What's new

Oliver Acoustic Research AR-94 specifications
Daniel Azziorri RCA new topic
Roy A. Esposito Acoustat topic reply
Joe Kasperowski Acoustat topic reply
Ed Optimus topic reply

Introduction

Somewhere in 2001, I bought some surplus drivers from stryke audio. Among these, were a couple of vifa P17WG woofers. I had planned to build basic bass-reflex speakers, which would be powered by the alesis modules I bought from Apexjr. But my plans changed. I have a fine amplifier for my speaker, so there was no direct need for active speakers. And in the meantime, I had heard some transmission line speakers and fell a little in love with the sound. And when Martin King released his TL papers and computer-model, I decided that I wanted to build TL's.

I measured the parameters of my woofers and compared them with the factory specs. Just as with my Scan Speak woofers, there was a big difference. However, I did several measurements, at different times, and got consistent results, so I decided to trust my own measurements over the factory specs. I also added a compensation magnet (before determining the parameters) to the woofer to cancel out the stray magnetic field, because these speakers will be placed close to my pc.

Thiele/Small parameters of my Vifa P17WG woofer

Thiele/Small parameters of my Vifa P17WG woofer

Fs=46Hz

Re=5.4ohm

Vas=20 l

Qes=0.74

Qms=1.6

Qts=0.51

Sd=137cm²

BL=5.7Tm

Le=0.58

The choice for the tweeter followed after some emailing with Steve Sedmak. I had limited my choice to Peerless and Vifa tweeters, because they're affordable and their proven quality. At that time, Steve just had purchased the same tweeters. He measured them and had a preference for the Peerless tweeters. They have lower distortion, and a better frequency response. And on top of that, they somewhat cheaper too! So I settled on the Peerless 812978 (WA10/TV)

After studying all the papers that Martin King had written, and checking out the worksheets, I fired up Mathcad and entered the parameters into the worksheet. And after a few hours of simulating and messing around with the sheets, I hadn't gotten anywhere! This was proving to be a lot more difficult to get working than the usual box-design software. I searched the web, and found a website on which several designs were featured, which were based on Martin's worksheets. I asked the writer of the website -Bob Brines- for help, and thanks to Bob, I'm now the proud owner of a pair of properly designed TL's! Bob had some rules of thumb, that usually give a very good staring point for optimizing a TL design. In my case -I wanted to have a relatively small enclosure, with an F3 around 35-40 hz- a conventional TL would work best, with a starting width of about 4 times the surface of the woofer(Sd), and a terminus of about 0.5Sd. Length would be around 60-70 inch, with about 0,5lb/ft³ of stuffing. When I entered these figures into the worksheets, I already had a better design than in the hours I was trying to get results myself!. Now that I had a good starting point, it was a matter of tweaking. I increased and decreased line lenght, tried different stuffing methods, different sizes for TL top area and terminus and different positions for the woofer. In the meantime, Bob has written a page about this on his website, which is very helpfull information for anyone trying to design his first TL. Just visit his website and go to the "design a TL" page.

I first planned for a design that uses two woofers per box, because of the rather low efficiency of the woofers, but that makes for one very large box, so I quickly gave up that idea. After several emails and some evenings of simulating, I decided on the following geometry: Line length of 62", tapered from 3Sd at the start to 0.5Sd at the end. Some heavy stuffing (1lb/ft³) in the start of the line, some lighter (0,5lb/ft³) in the middle and no stuffing at the end of the line. TL's are very sensitive to the position of the woofer. By careful positioning of the woofer, it is possble to shape the response above roughly 100Hz. In my TL, I positioned the woofer 12 inch from the start of the line to cancel out the first glitch in the response.

