Introduction:

This page describes the construction of my front array of speakers (Left, Center, Right): a set of open baffle, floor standing Avro Part speakers that were designed by ThomasW and Jon Marsh of HTGuide.com. The compliment of all-aluminum cone drivers consists of a Dayton RS28a tweeter, two HiVi M8a 8" midwoofers, and two TC-1000 12" woofers. The TC-1000 drivers are no longer available; a suitable substitute is the Dayton RSS315HF-4 12" woofer (necessitating a capacitor change in the XO for the woofers). These speakers are quite large - the left and right speakers stand just about 54 inches tall and just over 16 inches wide and weigh in at about 120lbs each (the center speaker weighs closer to 150 lbs). They are quite imposing speakers by any standard and need a fairly large room in order to perform well. They present a 4-ohm load from 50-150Hz and then impedance rises as a function of frequency for the rest of the audio spectrum. Due to the 4-ohm dip of the woofers, these speakers require a fairly substantial amplifier (probably NOT your average receiver) in order for them to perform at their best. Think about driving them with an amplifier that can deliver a minimum of 75w-100w into a 4 ohm load. 150w is even better - these speakers are power hungry! I am driving them with my new Aleph-X amplifiers and they are incredible beauties! My preamp crosses the Avro speakers to the sub at about 60Hz. The overall cost to build these is approximately $850-1000 each (driver cost alone is $500 each), depending upon the cost of your crossover caps, coils, and choice of finishing materials (glue, mounting screws, wood veneer, paint, stain, trim, etc). The primary philosophy behind these speakers is to produce clean, distortion free, full range output to a level of 105+dB, hence the dual mids and dual woofers. Their capability for undistorted, high SPL output is the primary reason that I chose these speakers for my home theater system and now that they are completed, I can say that they deliver well on this promise!

Drivers:

Drivers for this project are readily available from Parts Express and include the Dayton RS28a tweeter, two HiVi M8a 8" midwoofers, and two Dayton RSS315HF-4 12" woofers. Each of these drivers features an aluminum cone, likely chosen for this design to help achieve low distortion/high output performance since the aluminum cones are very lightweight and highly rigid.

Crossover Choices:

There are two primary configurations for the Avro crossover. One option is to run fully passive and drive each speaker with a single amplifier. The second choice is to bi-amp each speaker, driving the woofer with one amp and driving the mid + tweeter with a second amp. Since I am running three of these speakers for my theater, I decided to go fully passive to reduce my needs for additional amplifiers. In comparing the two crossovers, you will see there are two different tweeter networks. I started by using the tweeter XO from the passive design, but in comparison to other speakers I enjoy, the somewhat brighter tweeter network from the bi-amped crossover sounds better to my ear and in my room when listening without Audyssey MultiEQ from my preamp. Note the slightly raised response in the 8-9kHz region and the slightly depressed response in the 11-12kHz area in the graphical chart of SPL levels for the original crossover. The newer crossover smooths these out a bit, but your own experiences may differ as the effects of your listening room will also have a large impact on the sound that you hear. Its easy enough to build both and give them a try, though if your main interest is watching movies and you use Audysses, it's probably not worth the bother of comparing. If you are purchasing your inductors, get ones with a higher initial inductance and then unwind them until you hit your target value. You'll need a good meter that can measure inductance for this task.

Fully Passive Crossover:

The crossovers are constructed from Mills 12w non-inductive wire wound resistors, 15ga DIY inductors, and a mix of Obbligato and Dayton capacitors. The Dayton's are used throughout, with the exception of Obbligato Gold Premium aluminum foil caps which were placed in the direct signal path for the tweeters and Obbligato Premium Film in Oil caps in the direct signal path for the mid-woofers. Given the relatively higher price of the Obbligato caps, I only used them where the signal passes directly through the cap to get to the drivers. Dayton caps are readily available from PartsExpress.com and the Obbligato caps were sourced from the nice folks at HawthorneAudio.com (at a worthwhile savings over the price directly from diyhifisupply.com).

