak at approximately 2500 Hz, which proved to be so with measurement.
However, the off axis response drops off due to the beaming of the driver. i.e. at 2500 Hz the pattern is narrower than the dipole pattern.
The waveguide can maintain pattern control down to 2200 Hz  i.e. 80 degrees.
But there will be a frequency below that where the included angle is equal to the narrowed, due to beaming, dipole pattern of the mid.
With the setup I used, and with the non- ideal rear radiation pattern, this was at 1800 Hz.
Very steep digital filters were used e.g. 96 dB per octave, so that, essentially, no frequency was produced simultaneously by the two drivers.
So, the dipole pattern was maintained up to the onset of beaming where it progressively narrowed to become the CD pattern.
I so wish I’d saved the polars.
Whether this is a desirable outcome is obviously a matter of preference.
It sounded good to me, and the imaging was especially good, but as I said, I’m on to the next iteration.

Regards,

David
 

[Rudolf]

For a dipole the SPL does not change linear with angle. There is almost no change from 1-30°, but dramatic change from 60-90°. It is somewhat progressive.
If you look at Geddes WG speakers - they show an almost linear SPL change with increasing off axis angle.

So you can merge the SPL diagrams of both systems only for one angle exactly. They will diverge at lower and higher angles. This makes me wonder why some people are so eager to combine horn-style waveguides with true dipoles (apart from having no rear radiation with the WG).

Rudolf
www.dipolplus.de

[Saurav]

Because there is something special about the sound of a good CD tweeter When I had a compression tweeter, good recordings reminded me that a cymbal has a drummer hitting it with a stick, it's not just something that goes psssh, there's a percussive attack to the sound. It's hard to describe. As much as I like the sound of the dipole Neo3, it does not sound as dynamic as the horn tweeter it replaced.

So you can merge the SPL diagrams of both systems only for one angle exactly. They will diverge at lower and higher angles.

True. So the question is, how far off can they be before it starts to matter?

[D OB G]

Hi Rudolf,

I take your point.
However, the 6" X 6" waveguide is surprisingly close to linear to almost 30 degrees.
It's my fault for not including the polars.
The success of this experiment relied entirely on serendipitous geometries.
The driver had to be small enough for the dipole peak to be sufficiently above the waveguide cutoff.
The driver had to be large enough to beam by the crossover freq.
The waveguide had to have desirable coverage.
We start at 200 Hz with full 120 degrees coverage.
We end up with 80 degree coverage above 2200 Hz.
But the transition does not show the normal "pinched" polars that a non-optimal coupling of waveguide with dipole would show.  i.e. no flaring.
I'm not saying the off axis pattern of the waveguide is identical to the dipole pattern, but with this configuration, which relies on the beaming driver, and the steep filters, the responses track beautifully.

Regards,

David

[Rudolf]

Because there is something special about the sound of a good CD tweeter ... It's hard to describe. As much as I like the sound of the dipole Neo3, it does not sound as dynamic as the horn tweeter it replaced.

Aaah, someone who can compare (which I can't). But I believe that I understand what you are talking about. Who will be the first to start a thread about two compression drivers with WGs in dipole alignment? 

But the transition does not show the normal "pinched" polars that a non-optimal coupling of waveguide with dipole would show.  i.e. no flaring.
I'm not saying the off axis pattern of the waveguide is identical to the dipole pattern, but with this configuration, which relies on the beaming driver, and the steep filters, the responses track beautifully.

David,
I see that you got my concerns covered. If you have arrived at a smooth sonogram (or a smooth couple of polars along the frequency band), I am the last to argue. You say that you are on to the next iteration. Will be fun to see what kind of system emerges.

Rudolf

[Saurav]

two compression drivers with WGs in dipole alignment

I've seen some people who have tried that. This is where I would also throw in my "How much does real dipole response (or identical front and rear response) matter, vs. having a general match in front vs. rear power response".

I've tried dipole midrange with just a front-firing WG tweeter. Didn't like it. Of course, that was a long time back, and for most of that time I had a passive XO, which I'm sure was less than optimal. I didn't spend much time with that driver combination and the DCX. But I still think it needed *something* on the back. But instead of trying rear-firing tweeters, I moved on to the Neo3.

If (when?) I try this again, I would probably try something like a bullet tweeter on the back. Hopefully that would work better than a regular dome tweeter on two fronts - closer sensitivity match to the front tweeter (I'll probably have both paralleled on a single XO / amp channel), and narrower dispersion around the mid-tweeter XO region.

[D OB G]

I’ve tried to do the same thing with the 6ND430.
The first breakup is at 3.5 kHz.
The major breakup is at 5 kHz.
The driver is mounted almost nude, which gives a baffle peak of 1 kHz.
This makes integration with the waveguide difficult.
There is a peculiarity in that no matter how wide the baffle i.e. with a baffle peak anything below 1 kHz, there is a bump or peak off-axis at 1 kHz.
Is it a peculiarity of the driver?
Is there actually a resonance at 1 kHz, which only appears off-axis (the driver does start beaming at 900 Hz).  (there is a hint of something going on at 1 kHz on 18sound’s graph).
Is it the result of some resonance in the basket at the rear?
Would I expect a suck-out at the front instead if this was the case?
Any ideas?