This geometry gives the following response (Blue is infinite baffle response, red is TL response):

Advertisement

The design

After I decided on the geometry of the speakers, the actual design of the box itself was pretty straightforward. A lot of design aspects are already determined by the TL geometry, thus that makes the design a lot easier (or more difficult, depending on your point of view...). About the only thing I needed to do was to determine the place of the tweeter. For this, I used the baffle diffraction simulator by Paul Verdone. With this excel sheet it is possible to study the effects of driver positions on the output of the speaker. The woofer position is fixed in my case, because of the TL geometry, so the only thing i needed to do was to determine the position of the tweeter. Placing the tweeter as much to one side of the speaker as possible turned out to give the best results. The further I placed the tweeter away from the woofer, the better the results the spreadsheet gave were. However, placing the woofer and tweeter far apart spells trouble in crossover design. This was described (among others) in the AES paper by Linkwitz on active crossovers for non-coincident drivers (1976). It is almost always the best choice to put the tweeter as close to the woofer as possible, however I left some space between woofer and tweeter, because placing the woofer and tweeter close together resulted in a large bump in the response around 1200-1500Hz. And indeed, this came around and bit me in the ass when I designed the crossover, so next time: woofer and tweeter close together and only check the effects in the excel sheet of offsetting the tweeter.

The design(all dimensions in mm):

I also made a drawing of each individual panel in the speaker, with comments in dutch. Click here to download the PDF.

This doesn't give a good impression of how the speaker will look, so I spend an evening with a 3D package and that resulted in this picture:


Not only does this give a visual impression, but it also makes you "build" the speaker. And some design changes are a direct result from making this 3D drawing, because it was clear that some small changes would make sure that the speaker was a lot easier to build.

The construction

Now that the design of the box itself is done, it's time to start up the table saw!

I built a prototype first, and when all the test-measurements were done, I built the final pair of speakers. This page shows both construction pics of the prototype and the final pair.

Here you can see my nifty router jig in action. On my monitor speakers page you can read about how I was planning to buy one of these babies. Well: I did, and I'm very happy with it. It makes routing squares and long slots a breeze. I bought the Klang&Ton version, which uses a somewhat narrower aluminum profile and costs less (basic set about 250euroos, I bought some accessoiries). Here, I used the jig to route out the slot in which the dividing panel will be mounted. In the lower left corner of the image you can see this panel mounted in the other side panel (which already has this slot routed into it)

This is the box in clamps. Unfortunately, I did not take any pictures of my dowel jig in action. It has saved me a lot of time assembling the speakers. It takes some time to accurately drill the holes, but assembling the speakers with dowels is a piece of cake. The dowels accurately align all the panels, and also make sure they keep in place, leaving your hands free to place the glue-clamps. I won't build speakers without them anymore! You're looking at the front of the speaker, the baffle will be mounted later.

These are both speakers, primed and ready to be painted. I applied the primer with a brush: big mistake! The brush leaves such deep brushmarks that I needed to sand almost everything off again. Next time: roller or spraying.

Pretty ey?

I bought my paint at the local paint store, the darkest red I could find ("Histor zwartrood", if you're interested). I first practiced rolling paint on a small piece of scrap-MDF, and discovered that my painting skills hadn't improved after building my monitor speakers (yes, I suck at painting). I took this can of paint to another paint store and had them put the paint into spray-cans. These are "professional grade" spray cans, which means they're under higher pressure and have a fan spray pattern (instead of the usual round spray pattern, which sprays more paint in the center of the spray-circle than at the edges, which leaves noticable streaks). This costs about 8 euroos/can, and I had 3 cans filled. This is expensive, but I didn't want to ruin the mucho-expensivo brand-new spray gun we have lying around. I masked off a small area, which I declared my "spray booth" and also strictly offlimits for everyone else but me and my dad. Spraying the speakers was a breeze with these spraycans, no more brushes or rollers for me! I'm way better at spraying than brushing or rolling. It even looks good by objective standards (and not just my biassed opinion!)

Crossover

The first thing I did with the freshly built prototype was to measure the impedance of the speaker. This impedance corresponded very well with the impedance predicted by Mathcad. This is good to know, because this means that Mathcad has modelled this speaker correctly, and thus my measured T/S parameters were correct, and I could safely assume that other characteristics were properly modelled.

This image compares simulated impedance (top) with measured impedance (bottom). As you can see, the peaks are in the same place and also have the same amplitude (exept the larger peak, which is about 2 ohms lower, which is not really significant)

The next thing I did with the freshly built prototype was determining the effects of different amounts of stuffing on the bass-response. This is the measurement setup:

Although this was done in a small room, this doesn't have much effect on the measurements, which were all nearfield.

