My church in Austin, Hope Chapel, has a small sanctuary that seats around 300 people. When I arrived, there were a few studio monitors flying above the pulpit which covered most of the congregation but needed "fill-in" speakers on the side to cover some dead areas. Even with those, the coverage was not truly uniform.
To make matters worse, the studio monitors were not intended for PA use and had useful low frequency output only to around 150 Hz, which is nowhere near low enough even for a church service. To deal with this, a subwoofer was placed far away from them on the floor, but the combined sound from the two sources was nowhere near integrated, even sitting far away from them.
On top of it all, some of the tweeters on the monitors had blown, so the church desperately needed a better sound system! I love to design and build speakers, so rather than servicing the existing monitors, I volunteered to create a custom-designed speaker system for the sanctuary.
My budget for the entire project was on the order of $2000 and I wanted to get the most "bang for our buck" possible.
There were several goals guiding the design of the new system:
- 40 – 20,000 Hz (-3db) Frequency Response — I wanted the central cluster of speakers to produce the entire range rather than trying to integrate with a distant subwoofer.
- 100 dB SPL Sensitivity — This turned out to be more like 95 dB, but is efficient enough, given our budget and SPL needs.
- 300W RMS Power Handling — To match our exceed this so that sufficient levels can be achieved in concert situations.
- Uniform Coverage — I wanted uniform coverage of the entire seating area, with consistent frequency response: ±30° dispersion over entire range (-6dB) horizontal axis, ±25° dispersion vertical.
- 3-way, Tri-amped Design — For good midrange and good dispersion.
- Reasonable Size — Each cabinet no more than ~335 liters, so that the system is not too huge or too heavy.
- Cost — I couldn't use premium components because the budget for this was limited to around two thousand dollars.
Pro sound drivers vary widely in cost. At the time I was making my purchases, you can go to Parts Express and buy a compressor/horn tweeter combo that cost $15 or you could spend as much as $800 per unit. The trick was to find drivers that would meet our needs at a price we could afford.
I wanted a 3-way, tri-amped system for two reasons: better midrange and better dispersion. I hear lots of "tizz-bang" in public address systems that I abhor. They are generally created to produce loud sound efficiently: deep lows and sizzling highs without concern for subtlety. This is especially grievous in situations where a symphony orchestra needs sound reinforcement for outdoor concerts; the all-important mid-region where the instruments "live" is relegated to either a 15" low frequency driver which can't reproduce them well or to a horn-loaded driver which has huge 10-15 dB swings in its frequency response. I resolved that my system would be three-way so that the midrange could get the attention it deserves.
The 3-way design also helps with the goal of uniform coverage. When large low frequency drivers are asked to cover the midrange, the sound they produce in that region is focused within a tight beam directly in from of them. Smaller midranges disperse this same sonic information over a wider pattern.
The system is tri-amped, which means that there is no passive electrical network within the cabinets for dividing the sound into highs, middles and lows. Instead, an active crossover divides the sound and sends it to 3 separate power amplifiers which send the signal up to the mains via separate cables. I've chosen active 4-th order crossover points of 150 and 2500 Hz.
I ended up using the following total number of drivers:
- Bass: Eminence Kilomax PRO-15A (2)
- Midrange: Eminence LA6-MB (8)
- Highs: Selenium D210Ti compression driver (8), coupled with
Selenium HM11-25 horn (4) or
Selenium H395 horn (4)
The midranges are the weak point, power-wise, but still satisfy my design goals. The system goes way louder than it ever needs to — our church doesn't want a high-volume PA, but it does need finesse.
The polyhedral design (detailed in the above video) calls for the production of 3 separate cabinets with a total of 28 complex faces, most of which are unique. The mid- and high-frequency drivers are placed in the two mirror-imaged side cabinets and the large low frequency drivers are housed in the larger center unit.
The fully-mitered faces are joined with biscuit joints, and internal braces assure the structural integrity of the units. I could have built one large cabinet with internal walls dividing the inner volume into three parts, but I wanted the side units to be isolated acoustically and mechanically so they are hung separately (albeit close together) and spaced with rubber discs.