Saurav, this rear radiation is an interesting issue, isn’t it.
I have only tried front and rear dipole dome tweeters and waveguides.
I haven’t tried a true dipole.
My room is asymmetrical, with the speakers mounted asymmetrically in it.
This might account for the fact that while front and rear drivers have increased the sense of spaciousness, they have also reduced the image accuracy, and caused a deterioration in the transient response.
So, in my set-up, the front only waveguide actually gives the best results!  (especially when I can get the directivities to merge  ).

Regards,

David

[scorpion]

David,

I actually have the same situation with my 6ND430. It's an effect that I now observe with the 20 cm baffle width. It never called my attention when I played with the 40 cm wide baffle with assymetric placement. But now I have a sharp (Q=4) - 5 dB notch at 1 kHz in the DCX2496 to correct at the listening position.

The on axis dipole peak for a nude 6ND430 should be about 2 kHz, I suppose, because the approximative circular radius I calculated for 6ND430 is 8.7 cm. That's also why I don't consider my 20 cm wide baffle to be detrimental to CD performance if crossing sharp at 2 kHz as I do and listen a bit off axis.

/Erling

[D OB G]

I've had a go at optimising the Xover for the handover from the 6ND430 (6 1/2") to the Parts Express 6" X 6" 80 deg X 50 deg waveguide.
Ideally the mid needs to be used below it's baffle peak.
I'm testing with it set nude, mounted by the magnet.
The waveguide is also mounted nude, separated from the mid by 10 mm.
The baffle peak measures at 1400 Hz.
However, the rule of thumb with waveguides is that the mid and the waveguide should be about the same diameter if the waveguide has 90 deg coverage (equivalent beaming from the mid assumed), and this waveguide should be crossed over no lower than 2200 Hz to maintain pattern control.
With these measurements I've crossed over at 2400 Hz.
I'm prepared to say that it is not easy to identify that freq from the graph.   
Between 1400 Hz and 2400 Hz I'm under the impression that diffraction effects might show up off axis.  It seems to me that it is pretty consistent there, maybe due to the moving area being so close to the baffle area.  (In fact it is below the baffle peak that the response is "wobbly".  I don't know if that is just the test environment, or something else going on).




The graph shows 0 deg to 90 deg in seven steps at 600 mm on the tweeter axis, one third octave averaged.  (I can't seem to get Praxis to display more than seven lines on the same graph- anyone know how to?)

Regards,

David

[gainphile2]

Ideal OB is -6db at 60deg, while a 90deg waveguide would be -6db at 45 deg.




Polar plot 




Compare with using dome tweeters:




So certainly it makes sense to use a waveguide. The lack of rear radiation can be overcome by crossing low enough (like 1khz).

[Rudolf]

Gainphile,
of course your wageguide directivity is something to strive for. But your tweeter diagram looks a bit pessimistic. The diagram below shows what I get for my actual system. The tweeters are twice as large as my next iteration. So there is still room for improvement:



Rudolf

[Saurav]

The lack of rear radiation can be overcome by crossing low enough (like 1khz).

What does this mean? How does crossing lower overcome the lack of rear radiation?

[D OB G]

I've finally rearranged my room, at last. 

The distances to the rear wall (about 2.5 m), and side walls (about 1.5 m) are now pretty close to producing a symmetrical layout (room is still asymmetrical).

The system now sounds unequivocally better with rear firing tweeters.

Regards,

David

[gainphile2]

Rudolf, that's quite ok for domes. Could you please post a pic of the arrangement? I'm interested to know the dipole distance between front and rear tweeters. Mine is worse because the baffle width is only 19cm hence the "klingon battleship" polar response.

Saurav, the lower the M-T xo, the less sensitive the ears in "expecting" rear radiation. Same with the extreme highs. I observe that 1-3khz is a sensitive region which caused audible "discontinuity" when rear spectrum is different.

I recently built waveguide box system, which becomes omni at 500hz or so region and there is no such "discontiuity" audible. They do not sound "boxy".

Of course a full front-to-back would be much better but this is costly and presented construction difficulties.

[Saurav]

Saurav, the lower the M-T xo, the less sensitive the ears in "expecting" rear radiation. Same with the extreme highs. I observe that 1-3khz is a sensitive region which caused audible "discontinuity" when rear spectrum is different.

Interesting observation, thanks. I hadn't thought about that.

[Rudolf]

Rudolf, that's quite ok for domes. Could you please post a pic of the arrangement?

The tweeter dipole consists of two Monacor DT-25N www.monacor.de/typo3/index.php?id=84&L=1&spr=EN&artid=3574&typ=u&hkatid=&ukatid=&brand=

The tweeter dipole is 6.6 cm in diameter and 4.2 cm deep.

Next iteration is a swinging system with Peerlees HDS-PPB 4-25/08 and Dayton ND20FA-6 tweeters. The tweeter package is 3.3 x 3.3 x 3.0 cm:

.

Measurements of this package had been presented here:
http://www.diyaudio.com/forums/multi-way/161299-directivity-dipole-tweeters-4.html#post2125177.