This image gives an impression of how I stuffed the TL. I lined the walls of the front "chamber" with 28mm printex and filled the open space with baf wadding. I left the other half of the speaker completely unstuffed. This image shows the speaker with about 75 grams of baf wadding installed.

The next 2 images show the effects of stuffing on the output of the woofer and the terminus. As you can see, small changes have a large effect. One piece of baf wadding weighs approximately 150grams, and is roughly 50cm by 75cm. Half a panel is pretty small, but has a measurable effect on the bass-output.

I decided that using between 150 and 225 grams of baf wadding would give the best results, and that I would decide on the final stuffing by ear (and keeping within 150 and 225 grams)

With these results I was convinced that the basic design of the speaker was good, and started with the construction of the final speakers. While building those, I continued using the prototype for measurements.

Next, I mounted the tweeter in the baffle and did some frequency response measurements. After a false start (my first measurements turned out to be wrong, which also rendered my first crossover a fluke), I got some good measurements. I then started on the final crossover design. This is where the tweeter placement turned out to be a less fortunate choice. One of the tests that shows if you have proper phase tracking in the crossover region, is that the response shows a deep null if you reverse the polarity of one of the drivers (Good phase tracking is necessary for proper summation of the drivers over a large listening area). When I reversed the polarity of one of the drivers of my speakers in LspCad, it didn't give a deep null. Why didn't I chage my design at that point? 2 reasons: I sit at a very fixed point, namely behind my pc. Which lessens the need for a good response over a wide area. The other reason is that I was already nearly finished with the final pair of speakers, and I didn't want to go through the trouble of spray-painting a second set of speakers.

After some simulating in LspCad, I had some preliminary crossover designs. And to determine which one sounded best, I setup the speakers in the living-room and connected them to my pc through an Onkyo HT amplifier.

The crossover that sounded best was quickly found, and only needed a little tweaking (I changed two components one E12 step). However, the bass was a little weak. That was weird, because the bass was noticably weaker then before, when I was listening in mono to the protoype speaker. It tured out that they way I stuffed the speakers was the cause of this. In the open speaker in the image above, you can see one large piece of baf wadding that goes al the way from top to bottom. What's not very visible is that the bottom piece of baf wadding bends around the fold in the TL and lies flat on the bottom. In these speakers I folded that last piece of baf wadding 180 degrees, and added a small piece of baf wadding in the bottom of the speaker. After I unfolded this piece of baf wadding and removed the small piece from the bottom, the bass was back in full glory. Folding the baf wadding probably made the density at that point so high, that it caused the bass to suffer. There is still more damping in this speaker than in the prototype, but the bass is at least as good, probably better. The highest density is in the top of the speaker, behind the tweeter (more than 1lb/ft³). Behind the woofer it is a little less, and the bottom half of the front of the speaker has about 0,5lb/ft³. And all the walls in the front half of the speaker are lined with 28mm pritex. All in all, this makes for a very nice bass response, with that sweet TL flavour over it.

This is the filter I decided on:

It results in this frequency response (the output of the terminus is not added to this plot, thus the dip at the tuning frequency):


There is a slight lift in the lower frequencies (50-200 Hz) which is also noticable when listening to the speakers. It gives a somewhat "poppy" character to the music, which actually is kind of nice. If I get tired of this (which I probably will), this small lift is easily corrected by adding some more damping material. The impedance of the speaker is very friendly and never goes lower than 4ohm (at 5000Hz), the phase angle never becomes larger than 30 degrees.

Then: how do they sound? Well, first of all, they're a huge improvement over my previous pc-speakers (a pair of jamo compact 700, standing on the floor). I just love the bass from these speakers, very nice. At this writing, I have only listened a few hours to the speakers. And I like them, nothing really stands out, which I consider to be a good thing. And considering how they're set-up, not a small feat. In july, I'm going to take these speakers to a small gathering with more diy speakers. I'll add some of their comments here, then.

The speakers

Did I mention that they're difficult to photograph? It's because of their dark color (which isn't even that dark if you see them with your own eyes), the camera has difficulty to get enough contrast. Especially when it isn't dark outside.

Raymond van Weeghel