Side Cabinets (mid- and hi-frequency drivers)
Scribing and Cutting the Facets
Each side cabinet has 10 faces to it, each unique. And since the side cabinets are mirror images of each other, that meant the 20 unique, complex pieces needed to be accurately marked and cut. I decided to build the cabinets out of 3/4" plywood. I would have preferred to use MDF, but that would have made the units obscenely heavy.
By way of example, here is the Ligno3D output for one of the facets (Face 7). Most of the faces had 4 sides, but some had 5 or even 6.
The numbers near the edges are the lengths, and the angles indicate the bevel that is required on your table- or circular saw. So, the distance between verteces 9 and 15 (along the bottom of the illustration) is 20.782 inches, and the saw must be angled at 74.26° (90° would be vertical) in order for the piece to connect properly with Face 8.
At first I was using a labor-intensive and error-prone geometrical approach to marking the wood, measuring distances and triangulating. After a while, one of my friends volunteered the use of his architectual printer and I was able to make life-size printouts of these shapes. Then I was able to simply lay the printout over the wood, punch holes at the verteces and then draw lines to connect the holes... so much easier! I also made a jig for placing the fence so that I could make cuts more quickly.
Gluing up the Side Cabinets
This was the most frightening step of the entire project. The use of biscuit joints, the way that the facets were all mitered and the irregularity of the cabinet meant that there was no way to assemble two halves of the piece and then put them together after the glue had set on them. All the joints needed to be able to "give" a little in order for the final puzzle piece to fit in place. That meant that I had to use the slowest-setting glue I could find. I ended up using a hide glue that you mix with water - it has a set time of upwards of 45 minutes and I needed every single one of those to get one of these cabinets together!
Cutting Holes for the Drivers
After the cabinets were glued up, I had to fill in all the cracks along the edges with wood epoxy. This was necessary for many reasons: visual, structural and acoustical. I also used silicone sealant along the inner edges. I have a nifty store-bought jig for cutting precise circles with my plunge router. The rectangular holes were cut with a jigsaw.
Finishing the Side Cabinets
These speakers weigh a lot once the drivers are installed and the cabinets will be flown directly above our stage. I couldn't count on the biscuit joints to reliably hold all of that weight, so I custom-cut a bunch of braces which would help distribute the stress from the top of the cabinet to the various facets.
Mounting hardware and electrical connections
I decided to use three large U-bolts to fly each of the cabinets. Since I wanted the weight distributed evenly amongst these three points, I turned the cabinets upside-down, loaded them with their drivers and placed some small washers underneath the cabinet to see where the center of gravity was. I then marked the locations for the U-bolts at three points equidistant from that CoG. Also, a small rectangular hole needed to be removed from the back of the cabinet for mounting the Speak-on connectors.
Stuffing and Padding
I lined the interior walls with acoustical foam and after some trial and error decided to fill the cabinets with Acousta-stuff as well.
I took a couple of plywood samples into the church after I had used various stains on them and picked the one that most closely matched the beams that we have near our ceiling. After staining them, I applied a few coats of satin polyurethane varnish and sanded them smooth with very fine sandpaper. Not quite sure why I did that, since no one will ever touch them until the day they're removed from the sanctuary, but I prefer to know that they were done right.
Center Cabinet (low-frequency drivers)
Although larger than the side cabinets, the center cabinet is simpler and was easier to construct and glue together. It did not need to be glued together "at one go" so I was able to do the bottom half, cut holes for the drivers and add some bracing and then finally add the top and some additional top-to-bottom bracing.
The center cabinet uses what's known as a "bass-reflex" design, meaning that it requires a port for handling the lower frequencies. I had planned to use a distinctively-shaped port opening ( ) that would be visible above the front-firing woofer and I crafted a special shelf to fit the oddly-beveled space.
