By James Lehman
11 Feb 1996
Printed from SpeakerBuilding.com, 09 Sep 2010 06:24
URL: http:///content/diy/1035/
I would estimate the total cost of the stereo pair, all materials included, to be less than $600, and an awful lot of work.
The major concept of this design was the idea of putting four woofers in an actual mechanical, and electrical, series parallel arrangement. The benefits of this design are that 8 ohms impedance is maintained. The box volume required for all four of these woofers is the same as that required by just one of them. The power rating of the woofer system is quadrupled and the effective cone radiating area is doubled.
If the MCM 55-1190 (woofer) is put in a box of the right volume with a properly tuned port, it can respond down to about -3 dB at 19 Hz. This doesn't mean a whole lot coming from such a small cone, but if you double the cone area, it becomes quite a bit more effective. Putting four woofers in a dual isobaric arrangement also offers the benefit of averaging any variations in each of the woofers rated specifications.
During the design phase of this project, I had the added advantage of having some friends at Audio Technica. I was able to take the baffles into AT, with the four woofers mounted on them, and a small test box, and test them as one working woofer unit, to determine their actual T/S parameters. Before I took them in to be tested, I exercised them with about 18 volts rms of sine wave at around 30 Hz for 3 1/2 hours or so. The results that I got were a bit different than the parameters listed (for just one of these woofers) in the MCM catalog. I would assume that the difference is a result of the mass of air inside the chambers being added to the mass of the cones. With just a slight adjustment to the port length, I was able to get the low end response curve that I was looking for.
The picture of the response curve Fig. 3. The response curve, is a mathematically derived prediction, and not an actual measurement.
The estimated effective cabinet volume is 5.55 cu. ft.. The inside dimensions would indicate a volume of 7.02 cu. ft. (17.5" W x 36.5" H x 19" D), but estimated volumes of all of the stuff inside the box were subtracted from this figure. I used 5.55 cu. ft. as a good guess, and a 10" long, 4" id port to tune the box to 21.75 Hz. I figured the response curve for a box 20% larger, and 20% smaller, and it had little averse effect on the curve.
The boxes were made of 3/4" thick particle board with dado-rabbit joints in the corners. A second layer of 1/2" thick particle board was laminated onto the outside of this box - making the total thickness of the box walls 1 1/4". The woofers, mounted on the fronts of the baffles, are mounted onto the surface of the 3/4" thick inner box. Holes, big enough to fit around the woofers' outer flanges, were cut into the 1/2" thick front lamination before it was applied. A layer of wood veneer, formica, paint, or some other covering may be applied. I'd have to take a wild guess and say they weigh about 170 lbs. each, or more.
The mid bass is isolated from the woofer enclosure by placing a 6" id plastic tube 19" long between the back of the baffle (behind the mid bass) and the back of the box. It was glued in with epoxy and sealed with siliconised acrylic caulk. A small hole was drilled in the side of it, the wires were fed through and sealed, and then it was stuffed with poly fiber fill.
The fiber fill has not yet been added to this box, nor has the 1/2" of particle board been laminated onto the outside of the box. You can see the bottom of the two rear mounted woofers, a couple of the 3/4" thick laminated (isobaric chamber) wood layers, the crossover, the port tube (4" id white) and the mid bass tube (6" id green). The plastic tubing used in this project is very heavy grade stuff - 3/8" to 1/2" wall thickness.
Once I am completely satisfied I do not need to get inside the box for a while, I will seal the gaps around the bottom panel with caulk. If needed, I could still get into it with a razor knife and a screw driver.
The baffles
Fig. 5. The baffles and Fig. 6. The baffles with drivers are pictures of the backs of the baffles in progress. The isobaric woofer chambers are made by stacking four layers of 3/4" thick particle board onto the back of a 3/4" thick baffle - making two tunnels the same size as the woofer holes, 3 3/4" deep.
Each layer is a cutout of the same figure-8 shape. The inside of the holes is 7 1/8" in diameter and the outside of the circles is 10" in diameter. The centers of the circles are 9" apart. The same hole pattern is cut into the baffles. They are glued together with wood glue, and clamped to dry. After they have dried, a generous layer of clear siliconised acrylic caulk is smeared all around the inside and outside of the chambers. The surfaces that the woofers mount against (front and back) are scraped smooth with a putty knife before the caulk sets.
In Fig. 5. The baffles, I used some other kind of white caulk, before I found the good clear stuff, which is water washable, will not irritate bare skin, and can be rubbed into the wood by hand.