Tweaks to the Passive Woofer XO:

If you end up using the Dayton driver in place of the TC Sounds driver, you need to add 50uF to each of C1 and C2 that are shown in the passive crossover schematic. This compensates for small differences in the specifications and behavior of the drivers. In addition, if you remove R2, this will increase response in the 80-150Hz area a little. With R2 removed from the crossover, reducing R1 by 1 ohm increases woofer output by 3dB. Conversely, increasing R1 by 1 ohm decreases woofer output by 3dB. Don't change R1 with R2 present, though, or you lose output in the 60Hz region.

Bi-Amp Crossover:

Below is an update promised to a few folks who want to run the Avro Part "Classic" bi-amped between woofers and midrange and passive otherwise- sometimes known as the TW Special Edition. Jon Explains:

This looks a little different from a pure passive crossover in LspCAD. The TF1 block on the left in the woofer network represents the active crossover low pass- it's two 2nd order filters with Q of 0.707 cascaded, which gives a Linkwitz Riley 4th order Low Pass, such as you might get from a Behringer CX2310. That's followed by a buffer G2 with gain of unity.

Next is the secret sauce- the outboard inline passive LF equalizer, such as might be inserted between your active electronic crossover and power amp. Here, it's been developed for a nominal load of 22 kOhm, which happens to be the input impedance of Aragon power amps. (that's R14; in practice, R14 is omitted, or if you have an amp with higher impedance, a resistor is used in parallel so that the resistor plus your amplifier's input resistance equals 22 kOhm). This can be built in mono or stereo into any kind of small chassis with input and output connectors such as RCA; to do balanced, just build a circuit for the hot and low XLR pins. Resistor should be 1% for EQ tracking precision; caps 5%, preferably film and foil or metalized polypropylene film, at least. Total insertion loss is about 10dB, so you'll have to have your LF crossover output cranked up, and your HP output turned down to balance.

That's followed by another buffer, because the passive filter can't drive a driver; that represents your power amp. Gain is unity again; in practice in the real world, each power amp stage might have a gain of 26 - 30 dB; we don't need to represent that here.

The corner frequency to the LP is set to 150 Hz. How come? Well, to get things to work nicely with those fixed inflexible active crossovers, given the transfer function of the drivers, we have to play some games a bit to get a nice crossover behavior. Due to the peaking of the woofer in the U-baffle, and the 8" open mid bass starting to give out, we can achieve a good acoustical transfer function crossover at around 200 Hz by having the electrical set to 150 Hz; in combination with the driver performance, we get more or less what we want acoustically, though I have to run inverted phase on the midrange for it to sum in the crossover region. Can't fix that in the active fixed crossover, whereas with a passive I stagger poles and zeros to get the ideal acoustic transfer function, or pretty close. (see plots further down, SPL and transfer function.

In the middle section, we take out the series cap and parallel inductor, and have to adjust impedance compensation and other network values to get the desired midrange bandpass. As much as possible I optimized for standard sized components and smaller caps or impedance compensation networks. The high pass is a LR-4 also set to 150 Hz, since we have to keep the frequencies the same with a standard analog electronic crossover.

The tweeter network isn't modified from the last Dayton RS28a design; it has the upper range response EQ tweak (R6, R10 and C12) instead of plain Lpad; this gives the RS28a a little more life in the top octave.

Crossover Construction:

I began my construction with the crossovers and saved the woodworking until the electronics were complete. I wanted this to be a DIY project as much as possible, so I opted to wind my own inductors. In the end, it ended up saving me about 40% of the cost of purchasing inductors on the retail market (which worked out to several hundred dollars) and provided a few evenings worth of DIY satisfaction.

I started with a coil winder that I picked up on E-Bay for about $50. Its rather crude, but it has a sturdy crank and a turns counter to make life easy. I took a few scraps of wood I had laying around, squared them off, and ran them across the table saw to groove their inside faces. These recesses worked very well to hold plastic wire ties I used to bind the coils when I was finished winding them. Drill a hole in the center for mounting them on the coil winder and drill a small hole to let the wire from the beginning of the coil poke out as you wind them. Next, I cut a few pieces of styrofoam board (left over from insulating the finished basement) to act as the former or core for my coils. They were cut to specific thicknesses and diameters. I used a small scroll saw to cut the cores, but you could also cut them by hand with a razor knife or coping saw and sand them smooth. After your cores are cut, its worthwhile to use a pen and label them so you don't accidentally mix them up down the road.

