The Repository->DIY->Tubifex pt. 1 (Design, Crossover, Measurements)

Who says a speaker has to come in a rectangular box?  Who says a speaker has to be a box at all? 

What's fun about this project is the "MAKE" or "hack" quality of it.  Using fairly ordinary objects and tools, one will end up with a very unusual, very high-quality pair of speakers.  The majority of the Tubifex are made from SCH 40 PVC pipe, and the endcaps and support pieces are cut from MDF.  This project could be done with very ordinary tools: a jig saw and cordless drill would suffice, though a router might be nice to flush-mount drivers.   Crossover construction will take some knowledge of how to use a soldering iron.

The Tubifex should cost somewhere around $215 for the drivers and crossover components.  Bill of Materials here.

Before we continue, I feel I need to stress the point that this project is far more than your average "let's slap a driver in a weird plastic thing and see what happens."  I have seen several different projects of this genre around the internet, and while I would call the end results "art," I have very low expectations of them in terms of sound quality.  A high-quality loudspeaker has drivers that were specifically chosen to meet project goals, a well-braced enclosure which is tuned to get optimal bass, and a crossover which achieves a flat Frequency Response, tapered Power Response, good Acoustic Phase tracking through the crossover region, and is a safe Impedance load for a normal amplifier.  Meeting all those goals is usually beyond the scope of speakers you would find at a Big Box store--as they usually compromise a few features to meet a price point.  A loudspeaker which meets all those requirements is more likely to be found at a HiFi dealer, at a price point which starts at the 4-figure mark.  (If you've never heard such a speaker, you owe it to yourself to go hear the difference.)  The Tubifex do meet all those goals, and will give you audiophile-quality sound at a fraction of retail price; the fact that they are unique in their design is merely a fun bonus. 

OK, I'll get off my soapbox now.  Let's get down to the details. 

Enclosure Design

The Tubifex are based off a project by Parts Express' Mike V. called "Totally Tubular" (left)  Mike's intention when he created his design was to find out what would happen if a speaker had "no baffle."  This is an interesting idea, since many speaker designers are always looking for ways to ameliorate unwanted baffle diffraction from a driver, whether by using large-radiused edges, faceted corners, or careful placement of felt.  It's surprising that he was one of the very few people in the world to just "get rid of the baffle altogether" and use tubes for the enclosure.  Did Mike's design accomplish anything in terms of eliminating unwanted baffle diffraction?  I don't know.  Did it look cool?  Hell yes! 

So why haven't more speakers been created that look like this?  I have no idea.  Maybe people think it'd be too hard to construct (it isn't).  Or maybe the looks are just too unusual, and most people just prefer something boxy in their living room.  Who knows?  All I knew is that ever since I saw and heard the Tubulars in-person, I knew I had to do something like it.

So, as you can tell, my Tubifex were based off the Tubulars, at least in shape.  The top and bottom tubes of the Tubifex still use 6" diameter PVC pipe, however the mid tubes of the Tubifex use 4" diameter PVC pipe. 

The inner volume of each woofer tube is approximately 9 Liters.  Each tube is tuned via a piece of 1.5" diameter PVC attached to the back endcap.  The vent length is approximately 4.75 inches (including the endcap), which tunes the enclosure to 49 Hz.  The inside of each tube is lined with eggcrate foam (eg: a foam mattress from Wal Mart), and there is a small puff of poly fil behind each woofer.

You could technically call these "bookshelf" speakers, however at 22" deep, I don't foresee them sitting on any actual bookshelves.  Sitting them atop a speaker stand, end table, or sub (or even flown from the ceiling--I've seen it!) should complement a modern or post-modern decor. 

Driver Selection

I went through a very exhaustive process of finding just the right woofers for this project.  I wanted something that:
  • Had a frame diameter smaller than 6.5"
  • Could play up to at least 2000 Hz
  • Would achieve the lowest possible F3 in a 9 Liter enclosure (the internal volume of each tube)
  • Looked cool!
After modeling pretty much every midwoofer available, I finally settled on the Dayton RS125-4.  It's an exceptionally good woofer for the price.  As an added bonus, since the drivers are run in series, they present a safe impedance load for any amp.  The bass extension is really spectacular for a "bookshelf-size" speaker, with an F3 of 40 Hz.  Additionally, the RS125-4 can play pretty high before cone breakup becomes an issue, which opens up a lot of choices for tweeters.