A small hole must be drilled through the walls of the chambers to feed the wires in to the terminals on the front mounted woofers. The air gaps around the wires leading into the front woofers' terminals must be sealed. I used a silicone liquid gasket compound, used to seal engines - the blue stinky stuff.
I used the same blue stuff to seal the front mounted woofers and mid bass into the baffle. I applied a bead around the holes, and let it cure until it seemed dry to the touch, but still very pliable (one or two hours), before I mounted the drivers, so that they would not be permanently glued into place. The rear mounted woofers did not require any special treatment to make a good seal, because the woofer, itself, has a dense, closed cell foam gasket on the front of its flange.
The seal between the isobaric pairs can be tested by pushing on the cone of the rear woofer to displace it forward. The front woofer should move forward with it, and stay there until the rear woofer is let go.
I used a similar wood stacking technique to make small chambers behind the dome midranges and tweeters. The only differences are that there aren't as many layers of wood thickness and the last layer of wood in the stack has no holes cut in it. Both of these drivers have sealed backs, and require no box behind them at all. The reason that I made small enclosures behind them is that only a small, drilled hole (big enough to feed a twin lead wire through) needs to be sealed to maintain airtight box integrity. (Even a ported box requires air tight properties to work right.) It eliminates the need to goop up the backs of these drivers with caulk, and makes them easy to remove.
They are 12 dB per octave, fully cascaded networks. Each high pass filter section (from the bottom up) dumps into the next one above it. The center frequencies are: F1 = 250 Hz, F2 = 1080 Hz, and F3 = 4224 Hz. This design offers tweeter protection that is 12 dB/oct from F3 to F2, 24 dB/oct from F2 to F1, and 36 dB/oct from F1 down. The midrange also has the advantage of 12 dB/oct protection from F2 to F1 and 24 dB/oct from F1 down.
The coils were wound using AWG 16 magnet wire and a hand cranked machine that I made, with a turns counter. The coil formers Fig. 9. Coil formers are 1 1/4" diameter wood dowels with 3/4" thick particle board sidewalls glued together with epoxy. Epoxy was spread over the entire surface of the former to prevent the wood from absorbing moisture from the air.
All of the coils and capacitors were carefully checked for value accuracy. This was done by placing the part in series with a precision power resistor of a known value (in this case, 24.9 ohms). A sine wave, of the frequency that would make the part (L or C) of the correct value impose the same impedance on the signal as the constant resistor (R), was applied to the circuit. Then, a voltage reading was taken (with a digital AC volt meter) from across both the part in question, and the resistor, individually. When the voltage drops read the same, the part was the right value. The sine wave signals came from my computer. I wrote the sin("time") function at the needed frequencies to WAV files, and played them back through my sound card.
All of the coils that I made for these crossovers were tuned to within less than a half of a turn of wire (out of about 100 to 400 turns, depending on the value of the inductor). All of the capacitors in this crossover are rated at 100 volts and are groups of at least three. I used the values printed on the caps only as a guideline, and picked combinations of caps, that I had, to get as near to the needed values as I could (within about 2 or 3%).
The isobaric coupled woofers are wired in series with each other, and the two isobaric pairs are wired in parallel. This has the effect of joining the voice coils of the isobaric pairs end to end. Any bucking voltage (reverse damping factor), coming back off of the voice coils, is distributed across the pair, instead of to each woofer separately.
I put a switch in the line leading to the woofers' section of the crossover to be able to kill the woofers' output. I used L-pads to accommodate balancing the outputs of all of the drivers, of higher efficiencies, above the woofers and included SPDT switched non-inductive 8 ohm loads that can be switched in to replace the load and kill the output of any one of the drivers above the woofers. This makes it possible to hear the output of each stage of the system individually, with the effect of the L-pad settings, and without any interaction of the lack of the other drivers loading the crossover.
My next project is going to be The "ES Twin Iso 10". Basically, it will be a very similar design, using the 10" version of this style of woofer (MCM 55-1215), an 8" carbon fiber mid bass (MCM 55-1550), and probable the same midrange and tweeter. Of course, I'll have to get a better, more powerful amp.
Thanks for checking this out.
James Lehman
Fig. 1. Picture of the speaker
Fig. 2. Drawing of the frontplate
Fig. 3. The response curve
Fig. 4. Inside the box
Fig. 5. The baffles
Fig. 6. The baffles with drivers
Fig. 7. Picture of the crossover
Fig. 8. The crossover circuit
Fig. 9. Coil formers
