I set up a new system in our temporary rental flat over the weekend, using Dr. Geddes’ multisub setup procedure. The results, I think, show first the basic efficacy of Dr. Geddes’ methods (even with likely performance-degrading liberties, taken on time/cost/aesthetic grounds), and second, show that even someone with relatively modest means or rigid placement constraints can achieve clean bass if they follow a simple 4-step plan: first, commit to multiple subwoofers; second, learn how to take good spatially-averaged measurements; third, follow proven setup/placement methods for multisubs; and fourth, use minimally suitable components, i.e. woofers with shorting rings in their motors that have enough volume displacement to reach the SPL you want, and are powered sufficiently to reach that SPL.
I only want to say as little as required about the gear itself, before getting to the measurements.
I wanted to play with the miniDSP because it seems to offer a lot of cool functionality for minimal cost in an easy-to-hide package. So I bought an miniDSP-in-a-Box with unbalanced I/O for sub EQ. I used their “Advanced 4-Way Plugin” to EQ and provide level/delay adjustment. Interestingly, I found I needed to step down the jumpers to “rev. A” (0.9Vrms sensitivity) position, even though my AVR has been tested to put out 4Vrms from the preouts.
Note that the “full Geddes” would use 4th-order bandpass subs for every position. However, mine are all direct radiating passive radiator (PR) units, not bandpass subs. The main sub (S1) is a DIY unit, 40L cube with a Peerless XLS12 (830500), Peerless XLS12-PR, and Dayton 240W plate amp. I built it maybe 7 years ago, along the lines of the Peerless XLS12 Application Note but with more mass on the PR for a lower tune. It has been in storage for the past five years or so. I put a 48dB/oct highpass on it at about resonance (17Hz). The second and third subs (S2 and S3) are both KEF’s HTB2. These are thin UFO-shaped things with 15L cabinet volume, a beefy-looking 10” woofer, a 10” PR, a claimed 250W, and basically no onboard controls. I used them for two reasons: first, they both came with the mains in use (a recent craigslist purchase, KEF's KHT3005SE set), and second, they are small enough to easily hide. Also, one thing that interested me is whether having 4th order rolloff on all mains would hurt the deep bass because of interference. I hoped the KEF saucers would be OK, because I didn’t want to build something I would just scrap when I pull out my big Aurasound-based subs in a few months.
Basic placement (one in a corner, one far away, the third far away from both) and measurement procedure (adding in subs progressively, based on distance from mains) are per Geddes as documented by Mehlau. However, I could not place the subs ideally. Because the floorplan is open, I couldn't get a sub behind the listening position. Also, for aesthetic reasons I couldn't get a sub above the room's centerline.
I use a receiver with Audyssey’s MultEQ XT room correction program. All Audyssey bits were turned off for the below measurements, except where indicated.
I only want to say as little as required about the gear itself, before getting to the measurements.
I wanted to play with the miniDSP because it seems to offer a lot of cool functionality for minimal cost in an easy-to-hide package. So I bought an miniDSP-in-a-Box with unbalanced I/O for sub EQ. I used their “Advanced 4-Way Plugin” to EQ and provide level/delay adjustment. Interestingly, I found I needed to step down the jumpers to “rev. A” (0.9Vrms sensitivity) position, even though my AVR has been tested to put out 4Vrms from the preouts.
Note that the “full Geddes” would use 4th-order bandpass subs for every position. However, mine are all direct radiating passive radiator (PR) units, not bandpass subs. The main sub (S1) is a DIY unit, 40L cube with a Peerless XLS12 (830500), Peerless XLS12-PR, and Dayton 240W plate amp. I built it maybe 7 years ago, along the lines of the Peerless XLS12 Application Note but with more mass on the PR for a lower tune. It has been in storage for the past five years or so. I put a 48dB/oct highpass on it at about resonance (17Hz). The second and third subs (S2 and S3) are both KEF’s HTB2. These are thin UFO-shaped things with 15L cabinet volume, a beefy-looking 10” woofer, a 10” PR, a claimed 250W, and basically no onboard controls. I used them for two reasons: first, they both came with the mains in use (a recent craigslist purchase, KEF's KHT3005SE set), and second, they are small enough to easily hide. Also, one thing that interested me is whether having 4th order rolloff on all mains would hurt the deep bass because of interference. I hoped the KEF saucers would be OK, because I didn’t want to build something I would just scrap when I pull out my big Aurasound-based subs in a few months.
