Rhyos
Drivers: The Vifa M18 woofer was chosen because of its robust construction and reputation for possessing sound qualities similar to the venerable Scan-Speak 18W8545 at less than half the cost. The Scan-Speak D2905/9000 tweeter is a classic with a rich and dynamic sound. It's ability to easily handle a low crossover point made it an ideal mate to the M18. The S-S 9000 tweeter is, however, infamous for the sticky coating on it's dome. Mine had collected a nice assortment of lint, fuzz, and debris. The 9000 (which is now out of production) and the more expensive S-S 9300 share identical motors. For a modest fee the good folks at Madisound did a great job of replacing my old sticky domes for new 9300 domes. Now I've got nice, clean domes, a lower Fs, and better overall sound to include more high frequency extension. For all intents and purposes this tweeter is now a 9300.
Enclosure: Construction of the enclosure began in the Spring of 2000 and did not finish until later that summer (this was an on and off project). The vented 20 liter enclosure was built into the upper third of the floor-standing unit. The middle section holds approx. 25 lbs of dry sand and the bottom section holds the crossover (with removable rear cover for easy access). The cabinet was built out of .75" MDF with a .75" plywood brace (bracing sides, top, and bottom). The front baffle, however, was built out of 1.75" MDF. To get this thickness I laminated one .75" and two .5" sheets together. The entire enclosure is held together with aliphatic (yellow) wood glue and soooooo many biscuits, I lost count. No screws. Not that I have anything against screws, it's just that I did not use them.
The Rhyos uses a stepped baffle (not to be confused with a baffle step) to help bring the impulses of the woofer and tweeter closer together. However this step is only 1" and does not fully bring the impulses into alignment. But it helps a little and it kinda looks cool. Unfortunately, a stepped baffle creates a lip below the tweeter. With gated time window measurements, this lip can be seen as a major source of diffraction. However, I could not hear this during listening tests. Nevertheless, I went on a quest to eliminate the diffraction caused by the lip. All types and shapes of foams, felts, and rubbers were tried but the best, by far, was a wedge of MDF. The key was to start the wedge just below the dome of the tweeter and finish it flush with the lip. This semi-circular wedge was sanded smooth and painted to match the cabinet.
Since the front baffle is so thick, I used a 45
deg chamfer bit in my router to carve scallops around the woofer cut out on the
rear of the baffle. I left the area near each T-nut untouched. I
like to call this "back-chamfer".
Woofer Brace. Just for fun I added a woofer brace. This is a device that applies outward clamping pressure to the rear of the woofer magnet as well as the plywood brace and the rear of the enclosure. The idea is to add extra stiffness to the enclosure in and around the woofer as well as hold the woofer absolutely still. It is said that a well implemented brace can increase bass attack and midrange clarity for far less money than fancy cables or interconnects. I found only a small difference in sound quality with my woofer brace engaged. This could imply that woofer braces don't do much in most situations. However, I believe that my box is very sturdy to begin with and the heavy cast frame of the Vifa M18 does not need extra bracing. Further tests on stamped frame woofers might prove worthwhile.
From top to bottom: 20L Woofer enclosure, 25 lb. sand enclosure with PVC
fill port, crossover enclosure with rear-facing entry panel removed.
Woofer Brace details. The brace is constructed like a 3-point floating mirror cell. A small triangular piece of wood holds three .75" dowels, 1.5" long (one at each corner). The rear of this triangular piece of wood rests centered on a long 3/8" carriage bolt which runs through a T-nut in the plywood brace and through another T-nut at the rear of the enclosure, terminating in a knob outside the cabinet.
Stuffing. I used .5" to 1.0" of natural combed cotton batting on all interior walls of the enclosure (and some on the brace).
Paint. My first tests with Duplicolor Truck Bed Paint failed miserably. If you live at high altitude and try to use this stuff on a hot, dry day, forget it. In order to get an even coat, you have to hold the can about 2 feet away from your work. This stuff will literally dry in the air before it hits the work piece. Most of it ended up as dust on the floor and the paint that did make it to the test piece created an very uneven texture and appearance. Instead of high volume/low pressure, I got low volume/high pressure. Maybe I got a bad batch - who knows. I did have excellent results with Behr Premium Satin latex paint. After thoroughly sanding and preparing the box I put down 4 coats with a small roller and then a final coat with a sea sponge to get a very attractive textured effect (stipple). People who look at these speakers say "Whoa, what what did you make the box out of, metal?" Or, "Is that leather?" I take that as a compliment.
Grill. The grill is made from .25" thick Cherry panels. As you may already know, just about any hardwood board .25" thick and 7" wide will have some cupping along it's long axis. I used this to my advantage to create grills that have a nice curve to them - as if they are trying to wrap around the front of the enclosure. Not happy with the complexity of my enclosure so far, I decided to attach the grill not with hedlocks, or velcro, but with magnets. Four .5" magnets were recessed into each baffle then filled, sanded, and painted. The are totally invisible to the eye. Each grill has four .5" cherry "stand-offs" (dowels) that are tipped with a mating .5" magnet. Just hold the grill right over the baffle face and "fwip!" the grill is securely pulled onto the baffle.