I ordered 4 spools (approx 4000 feet) of single build wire, 15g solid core enamel coated copper wire (nominal outer diam of 0.0587") from EIS/Fay Electric Wire in Elmhurst, Illinois. The total price was about $330 including UPS delivery. I started by inserting the wire ties into the grooves I cut with the table saw and wrapping a strip of paper around the styrofoam core and on top of the wire ties. This helps when it comes time to remove the completed coil after it was wound and helped keep the wire ties out of the way while I wound the coils. I used tape to hold the end of the wire ties out of the way as I wound and worked carefully to keep the wire as neat as I could. Using 15ga wire, I was able to keep the first 2-3 layers of wire very neat, and things started to degenerate a bit from there. I stopped every 2-3 layers to drop some super glue onto the coil to keep things from coming apart when I removed the coil from the form. When the counter hit the target number of turns, I used my LCR meter to confirm the target had been reached. With the wire I used it was as easy as using the alligator clip to scrape some of the insulation and measure the inductance. Once I confirmed the target reading, I added about 5-8 additional turns for good measure (a MUST in my book) and then closed the wire ties, snipped the wire, and removed it from the winder. If you don't add additional turns, the process of tightening down on the wire ties and removing the coil from the former reduces the measured inductance by a significant amount, requiring additional turns to get back to your target value. Adding extra turns after cutting the wire is a major pain in the neck (you have to join and solder the wires back together), its SOOOOO much easier to over-shoot the target by a few turns (maybe 5-8 turns) and then unwind it a little. A great resource for winding your own coils is the Shavano Calculator. You simply input what inductance value you want, and it spits out a table with various wire gauges, the dimensions of the coils, number of turns, length of wire required, and the corresponding resistance value of the coil. A bit of caution, these coils come out LARGE (diameter). They are much larger diameter than, say, the Erse air core coils. But, they are also much less expensive when you need to wind 24 (8 coils * 3 speakers) of them! I'm not sure its worth the trouble if you are only going to make 5 or 10 coils - it might be cheaper/faster/easier to just purchase them. Also, the coils that result from this calculator are quite large in comparison to commercially available coils. This makes placing them inside a speaker cabinet a bit more complex because you don't want them to interact with one another or with the magnets from the backside of the drivers. This wasn't so much of a problem for me since the cabinets are large and open for easy experimentation for placement.

Once I removed the coil from the foam core, I tightened up the wire ties some more and remeasured the inductance. After confirming that I had reached the target inductance value, I coated the remaining three sides of the coil with some good old fashioned Elmer's Glue to hold everything in place. After a few hours of drying in the sun, I had a pile of perfectly matched inductors.

I used 15ga wire that comes in 10lb spools (about 1000 feet). I carefully planned the usage of the wire across the various size inductors that need to be made in order to minimize the left over wire on each spool. Overall, this strategy led to very efficient use of the wire with minimal scrap at the end. Once all of the coils were wound, I assembled the crossovers. Where the caps are directly in the signal path, I used gold Obbligato foil caps for the tweeter section. The remaining caps are Dayton and all resistors are Mills (except the temporary sand cast ones in the tweeter network because I ended up short on resistors during my first build - these have since been replaced). In the XO for the mids, I used Obbligato paper in oil caps for the signal path and Dayton's elsewhere. Finally, the woofer XO uses Dayton caps exclusively. You can see where I had to add some extra wire in order to get the right inductance in the woofer XO (this is how I learned that the inductance drops once your remove the coil from the winder!). Each of the configurations shown above physically orients the coils to minimize interactions effects both across the coils and with the magnet of the drivers. I played around with my LCR meter and positioning the coils for minimal interaction before assembling the XOs. A very good resource for understanding the impact of coil orientation comes from Troels Gravesen in Denmark.

Left and Right Channel Cabinet Construction:

Next up was constructing the cabinets. The front baffle is constructed from 18mm void-free marine-grade Baltic birch plywood that is laminated (glued and screwed) to 3/4" MDF. As an aid in planning the most efficient use of your wood, you might want to have a look at Gary Darby's Cut List utility. Its a free download for a little utility that allows you to enter the size and quantity of panels you need and provides the optimum cutting layout for your wood. The local hardware store where I purchased the wood cut everything for me for about $10 - a real deal since I lack the equipment for precisely cutting large chunks of wood. Be sure to use plenty of glue when laminating the two pieces of wood together (MDF + birch ply). Use lots of clamps around the outside perimeter and plenty of heavy objects stacked in the middle of the wood (cinder blocks, quickrete bags, barbells, transformers, etc) to help squeeze out all of the air bubbles. If you don't have glue oozing from your clamped seams, separate the pieces of wood, add more glue, and re-clamp them. I used TiteBond glue that you can find just about anywhere.