I tried a few tweeters, but the one I used in the final design is the Tang Band 25-1719, an interesting little guy.  The dome material claims to be "ceramic"--however, true ceramic dome drivers are ungodly expensive.  More likely, it's some sort of aluminum which has been coated or anodized to give it the especially rigid qualities of a ceramic dome.  I did some distortion tests of it and found that, above 2500 Hz, it's a very low -distortion tweeter.  I did try crossing it below 2500 Hz, but I found the odd-order distortion to be very irritating and "tizzy," and the tweeter basically sounded like a piece  of cheap junk.  Once I raised the crossover point, everything sounded clean and the tweeter had no "sound" and drew no intention to itself (which is what you want in a loudspeaker). 

Crossover Design

The crossover uses a 2nd order electrical Low Pass on the woofer, with the .1 uF capacitor on the inductor as a notch filter. The 2 Ohm resistor is there to slightly adjust the woofer roll-off, and finesse the midrange. I decided to go with the iron core inductor to gain a dB of sensitivity.

The tweeter filter is 3rd order electrical. The 1.5 Ohm shunt resistor is there to tame the ripple just above the XO point. The only padding on the tweeter is the 8 Ohm resistor in series just before the tweeter; this did an excellent job of taming its rising response on the top end.

Laying out the crossover is a bit of a trick.  It has to fit inside the 4" middle tube.  Here's how I did it. 
Note: in my photo, I used a pair of .2 uF caps [yellow] in series to achieve the .1 uF across the large inductor.  Hope that's not too confusing

Here are some graphs...

The acoustic slopes are pretty steep. It's not how I would have originally chosen for things to turn out; but the inductance on the woofer gets REALLY intense by about 2500 Hz, so there's no real way to get the woofer's acoustic slope any shallower without negating the notch filter--and you still have a massive impedance spike to deal with.

The impedance load of the Tubifex should be fine for pretty much any amp (to my limited knowledge, anyway).

Listening Impressions

Now is the time I get controversial: there is some very vivid, pinpoint imaging going on with these. I didn't want to believe it at first; thinking it might be a psychosomatic effect of seeing a "baffle-less speaker."  In reality, the lack of a baffle might have very little to do with the imaging--in comparison to the quality of the drivers, and my voicing of the crossover in regions are ears are most sensitive to localization cues.  Whatever is doing it, the imaging is so precise that it might actually be unnerving to some listeners. If a singer is panned "center," his or her voice will "float" about 6 feet in front of me--with a width and height of only a few inches! Yeah, it's weird; some would call it "exaggerated" or "novelty." Whatever. Can't please everyone.

The other thing that's very fun about these is the bass extension, which was one of my biggest goals with driver design. The F3 is really 40 Hz, just like Unibox predicted. I can't help but smile that such a small, odd speaker can replay the "sonic booms" on metal recordings.  It should turn a few heads.

Now for the downside: these are terribly inefficient. I'm guessing 82 dB tops. I am running them through a 50 watt amp, and can push the amp full throttle and not get the speakers to their limits. On the plus side, I don't know where the limit of the speaker is, but it's loud! I listened at 90 dB for a while one day, then thought, "What the heck, lets go all the way," and that's when I turned the volume knob up all the way and left it there. The dB meter sat around 93 dB, with occasional hits around 97 dB. Weird thing was, my ears weren't hurting. I'm guessing that's the "low distortion driver effect." Still I wanted to know if I could hit xmax, so I just sent full-blast sine waves through at 40, 45, and 50 Hz. The dB meter read 103 dB, but I still couldn't breach xmax. So, make of that information what you will.

And finally, here are some more wiggly lines...

There's a null created when the tweeter's polarity is reversed. Must mean the XO is "right"

And finally, here's the measured response at my preferred listening position (speakers pointed straight out, me sitting at the top of an isosceles triangle)

On to Part 2: Materials and Construction Tips... ->
Jun 1, 2010, 12:37 AM