I haven't really got a great set of photos to go with this yet, but I thought I would post this up here since some of the old Data-Bass tests were something of an inspiration for doing this project. So you all get first crack. Here's the teaser photo:
It's been years since I had a decent subwoofer. My last project was a Resonant Engineering XXX15 XBL^2 in a huge cabinet tuned to about 16Hz. But that was more than 15 years ago now, and it's been mothballed for the better part of a decade. Still, at the time, it was a "who designs something tuned like that with a 6dB drop?" sort of project. Now everyone does it that way, and just EQs it flat. Yeah, you lose 1 or 2dB up top, but you pick up 10dB down low. I said "I'll take that math then" and I'll take it again now. But even more nutty this time.
Planning started about a year or two ago. This subwoofer needed to go in my living room theater, which is about a 16x21' room in what I think most people would probably regard as "Victorian mansion". The house is enormous, doorways are 8' high and 6' wide, the ceilings are 11 feet high, and nearly every surface they could find to cover in quartersawn oak is covered in the stuff. The floors, 3' up the walls, the doors, the stairs, ceiling beams... Putting in a huge ugly subwoofer was not an option. And room gain somewhat questionable. I already managed to sneak in a full set of Polk LSiMs, under the theory that we needed some cherry to contrast with all the oak... . I did the electrical design for the subs, bought the parts, bought the amplifier ... and then never build the subwoofers. I finally got tired of looking at my parts and slapped it together a few weeks ago.
No one's really done a project like this in years (if ever), so hopefully someone will find it interesting. For a PR, this is a big cabinet. The project uses a pair of Earthquake Sound SLAPS-M12 passive radiators, and a single 15" Alpine SWR-1522d.
The design goal is for a pair of subwoofers (to help flatten response) with solid extension and output of at least 110dB at 15Hz, ideally still with 100dB anechoic output down to 12 or 13Hz, and to pull this off using less than 4 square feet of floor space. 20Hz and up is easy stuff. It was pulling off the ULF in a small cabinet that was a challenge. Port velocities and resonances were not working for me in the allowable cabinet sizes. Port compression was sucking away 6 or even 8 dB, and trashing the design goals. So I fiddled with passive radiator designs. Now that's interesting... Passive radiators, here we come. The only problem? No one was selling passive radiators that would work. TC Sounds was long dead. Creative Sounds had a few APRs left.. but those things only had 30-something Xmax. Pickings were slim. Then I stumbled across the Earthquake SLAPS. A claimed 4" inches of "peak excursion" (in typical Earthquake car audio vagueness), parameters that appeared to allow a very low tune. But it's just a plastic disc with a fat surround? Eh, can't win if you don't try. Plus, if there was one thing I figured that Earthquake knew how to do, it was put a bit fat surround on something that would allow a cone to flop around wildly. But no spider? Would it rock? Would it actually have 2" of travel (one way)? Would it take over a 1000g of added mass? Who knew? I didn't. So I ordered four. For about $60 each on Amazon. There they sat on the shelf with the other pieces and parts for a good year.
Another long unfinished project was getting my Atmos up and running. Due to whole "victorian mansion" living room issue, whacking holes in lath and plaster was a bad option. So I bought some Polk OWM5s with the intent of hanging them off the ceiling, somehow. Those sat for a year too. Then I got sick of looking at them, bought some mounts, and went to hang them. And promptly realized I could manage to hang them using hooks on the existing picture rails, and conceal all of the wires in the picture rails (a type of molding in old houses intended for hanging pictures without trying to hammer nails into plaster walls). Perfect. My better half was thrilled, surprisingly, and invited everyone over for a "movie night". Aha! A perfect excuse to spend a few nights building subwoofers.
I don't particularly care for threads with dozens photos of people building subwoofer cabinets. It's just.. a box. But here's a couple of obligatory photos anyway:
I built the box in another house I'm working on. Not my living room. Although the site of sawdust in the living room probably makes some of you jealous.