Below are links to the PDFs of the baffle design, officially referred to as the mk II, v2.5 cabinet. This cabinet design is specific to the crossover posted above (and vice versa). So, if you change either the crossover, the dimensions of the front baffle, or speaker placement on the baffle, you are likely to have unpredictable results.

After letting the baffles sit and dry for 3 or 4 days, the next task was making the recessed holes for mounting the drivers. ThomasW has some great advice on how to use your router to cut recessed holes for flush mounting drivers. You need a router, a circle jig (I used the Jasper Circle Jig from PartsExpress), and two router bits: a dual fluted 1/4" spiral cut up-twist bit, and a 3/4" flat bottom carbide bit. The technique is different from the one described in the Jasper instructions. It's slightly more labor intensive but doesn't require an expensive and difficult to manage 1-1/4" router bit. It also generates significantly less sawdust.

For additional details and pictures of cutting recess holes, be sure to check out Thomas' web page on flush mounting drivers.

A word of caution about routing/cutting MDF -its VERY nasty stuff that generates ENORMOUS amounts of very fine dust that blows around very easily. MDF is held together with all sorts of glues that are best kept out of your eyes, nose, mouth, and lungs. Be sure to wear eye protection and a face mask or some sort of breathing filter. Finally, you shouldn't be ANYWHERE NEAR a power saw or a router without ear protection. There is little sense in damaging your hearing before your have the opportunity to enjoy your new speakers...

Routing out the driver holes made the baffles much lighter and easier to move around. Once the recessed driver holes were cut, I placed the baffle on a set of saw horses and dropped in each driver to mark the position of the mounting holes. I used a pen that just fit through the mounting holes in the perimeter of the drivers to mark where I needed to drill to secure the drivers to the baffle. I then removed each driver and drilled the mounting holes through to the backside of the baffle. Its important to make sure your holes are as close to perpendicular to the baffle as you can make them so you get even mounting pressure on your drivers.

For the mids, I rounded over the back side of the holes on the baffle as well. This provides a little more breathing room for the back wave of the midrange driver. See the picture below for a close up image of this. I didn't round the entire backside of the hole, because I still needed a flat mounting place for the washer and nut that would hold the drivers in place. Next, I lightly sanded the front, and painted them with a satin black paint that I grabbed from the local hardware store. The paint will raise the grain of the wood (which I why each baffle only received a light sanding prior to painting - don't kill yourself on this one). Let the paint dry, lightly sand it smooth, and then apply a second coat of paint. Be sure to work in the direction of the grain for a nicer finished appearance. Also, try to work quickly so that when the entire baffle has been coated, you can do a quick end-to-end brush of the entire baffle to smooth out brush strokes or other lines. This helps make a nice finished appearance.

To make the front baffle more rigid and to enhance the visual appeal, I glued and clamped 1x2 oak trim around the top and sides of the baffle. I stained this trim wood with a dark red colored stain, mitered the corners to 45 degrees, and then clamped the pieces in place. With the upright speakers, you can glue the three trim pieces all at the same time. Work quickly and carefully once you apply the glue. Wipe up any excess drips and adjust the alignment of the edges so the resulting front baffle is nice and smooth. It might help to use a small C-Clamp at the corners to keep the trim nicely aligned with the main baffle piece.

For appearances, I rounded the edge of the trim with a 1/2" round-over bit. Staining/painting both the baffle and the trim prior to gluing created nice, crisp edges for the two pieces of wood. Before using the router with the round-over bit, though, I applied a few strips of wide masking tape around the face of the baffle so the router wouldn't mark up the nice paint job that I had done as the wood chips flew. When you're done with the router, peel the tape, and use a small sanding block or sponge to give a final smoothing to the trim and then apply the final coat of stain.

