After hearing a demo of these speakers, I recommended them to a friend, as a good-for-the-money alternative for his stereo system. Nice highs, although no real bass. While I didnt really like the mid-range that much, I was confident I could get that working right. So, he got them, and after some breakin turned them over to me for modification.
Since then, I have thoroughly tested the speakers, using Liberty Audiosuite, and have completely dismantled them, to check out their construction. To summarize whats needed, in approximate order of importance:
Tweeter: this dipole has a fairly odd dispersion pattern. Straight ahead, there is a deep "suckout" in the response, centred at about 16 kHz which is 30 db down! Its very narrow in range, however, and even at 7 degrees off-axis, is mostly gone. In sum, the tweeter output is fairly smooth and stable up to about 10 kHz, above that it varies significantly with listening angle. (Its output is about the same, on average, as the woofer, so some of the usual crossover options to smooth the output cant be used.)
Woofer: pretty average, below about 3000 Hz. Above that, the response falls off rapidly, then rises again steeply, likely due to cone breakup (waterfall plots show a strong resonance centered around 5 kHz.)
Total amplitude of the dip and peak is more than 24 db!
Not much bass, but thats no surprise. There is a significant peak in output at about 100 Hz, since the box is really too small for this driver. Some effort to reduce that is worthwhile.
The stock crossover doesnt do much to control the driver problems. Its nominally first order on both drivers: a 1 mH coil on the woofer, and a 6.2 uF cap on the tweeter. What with that peak in the woofer response, its just as loud as the tweeter at 5000 Hz. Since the drivers arent even nearly in phase, the result isnt too impressive.
So, having thoroughly tested the drivers, and developed the Calsod models, I started trying various options. A quick examination of the test results convinced me there was no way to use my favourite first-order crossover designs: the drivers just dont have the operating range needed, and in any case are not time-aligned.
I worked for a bit in trying to smooth out the tweeter, but soon gave up. Its dispersion pattern above 10kHz is just too irregular to do much about. Below 10 kHz, however, its quite stable, so I decided to just work on a smooth transition with the woofer.
I decided to go as far as practical in the direction of steep rolloffs, to reduce the crossover range. A fair bit of Calsod simulation demonstrated that a sixth order crossover at about 2000 Hz was quite practical, by taking advantage of the driver rolloffs. Following my usual practice, I developed a crossover by adding the various needed functions one by one, then adjusting and optimizing to simplify. The result is quite simple, and meets the following criteria:
Tweeter crossover: (Note that the tweeter phase is reversed)
C1= 7.2 uF
| | + C2= 10 uF
L1 Twtr L1= 0.3 mH
| | -
L2= 1 mH (same as original, but I
| | | recommend an aircore )
| | | L3= 0.52 mH
| C4 | + C3= 22 uF
C3 | Woofer C4= 38 uF
| R1 | - R1= 1.5 ohms
| | |
(If anyone wants to use other types of components, of course, go ahead.)
C1 can be the original component, with a 1 uF polypropylene bypass
C2 should be at least a cheap polypropylene, such as Solen or Bennic
L1 can be a cheap aircore, such as Madisounds cheapest line.
L2 and L3 should be aircores, at least equivalent to Madisounds Sidewinder series
C3 and C4 can be something cheap, such as Bennic bipolars
R1 can be a cheap power resistor, 15 watts is plenty.
For the internal layout, I just separated the coils as much as practical, and arranged them so their axes were at right angles. All crossover components glued into the cabinet. For internal wiring, I used some teflon-insulated wire I had available, Id guess about 16 gauge. Im sure anything above 20 gauge or so should be fine.
The tweeter dispersion above 10000 Hz is not even, but it actually sounds pretty good, I suspect because of its dipole characteristic. Its putting more energy into the room than a normal tweeter would, at a given spl. Its worth playing around with speaker room position and toe-in. Based on my listening trials to date, I would aim to have the speakers at least 18" from any wall.
The woofer cabinet is cast aluminum, with a stiff plastic front. Its structure is fairly good, but the aluminum does tend to ring. I simply cross-braced it at a couple of points, with 1/2" dowel, glued in place with "Liquid Nails" adhesive.
Cabinet stuffing was worth more attention, to get rid of the backwave, while maintaining the basic box resonance, for some semblance of bass. I used pieces of 1/4" thick felt to divide the cabinet up into sections that had non-parallel sides. Many possible variations exist: the basic idea is to eliminate the parallel wall resonance problem, by positioning felt at angles across the cabinet. Of course, you must not seal off any portion of the cabinet completely. More Liquid Nails to hold the felt in place.
Since my friend wanted to use the speakers alone, without a sub, I tried to maintain some bass, while reducing the 100 Hz peak as much as possible. I liked the effects of plugging the shorter of the two speaker ports, and leaving the other operational.
If you plan to use the speakers in combination with a sub, then by all means plug both ports, and stuff the box with polyester batting or similar. That will reduce the 100 Hz peak by several db, at the cost of any bass below 80 Hz or so.
The top tweeter location results in several diffraction problems: my friend insisted on maintaining the stock cage arrangement, but I was able to test out several variations. Felt on top of the cabinet helped a bit, but high density felt is needed to get a significant reduction in the reflected sound. (BTW, Libertys impulse measurements are *very* nice for testing the various options!)
The final result: very nice! A very clear midrange, and the dipole tweeter has a very "open" sound to it. Overall, well worth the cost and effort.
Larry Van Wormer
Last revised: August 12, 1999