Much of our first floor has a quartersawn oak wainscot that runs about 42" high. This cabinet is precisely as high as the wainscoting. I built a box out of Home Depot oak plywood (which was fairly decent and void free, actually), and slapped some Golden Oak on it. Oh, so ugly. Then I trimmed it out with various pieces of molding that I had lying around, matching the top (roughly) to the wainscot. Grain orientation on one panel is wrong because I had Home Depot rough cut this stuff so I could toss it in the back seat of my car. I didn't spend enough time on the cut list. This whole thing was done in about two days, finishing included. I pulled three side pieces off the sheet vertically, and one horizontally, then trued them up on the table saw when I got home. The actual box is about 1/4" larger on each side than the 15" driver, which is bottom mounted. I glued it up with wood glue and used finishing screws to hold the box together. No clamps. Braces were left over scraps glued as appropriate. Then I grabbed a few pieces of trim molding I had and glued and brad nailed them to the box to build this pedestal design. 2x4s are glued into place for the legs. Finishing was a Shellac seal coat, then golden oak stain, then a coat of shellac, let dry to sort of tacky, then a darker gel stained brushed on and dry brushed off, dried for 30 minutes, then a coat of shellac, then another coat. All applied in the course of a day, loaded into the car, and then moved. Not exactly according the finishing schedule on the can... But movie night beckoned. The finished product is on the top.
The performance is actually quite spectacular. There is zero noise from the system, and the bass below 20Hz is as (ultimately) modeled. Unfortunately, I haven't been able to take any really good measurements yet. The radiators perform extremely well, and do not rock or exhibit other bad behavior. There was one small problem though, and that was the box tuning. The original design was supposed to be tuned to about 13.5Hz. It wasn't. The passive radiators were loaded down with the factory supplied weight, and a stack of 3" washers from Grainger on top of that. I measured the movement of the woofer cone, and came up with about 15Hz. Uh-oh. Earthquake's parameters for the passive radiators were off. At one point, I knew about this, and then promptly forgot about it. They specified a Vas of about 4.5cf which, when loaded into WinISD, causes it to autocalc the Mms at 100g. The actual Vas, if you derive it from the Sd, Fs, and Mms is about 1.5. Jeff Bagby's Woofer Box program will consistency check the PR parameters. WinISD does not.
[I lost about a page of technical analysis in here when I bumped the back button my mouse. I don't feel like retyping all of it right now, so I'll give a quick summary and hit the high points:] The Data-Bass models for the LMS-R system are fundamental to understanding some of the unusual properties of passive radiators. You have to get the properties of the radiator right to model these things. Many of the passive radiators being sold have bad parameters. Earthquake was not alone in this. TC Sounds had bad T/S parameters. PSI has bad ones. They will not mathematically reconcile. Earthquake was helpful in getting to the bottom of it, even cutting a radiator apart for me to weigh the actual Mms. With that in hand, I could measure the Fs by resonating the radiator in free air, and guesstimate the correct Sd with a tape measure. The other complication is that parameters which are normally assumed to be static clearly ARE NOT. They suffer from the same large versus small signal issues as subwoofers. Once you take that unloaded radiator and slap 2 pounds of weight on it and get it moving a lot, the small signal parameters change. Using the measured small signal parameters, actual system tuning should have come in around 16Hz. Fortunately, it came in at 15Hz. A far cry from where the factory specs would have put it, but better than 16Hz! At low tuning frequencies, the weight that has to be added to a PR can be huge to get a slightly lower tune. Hundreds of grams hanging off the suspension and sucking efficiency. The 15Hz tune with 960g left enough washers in reserve to get the weight up to 1240 grams each, and tune down to 14Hz.
Data-Bass shows you the weirdness that happens with passive radiators. Go here, https://data-bass.com/data?page=system&id=49&mset=47, and then Static Charts, Impedance Plot. See it shift lower with the higher power? That's the tuning frequency of the whole system going deeper the harder it works. Granted, other systems also do this, but the passive radiator system is particularly notable for how orderly it seems to happen. And, I suspect, also notable for its ability to matter since the port isn't loading up. Now go to the Extended Charts and look at the long term power compression. These are the only system which ever showed that bizarre peak--and all of them did it, regardless of the weight on the radiators. Again, I suspect this is because the tuning goes deeper the more the radiator is moving. In my discussions with Earthquake, the indicated they had also observed this behavior. So here you get a system that shift the tune lower the harder you push it, and doesn't conceal it with port compression. Now that's useful.