Once the baffles were completed, it was time to glue the rest of the cabinet together. In the rear view of the cabinets (above), you can see the wood screws that I drove in from the back side of the baffle to further reinforce the lamination of the two pieces of wood. Since there are only five pieces to join together for this speaker design, its easy enough to do all of the gluing in a single step (see left most image above). I let them dry overnight and then tipped the speaker forward and drove a few screws through the bottom plate into the front baffle and uprights on the sides of the speakers. I recessed these screws so they wouldn't catch on the floor. The completed cabinet is shown above along with the corner trim detail. After each cabinet was completed, I applied a single coat of satin finish polyurethane to smooth things out and help make them easier for periodic dusting/cleaning. I didn't want to add another coat for fear of making the baffles too reflective to the point where they interfere with the projector in a darkened room. Then, the crossovers for the mid and woofers were mounted on the bottom plate of the cabinet and the XO for the tweeter was attached to the underside of the top plate of the back of the speaker. I used some hot glue and plastic plumbing strap with some screws for this. The result is a mounted crossover that won't buzz or rattle along with the music.

The base, side panels, and horizontal plate between the midrange and woofers are cut from a single 18mm thick panel of plywood. For as rigid as the cabinet is, the side panels vibrate more than I expected when the speakers are driven hard. Looking back on things, I would recommend doubling up on all of the panels for the speaker (not just the baffle) to help reduce vibrations when they are driven hard. The final step was to install the drivers. I used 2" 10-24 screws for the woofers and mids, and 8-32 screws for the tweeter. On the backside of the baffle, I used a washer and a locknut with a nylon insert so the screws wouldn't work themselves loose over time. The finishing touch was to add a dab of black paint to the screw heads to make them a bit less conspicuous on the finished speaker (this was cheaper than purchasing screws with black oxide finish and allowed a greater range of screw choices).

Center Channel Construction:

The images below show some of the center channel construction details. Before I began this part of the project, I spent some time asking questions of Jon and ThomasW over at the HTGuide forum concerning the form factor for the center channel. Since the original design was for tower speakers, no one had actually built a center channel for this design yet. The consensus was to separate the two woofers, one on each side of the speaker, and then slide the MTM section down to the floor, thus making an inverted "T" shaped speaker. In order to maintain the relative isloation from the front and rear waves of the woofers in the tower version, each woofer would each need to be boxed in on the back of the center channel. Jon further recommended making the main baffle a little larger than the original tower baffle.

The baffle for the center channel measures 55" wide by 26" high and has two symmetric cutouts at the top that measure 9"x20". Below are some additional in-progress pictures of the center speaker build. The center speaker gets the same two-layer lamination and finishing treatment as the left and right pair: 1x2 oak trim, 1/2" round over around the front edges, paint for the baffle, stain for the trim, and a coat of polyurethane. Cutting the top perimeter and attaching the trim pieces was a little more of a complex job than it was with the left and right speakers, but still pretty manageable if done in stages. The trick is to cut and test-fit all of the piece that will be glued at the same time as one another.

When the center baffle was completed, I cut the bottom edge at a 12 degree angle so it aims the drivers up toward the listening position while it sits on the floor. The first step in assembling the panels for the center channel was to attach the baffle to the base plate. I carefully marked the bottom plate prior to clamping to make sure that was aligned properly. After tightening the clamps, I drove a bunch of screws up through the bottom plate into the baffle to provide some extra strength to the glue joints and insure that they wouldn't shift while the glue set. After sitting for a day or two, it was time to create the boxes for the woofers. This allowed precise measurements for cutting the needed angles and test fitting before the boxes were glued and clamped in place. Each upright portion of the woofer boxes was also screwed to the bottom plate for added strength.