Crossover: I did not use any crossover modeling software to create this filter. Instead I arranged a test jig crossover board in front of my monitor so that I could create a real crossover then press a button on my keyboard and let LspLAB take an actual measurement of the test crossover with the speaker. In order to do this I first had to purchase a large collection of inductors, capacitors and resistors (which cost about the same as crossover modeling software). I feel this technique is a good "hands-on" learning aid but I will eventually want to purchase some nice crossover modeling software like LspCAD.
The tweeter circuit is a simple 2nd order electrical. I'm told that the S-S 9x00 series tweeters are a joy to work with. Mine proved to be no exception. I had no problems tweeking the roll-off to whatever I needed to match the woofer. It crosses over at 2k without batting an eyelash. However it does get a tiny bit soft above 15k. Response above 15k could be lifted by bypassing the series resistor with a 1 or 2 mf capacitor. I did this and heard no difference, though it can be easily measured.
The woofer circuit is a simple 2nd order electrical. Since the M18 has a continuously rising response, (well, sort of) I was able to use a large inductor with out a bypass resistor for baffle step compensation. I was able to get a decent response and good roll-off without a Zobel. For a short time, however, I did try adding a damping resistor in shunt just before the woofer. Why may you ask? The included plots show that the M18 has 3 dB hump centered around 1k. Adding a 30-40 Ohm damping resistor did drop the overall sensitivity of the woofer by about 1 dB, but it also had the added effect of almost completely knocking down the 1k hump. I think it has to do with the cone edge resonance that the M18 exhibits (like most woofers). This can be seen as a blip on the impedance plot at about 1k. On some woofers this will cause a dip (usually about -3 dB). On the M18 it causes a hump at 1k then a dip at 1.5k. My guess is that since the damping resistor is creating a mild short in the motor and increases electromagnetic "braking" which lets the motor really clamp down on the cone and keeps things from getting out of control. It damps at all frequencies, but for some reason it seems to damp more at the edge resonance point. Not a very scientific explanation, I know. I should also note that I tried series notch filters, parallel notch filters, placing strips of tape, foam, clay on the cone, etc... nothing works as elegantly as the damping resistor. If you build this circuit, I strongly urge you to try it with and without the damping resistor. If you remove the damping resistor from the woofer you might want to lift the pad on the tweeter by an Ohm or two.
Ultimately, after months and months of listening with various amps, I settled on one crossover design. This design sounds best with my current amp - a Golden Tube SE-40 Special Edition. Don't ask me why, but I like the Rhyos better without the damping resistor on the woofer - even though it measures better. Others (Dave Elledge) preferred the damping resistor. Try it yourself and see. I am very happy with this set up.
Tweeter:
Caps: North Creek (NC) Zen 6uf, NC Zen 1uf and NC Harmony .22uf.
Inductor: North Creek 14ga .3mH.
Resistors NC 10W 2.49Ohm and 10Ohm.
Woofer:
Cap: Bennic 16uf.
Inductor: NC 12ga 2mH
Measurements:
Combined response of near-field
port, near-field woofer, and 2 meter on axis. I wasn't too scientific with the
combining of the three plots, but it looks about right to me. Ignore dB
scale on all plots. I estimate the sensitivity of this speaker to be about
86dB.
Multi-plot showing on-axis response,
tweeter only, woofer only, and response with tweeter wired out of phase. I
was shooting for a crossover point of 2k.
Ignore dB scale.
Horizontal response plots over a 60
deg range. Not great, but not bad either. Sounds just fine in my
living room.
For fun I added in an in-room response (with some
smoothing). Things look pretty decent to me though the midrange.
However, around 500 Hz and lower, room interaction starts to become a factor.
System impedance. I hit my
target port tuning of 35Hz but I think 40Hz might have worked better. I
have no complaints about the bass response though. In a smaller room
this speaker would have too much baffle step compensation. Ignore the noise
above 10K.
Just for fun I superimposed a plot I
made of the Vifa M18 with a measurement Phil Bamberg made of the Vifa PL18. The two
plots have been offset for readability (ignore sensitivity scale). The M18
plot shows a drooping bass response due to the short gated time window I to
used. I really see no huge difference in the two plots other than the
location of the cone edge resonance. The M18 has it at 1.1k and the PL18
has it at 650Hz. I think the pentagonal shaped cone of the PL18 may help
in this regard. As you can see, the PL18 plot above is nothing like that
published by Vifa.
Sound: Subjectively speaking I think the Rhyos loudspeakers sound pretty darn good. Great dynamics and bass output for a 2-way. These speakers live in a fairly large living room so I've got a fair amount of baffle step compensation built into this design. The bass can get a bit on the full side if I push the speakers up near the front wall. (For critical listening, the speakers come out about 3-4 feet from the front wall.) I find the top end (treble) to be totally non fatiguing. I sense that there is a tiny bit of excess midrange energy that comes through on some recordings. I guess that's the thing about the M18 - just a tiny bit too much at 1k but excellent detail, warmth, and dynamics. Imaging and soundstaging seem pretty good but are very dependent on room placement. (Amplification for listening tests: Rotel RA 985BX 100w integrated amp; Golden Tube SE-40 SE. Source: Toshiba SD-1200 DVD player; Rotel RCD-971. Home-made interconnects and home-made CAT 5 speaker cable.)
One of these days I'll get a decent digital camera.