An exploration in eliminating driver lobing

[boris81]

So somehow I became interested in eliminating driver lobing at the crossover. It looks ugly in polar graphs I suppose..

I tend to be biased against the obvious solutions of using a fullrange or coaxial speaker so the most sensible approach is to use a very low crossover point and put the drivers as close together as physically possible. Basically what happens is as drivers are spaced further apart and the crossover rises in frequency cancellations begin to form on the vertical axis. With many popular designs the listening window can be as small as 20 degrees vertically. Xdir is a great FREE program that can simulate the directivity pattern of a given crossover scenario.

Below is a crude simulation of what this phenomena would look like if say you built a speaker that can color the air in neon green and blue. On the left, 2 point source drivers interact destructively at high frequencies to paint a nice psychedelic blooming neon.. semi-flower? On the right, the same 2 drivers interact well together at low frequencies.
2b.PNG


I built a quick prototype to test this idea. I decided to use the Vifa NE19VTA and the HIVI B3S because of their small sizes, low price and excellent reviews on Zaph Audio.

vifa-hivi.jpg

When it comes down to cramming 2 drivers close together how close is close enough? I'm going with the assumption that the lobing effects are negligible when the centers of the drivers are spaced less than a 1/4 of the crossover frequency wavelength. Following that rule Xdir informs us that cancellations will tame the response at 90 degrees off-axis by -3dB. I think that is acceptable even by hardcore audiophiles and as I will show later, real life results are often much better than expected.

The distance between the 2 drivers in my setup is 60mm requiring a crossover point of 1400Hz. I cut a piece of the tweeter flange but the measurements look so promising I think I could've kept it in it's entire glory. The measurements show 30, 60 and 90 degrees off center, 1m away on the tweeter. The on axis frequency response is missing because it looks almost identical to the 30 degrees curve and HOLMimpulse lets me have only up to 3 curves.
vf-vert-30-60-90.png

I like what I'm seeing (and hearing) a lot and I am going ahead with a 3-way implementation. I'd like to hear some comments from you guys!

[planet10]

Originally Posted by boris81

When it comes down to cramming 2 drivers close together how close is close enough? I'm going with the assumption that the lobing effects are negligible when the centers of the drivers are spaced less than a 1/4 of the crossover frequency wavelength. ...

The distance between the 2 drivers in my setup is 60mm requiring a crossover point of 1400Hz.

In my MTM i achieved <1/4 wl by moving the XO down to <400 Hz. (6.5" modwoofers/3" mid-tweeter)

dave

[boris81]

Originally Posted by planet10

In my MTM i achieved <1/4 wl by moving the XO down to <400 Hz. (6.5" modwoofers/3" mid-tweeter)

dave

MTM designs have the worst vertical directivity because the 2 woofers are spaced too far apart to work constructively into high frequencies. At 400Hz you can have the same vertical directivity with a single 10" to that 3". What was your idea about using 2 drivers instead of 1? Can two 6.5" woofer combine to have a lower distortion in the bass region than a single 10" at comparable cost?

[tuxedocivic]

How does the tweeter not fall off the chart above 10khz and 90 degrees off axis?

And what kind of XO did you use? I would assume that a shallow XO would extend the useable portion woofer higher into the frequency range causing lobing to occure, but if the XO is steep, then it'll be less noticeable. This this has to be a part of your experiment I think.

[planet10]

Woofers are less than 1/4 w/l apart at the XO (which calcs closer to 300 than 400 -- with 1st series XO i don't know exactly -- it would be considered a FAST). Midtweeters are closer than 1/8 w/l from each midwoofer. I've always wanted to do an MTM "right". Woofers were a nice price ($60/pr), i liked the way they looked. They give lots of nice finessed bass. Midtweeter has what is essentially a 3/4" tweeter in the middle, it has claimed extension to 32kHz, and has few issues with directivity. If they are here when you next visit (Chris is thinking of borrowing them to play with bi-amping using a no feedback SEP on the top, and then finish them)

Geddes work shows that concern over distortion is overrated.

dave

[boris81]

Originally Posted by tuxedocivic

How does the tweeter not fall off the chart above 10khz and 90 degrees off axis?

I specifically picked the Vifa-NE19VTA tweeter because of it's excellent dispersion at high frequencies. Below is the HORIZONTAL 0-60 graph from Zaph Audio.


My own measurements are very similar to his. Just for reference this what I get for HORIZONTAL at 30-60-90 degrees(the on-axis is not shown but is almost identical to 30deg). I have a shelving filter for baffle step compensation and a lowpass at around 15KHz to tame the rising high end.
vf-horz-30-60-90.png

The crossover is a 4th order Linkwitz-Riley at 1400Hz, all active in DSP. I definitely want to experiment with 2nd order and higher frequencies and will post my findings here.

[boris81]

Dave,
Awesome work, I'm very impressed!!

Is that 3" one of those drivers that has 1 motor to control both the cone and the tweeter-cap? What is your impression on those?

[planet10]

The 3" is the FF85KeN. If one ignores its obvious bass shortcomings, it is, IMO, the best driver Fostex ever made (and discontinued in their wisdom)





dave

[buggsson]

Dave,

Very interesting subject. You have managed a 1/4 wavelength at the crossover frequency. Is it still 1400 Hz, or did you lower it 400 Hz?

Do you consider lobing to be a more detrimental effect than crossing over in the so called "the ears most sensitive region"? Most people seem to cross in this region anyway, so is there a problem, and how do you in that case solve it? I have tried to get a grip on this subject, but I've failed so far.

I'am planning a MTM but have choosen to go above "the region" so I have only approx. 1 wavelenth at my crossing over frequency. C-C is 90mm.

[planet10]

My crossover is somewhere between 300-400 Hz. I'd not XO between 1-4K.

dave