Beyond the Ariel

[Lynn Olson]

Here's one of the Bjorn's models (not using the same BEM methodology shown above) that were used during the development of the AH-425. This shows the difference between a 420 and 440 Hz LeCleac'h horn with a T=0.707 and the internal profile of the Altec 288 as part of the model.

[Lynn Olson]

Originally posted by salas
So its a 40deg symmetrical dispersion horn at 3kHz? How high are you planning to go with it?

The target crossover is anywhere between 5 and 8 kHz, depending on the mass rolloff and onset of breakup of the compression driver, and the relative directivity of the AH-425/GPA-288 vs the RAAL 140-15D tweeter.

The HF directivity of the AH-160 and AH-425 are not likely to be the same, so the actual system will depend on what MLSSA (and direct audition) says is the best crossover. Like Dr. Geddes, I prefer to listen off-axis, and the crossovers will be balanced for that.

[Salas]

Aha, wish you best experiments and seamless integration. You have been investing lots of planning and $ in options, we all want it to succeed relatively easy.

[Lynn Olson]

I don't have as much trouble with driver integration as others, so I'm not too troubled by that - but then, I've been wrestling with these things since 1975, and follow my instincts for what will and will not work. It's part of the reason I prefer passive crossovers - I can make very small adjustments to slope contours and inter-driver phase that aren't as easy with off-the-shelf active crossovers.

The main thing I look for is very careful control of inter-driver phase over at least an octave (and preferably wider) and avoidance of breakup artifacts close to the crossover region. Modern high-end speakers routinely fall into the this trap, with their fashionable minimalist 6 dB/octave crossovers combined with rigid diaphragms that have aggressive breakup products close to the crossover region - this results in poor system integration, regardless of pretty-looking polar curves, and very fatiguing sound in the upper midrange.

The $60,000 TAD Reference One, like so many other speakers in the ultra-high-end, has this problem - the coaxial mid/tweeter driver layout does not rescue the midrange from obvious upper-mid breakup and poor integration in the 1~5 kHz region. Classical music, with its broad and dense spectra, just doesn't work if the mid and HF have audible breakups in this region. Driver breakup, regardless of coaxial or MTM mounting, destroys crossover integration, no matter how clever the crossover design is.

I am much less concerned about polar curves than driver breakup - sorry if the Toole-school disciples are offended, but I don't agree with them on this issue. I put driver characteristics first and foremost in where to choose crossover frequencies, or if the driver is even suitable at all. The performance of the GPA 288 remains to be seen in this context, but I am optimistic based on what I've heard and measured so far. Similarly, the RAAL has a plenty of power-handling at the lower frequencies, unlike other ribbon tweeters.

Some designers work from the outside in, starting with a clean sheet of paper and a set of preferred parameters for the overall system. I work the other way around, finding the most attractive drivers I can get, and fit the system design to what the drivers can and can't do. It is clear that compression drivers (and horns) have less desirable behavior above mass rolloff, and that some diaphragm materials are preferable to others.

[Kolbrek]

Originally posted by John Sheerin
Do you know how much of the driver that model includes?

The model is just the horn, no driver parameters included. The horn is driven by a plane piston the size of the throat, with constant velocity.

It is simulated with the axisymmetric version of Stephen Kirkups
BEM package, using 6 elements pr wavelength.

Bjørn

[Salas]

Originally posted by Lynn Olson
I don't have as much trouble with driver integration...

...The main thing I look for is...

...It is clear that compression drivers (and horns) have less desirable behavior above mass rolloff, and that some diaphragm materials are preferable to others.

Lynn, this is what I call a thorough answer. Good luck again for fast results with no costly surprises.

[fivestring]

Originally posted by Lynn Olson

Similarly, the RAAL has a plenty of power-handling at the lower frequencies, unlike other ribbon tweeters.

That`s because it is flat (non corrugated). I just finished my 1 meter tall ribbon and tried the "old fashioned corrugated" way and it sounded lifeless and with that typical nasty artificial high frequency sparkle and sizzle, common to all corrugated ribbons.
Then I just re tensioned the ribbons to the point of almost perfect flatness (the corrugations were still visible and contributed somewhat to the overall structural strength) and voila, a complete transformation of sound! The whole sound gained tremendously in control, weight and gone were the grayish sounding "sparkle and sizzle". Alexander is right, corrugated ribbons have no control in the lover passband of their operating range.
Now, you still want to get rid of that metal noise alu foils produce when totally flat, here embossed patterns help.

[Lynn Olson]

Run time for the AH-550 was 28 hours, simulated 100 Hz to 15 kHz in 250 steps. Here's the polar diagram:

[Lynn Olson]

As before, the black line is the percentage of power that goes into the horns and is radiated into the room, while the red line is the power reflected back to the diaphragm.

[Lynn Olson]

The black line is the resistive term, the red line is the reactive term. The flat portion of the black line represents the region where the diaphragm of the driver is in constant-velocity mode (direct-radiators primarily operate in a constant-acceleration mode).

To expand on this point, constant-velocity drivers increase the diaphragm excursion at a rate of 6 dB/octave as the frequency is decreased, while constant-velocity drivers increase their excursion at a rate of 12 dB/octave as the frequency is decreased.

One of the most important functions of a crossover, aside from driver integration and equalization, is control of excursion. Many designers forget that direct-radiator tweeters operate in constant-acceleration mode, thus excursion continues to increase below the crossover point if a 6 dB/octave highpass filter is chosen. The tweeter is in the awkward situation where excursion continues to increase at a 6 dB/octave rate in the region between the crossover point and the Fs of the tweeter - excursion only decreases below Fs. Although the acoustic output of the tweeter is decreasing in the two-octave band between Fs and the crossover, the excursion is unfortunately increasing. (This is why I've always avoided 6 dB/octave crossovers, and prefer Gaussian or Bessel highpass filters of moderately higher order, typically 12 dB/octave.)

This unwanted increase in excursion is the source of an annoying intermittent distortion in many high-end loudspeakers with minimalist or linear-phase crossovers, since the tweeter excursion is excessive in the very critical 500 Hz to 2~3 kHz region. The problem is especially troublesome because orchestral music has the maximum spectral energy centered around 500 Hz, the same as the Fs of many tweeters.

Similarly, with a horn speaker crossover, the excursion curve has to be kept in mind - the excursion of the diaphragm increases many times once resistive horn loading is lost, which is why I'm seriously considering a notch filter tuned to the Fs of the compression driver - not so much as to twiddle with the response curve, but to keep unwanted out-of-band excursion to a mininum.