Here are the models for the final product versus two comparison models for the Captivator 1400:
Sorry for the bad color selection. The Captivator was modeled using a FiSP4 which has a similar motor strength, in a 5.5cf box tuned to 17Hz. It was also modeled as a passive radiator with twin 12" passives. The last is my model. Not too ad, but are these reality? Modeling up the LMSR-12 PR system (not shown) indicates that it should start falling on its face at around 13.25Hz where the passives run out of gas. It doesn't. Well, at low power it does. The "knee" is around 13Hz in the low power power compression charts. But it shifts over to about 11.5Hz as the power level ramps up. It puts out about 94 dB at 11Hz. If you model it, it should only be managing maybe 88dB adjusting for 2m groundplane. That's interesting. The ported systems don't do that. The ported Cap model matches well around 30Hz, and then progressively falls away from the model as the frequency drops, until it's about 6dB off the model by 14Hz (adjusting for 2m groundplane). So if the goal is to squeeze out performance from 14Hz up through 20Hz, passive radiators are a fabulous way to do it. A smaller driver in a smaller box with the same power yielding the same output. It's a fine trick, really, but it has its limits. You can only load up so much weight onto the radiators, and the loss in efficiency elsewhere is very obvious.
Anyhow, perhaps I'll follow up on this later if anyone is interested in something that isn't a huge bass horn. But for now, this is long enough... Suffice to say that the Earthquake radiators do work, and if you play around modeling them (use Vap 1.6, Fs 21, Sd 483 to get somewhat close to accurate), and dig through how the the radiators perform in Data-Bass, you'll walk away with some interesting possibilities. 2" of Xmax for about $70. Two weeks ago, Amazon had them at $50. And since the Xmax is so high, the "two times" the cone area rule of thumb isn't that big of a deal if the goal is extension below 20hz. And the tuning shifts lower the harder you push them.
Yeah, you probably shouldn't tune a 4cf cabinet down to 14Hz, but you probably shouldn't tune one three times that size down to 10Hz either. That didn't stop JTR, so I figured I shouldn't let it stop me, either. Plus, I figured their port losses had to just dwarf whatever radiator losses I was going to get. Had the original Earthquake specs held (which they didn't), with a fantastic 25Hz Fs, and a 100g Mms, this whole thing would have been far more efficient and a pair would have schooled a Cap 2400ULF. As it stands, I probably can't load enough weight on them to get there. No big loss though. Even at 12Hz just the one completed subwoofer will make the floor start bouncing.
Radiator and Driver Notes:
The radiators seem strong and the rear surround does the job at keeping them pistonic. The surround is over an inch tall, which does seem to give a legitimate 2" of excusion. I have yet to observe any radiator rocking or other bad behavior out of them, and I've beat on them pretty hard. You do NOT need to worry about slapping a limit on the system to protect them. Just let them run wild and bounce off the suspension. When I weighted them, I bought a box of 100 3" washers with a 5/16" center hole from Grainger. No one else sells them in that size. The washers are 40g each, about $30 shipped for all of them. You could also use a standard 3" washer and increase the bolt size by drilling a new hole, but then you'll also have to drill out the supplied weight. I also turned the bolts around and glued them to one of my 3" washers so that I could (possibly) add/remove some of the weight outside of the enclosure and adjust it without taking the radiators off. We'll see if the glue holds. I tried to balance the weights about 50/50 inside and outside the cone to avoid creating a cantilever situation.
The Alpine drivers are NLA. That's a shame. I selected them because Alpine did a lot of Klippel work and uses a BL flattening mechanism for low distortion. I read their patent literature. It was the closest current analog I could find to the old XBL2 technology. Oh, and they were dirt cheap on clearance. There are not any terribly good priced 15" drivers on the market right now. An 18" makes more sense unless you're concerned about footprint. 22mm of Xmax that the Alpines have is pushing it. 30mm is a practical minimum if you're going to dump a full channel of a iNuke 6000 into them (which I am). The Alpine driver is a non-bottoming design, with the Xmech well over 30mm which makes it work. I figure I get maybe 26 to 28mm at 50% BL. Not ideal but I can live with it. I would probably select something like the Skar VXF-15 if I was doing it again, or these crazy eBay pawn shop drivers: https://www.ebay.com/itm/15-EXM-Audio-1500-RMS-3000-MAX-3-Coil-300-Oz-Magnet-259-FREE-SHIPPING-/122993448586. Those are legitimate products that are probable made by the Chinese build house that puts together the Sundown import drivers. This is not a project where you can just slap a driver in there and have it work. I modeled two dozen of them trying to find suitable replacements for the Alpines. You need a low Qes and plenty of BL. Something like the Ultimax or the old Thunder9500 models with an ugly drop in output below 20Hz, I assume since it runs out of motor. Whether this drop is real or not, I don't know.