The remaining images show the gluing of the rear braces/supports between the baffle and the base of the center channel. Applying the lessons I learned from the tower speakers, I doubled the thickness of the wood for the boxes on the back of the woofer. I glued an additional 3/4" thick panel of MDF to the insides and bottom of the top panels - you can see them if you look closely at the third image below. Since I was in no mans' land with this design, I didn't make any changes to the crossover to compensate for the different form factor from the left and right pair. I figured this could be tweaked after the fact if necessary. The crossover components were mounted to the top of the bottom plate behind the drivers as I did with the tower versions. I was rather pleasantly surprised by the sound of the center channel. The result is that the center channel develops a few more dB of bass than the left and right speakers and a pinch more treble as well. I suppose the additional bass comes from the woofers being individually boxed in and closer to the floor, but I'm at a bit of a loss to explain the additional treble. The overall sensitivity of the speaker is a bit lower than the tower versions requiring a few dB boost to the preamp output level - I'm not quite sure why... When I finally got the theater room finished enough to run the Audyssey MultiEQ function in my preamp, it seems that the frequency response of the center channel is fairly flat by itself, since very little EQ is applied to the center channel. The left and right speakers, by contrast, have a few dB of boost at 63Hz and 125Hz as well as at 10kHz and 16kHz. I haven't done much reading on exactly how Audyssey is supposed to work, but I wouldn't be surprised to learn that it seeks to match the left and right channels to the center rather than the other way around...

Some Cabinet Changes:

Since completing these speakers, I've made a few additions to the tower cabinet and some progress on the overall theater room.

In my initial construction of these speakers, the front baffle was made from two layers of wood (3/4" MDF + 3/4" Baltic Birch) while all other panels (side panels, top and bottom plates) are only a single thickness of 18mm Baltic birch plywood. Building these speakers again, I would have used a double thickness of plywood or a combination of plywood and MDF on ALL panels for better strength and reduced cabinet resonance. To make the cabinet more stiff after construction was largely completed, I added a U-shaped cross brace connecting the side panels with the back of the front baffle directly between the bottom two woofers. I also added two 1x4" cross braces that connect the side panels to help keep things a little more rigid. These are mounted behind each of the 12" drivers, though they are not as close to the drivers as they appear in the image. The braces are held in place with screws and glue. The final addition was a set of carpet spikes to the bottom of the tower speakers since they are somewhat front-heavy and as a result lean down into the rug more in the front than they do in the back. The spikes also help keep the cabinets from jumping around on the carpet when things really get going. The three cross braces that I added made a very significant improvement in the stiffness of the cabinet.

The room overall is 14' wide, 24' long and just about 7.5' high. After some paint (grey for the side and back walls, black up front), a carpet, and trim work, it's looking more like a finished theater. I later added some drapes/curtains around the screen to improve appearances and reduce some of the sound reflection in the corners. The screen area is white paint on a black wall. I thought about building a cloth screen, but decided to just go with paint after sifting through numerous forum threads about the tradeoffs. I started with a screen size of about 100" diagonal (pictured below). After living with this for a while, I increased it to 120" diagonal - which is a great size for a 15 foot viewing distance. Just for a quick comparison, a 120" screen is THREE times larger than a 65" television and nearly EIGHT times larger than the more common 42" television.

And finally, below is an early picture of the theater room. As you can see, finishing details like a ceiling, carpet, paint, and wall treatments make a big difference (the completed theater is pictured on my main theater web page). In the image below, you can see the left, center, and right speakers (the center speaker is still under construction in these pictures). The holes in the back wall behind the center speaker are the manifolds for the infinite baffle subwoofer. Each manifold holds two 15"subwoofer drivers and all four drivers are powered by a Behringer EP4000 amp (1000 watts of output power for the sub). The two silver boxes on the floor are my DIY Aleph-X amps, the black boxes on the floor (left of the center speaker) are my CD player and an old preamp, the black box to the right of the center speaker is the Behringer amp for the sub.

Conclusion:

I'm driving my Avros with 125w Class A monoblocks (DIY clone of the incredible Pass Labs XA series of amps) and they are just incredible to listen to! These speakers cast an absolutely HUGE sound stage and produce an impressive image in a room that doesn't even have any furniture. The Avros make my Atlantic Technology and HATT speakers sound rather thin, and, well... rather boxed in. I have spent a great deal of time just sitting and enjoying these speakers! I haven't measured in-room response just yet, but close-miked response indicates strong output right down to about 40-45Hz - an amazing accomplishment from open baffle speakers.

These speakers play very clean at low volume levels all of the way up through excruciating levels. With a stereo pair in a 14x24 foot room, I was able to average 105db (with peaks to 107-108dB) at the listening position over 15 feet away without a sub (meter set to "C" weighting, "slow" response) without audible distortion. I don't, however, recommend maintaining this level of playback for very long... Its always a bit of a leap of faith when building something that you can't really listen to first to decide if you like it. I continue to be completely thrilled with these speakers!