When I began, I used an oscilloscope to find the port frequency and it was showing up as lower than needed — around 33 Hz. This was according to plan, and the idea was that I would shorten it to tune it to the target frequency of 43 Hz. What happened instead is that when I shortened the port, the frequency only changed by 1 Hz and the port resonance became less well-defined. I realized at this point that no amount of adjustment would make it work properly, so I went through the additional trouble of cutting three specially-shaped pieces to fill the empty space I had created, and then just cut a large 6"x16" rectangular port in the rear of the cabinet. This new port did in fact work as expected.
I had originally toyed with several ideas for covering the drivers. One option is not to cover them at all, but the size and shape of the cabinets is attention-calling enough without revealing the driver cones. The option I had in mind during most of the construction was to make frames that would extend speaker cloth to within an inch of the edges of those facets with drivers, but then it seemed a shame to cover up so much of the wood which had turned out to be so nice to look at.
So, after getting the opinions of some friends, I decided to leave the tweeter horns as-is (they're flat black anyway) and just cover the midranges and woofers. The midranges all shipped with a separate paperboard gasket on top of them which you can use if rear-mounting the drivers to your baffle. So I cut slightly larger circles of fabric, stretched them over the drivers and stapled them into place. Once finished, I attached the grills to the front of the midranges with some silicone sealant.
The woofers didn't have an easy option like this, so I had to create circular frames for them which could fit snugly over the outside of the driver. I then used sealant to affix these as well.
Installation and Adjustment
Now it was time to hang these things 18 feet up in our sanctuary! Although lots of people helped make it happen, two deserve special mention — I couldn't have done it without the help of Steve Cain and his son, Brian, who were also installing a projector in the church building. They were the ones who rented scaffolding and put in hours of grueling work to see this project through. The process of putting these cabinets into place took several days, beginning with the removal of the previous speakers.
It took longer than I expected just to remove the old cabinets. We didn't have the scaffolding set up yet, so I perched on top of an enormous A-frame ladder that we use for such occasions and removed the brackets that kept the speakers in place. The next step was for me to actually prepare the new cabinets for being hung. I still had to install the speaker grills I had made and the stage left speaker cabinet had yet to have its drivers installed. After many hours of work, and lots of help, the new cabinets were hung roughly in the places they needed to be, but still needed to be hoisted a little further and leveled out.
The next evening I installed some new cable runs so that the new tri-amped system would get plenty of power. After that, we hoisted the cabinets a little further and got them nice and level and I put some cable wrap over the chains which were suspending the system.
Finally the time had arrived to power up the system and see what it would sound like. I had not yet heard all the cabinets together at one time, and have to admit that even without EQ, the system sounded better than our old one. I just barely had enough time before Sunday, Sept. 23, 2007 to run some pink noise test with a calibrated microphone and even out the equalization of the system. At this point, CDs played on the system sounded great and I was very happy that everything worked as well as expected. The final hurdle was to see if feedback was an issue with a live band on Sunday morning and everything went very smoothly indeed. No feedback at all, which was a concern for me since this new system has such better dispersion than the old.
I'm very grateful that it turned out well, and it's hard to believe that after several years of thinking about this project, it's finally done!
Special Thanks to: Steve & Brian Cain for going beyond the call of duty to help see the installation through.
I mentioned earlier that feedback is not an issue, but that's not strictly true. We don't run sound so loud that we get anything howling or overloading, but we do have an issue with certain frequencies "ringing", such that we can't amplify the the wireless condenser mics used in the podium area as much as we'd like. In hindsight, it's clear that the system (built back in '06) disperses the sound too well and is located too close to the persons speaking on the podium.
A future project, when I have time for it, is to reconfigure the system in two ways:
- Target the listening area more narrowly. I mean to do this by deploying a sort of Synergy Horn design. Tom Danley's idea is patented, but he's generously shared details of the design with the DIY community and we're fine using it for ourselves in projects that are not for sale.
- Run the sound in stereo instead of mono and place the speakers so they're not directly over the podium. They could be placed off to the sides and a little closer to the congregation.
I can reuse some of my high- and low-frequency drivers for these modifications and will have to order new midranges. I think the system will sound a LOT better when I'm done.