Basic placement (one in a corner, one far away, the third far away from both) and measurement procedure (adding in subs progressively, based on distance from mains) are per Geddes as documented by Mehlau. However, I could not place the subs ideally. Because the floorplan is open, I couldn't get a sub behind the listening position. Also, for aesthetic reasons I couldn't get a sub above the room's centerline.
I use a receiver with Audyssey’s MultEQ XT room correction program. All Audyssey bits were turned off for the below measurements, except where indicated.
Last edited:
Setup and Measurements
Enough boring stuff! Here is the measurement setup. I used a Velodyne SMS-1 to generate sweeps and record measurements. The SMS-1 was not in the subwoofer output chain, only on the input side. So ignore any slider or volume control setting you may see. To effect spatial averaging in a quick and easy-to-view manner, I used the SMS-1 in concert with Velodyne's MIC-5 spatial averaging kit. Basically, the MIC-5 is 5 Behringer ECM-8000s with Velodyne silk-screening, and a mixer/preamp box. Horribly overpriced in today's world of accurate-enough Audyssey tower mics and programs such as FuzzMeasure that can average multiple measurements, but six or so years ago it was by far the easiest way out there to take spatially-averaged bass measurements.
Here is the mic setup (our new kittens were banished during actual measurement!):
NOTE: I wore Etymotic ER-20 earplugs for all sweeps. I’ve no interest in exposing myself to non-musical signals from my system at full blast. I recommend you use earplugs while measuring, too!
And...finally, MEASUREMENTS
Mains with the 120Hz highpass suitable for their size:
S1 raw response, front-left corner, no EQ, 48dB/oct highpass at 17Hz.
(The notch at the top of the graph is from the mains' floor bounce, which from other measurements is centered at about 230Hz.)
S2 raw response, ~25% down the right wall, no EQ, no highpass:
S3 raw response, ~70% down the left wall, practically nearfield, no EQ/ highpass:
Summed response S1+S2+S3, after just optimizing levels and delays on the miniDSP, prior to application of any EQ
Summed response S1+S2+S3, after Audyssey calibration (MultEQ XT on, DynamicEQ off) and three bands of fairly mild EQ cuts
Enough boring stuff! Here is the measurement setup. I used a Velodyne SMS-1 to generate sweeps and record measurements. The SMS-1 was not in the subwoofer output chain, only on the input side. So ignore any slider or volume control setting you may see. To effect spatial averaging in a quick and easy-to-view manner, I used the SMS-1 in concert with Velodyne's MIC-5 spatial averaging kit. Basically, the MIC-5 is 5 Behringer ECM-8000s with Velodyne silk-screening, and a mixer/preamp box. Horribly overpriced in today's world of accurate-enough Audyssey tower mics and programs such as FuzzMeasure that can average multiple measurements, but six or so years ago it was by far the easiest way out there to take spatially-averaged bass measurements.
An externally hosted image should be here but it was not working when we last tested it.
Here is the mic setup (our new kittens were banished during actual measurement!):
An externally hosted image should be here but it was not working when we last tested it.
NOTE: I wore Etymotic ER-20 earplugs for all sweeps. I’ve no interest in exposing myself to non-musical signals from my system at full blast. I recommend you use earplugs while measuring, too!
And...finally, MEASUREMENTS
Mains with the 120Hz highpass suitable for their size:
An externally hosted image should be here but it was not working when we last tested it.
S1 raw response, front-left corner, no EQ, 48dB/oct highpass at 17Hz.
An externally hosted image should be here but it was not working when we last tested it.
(The notch at the top of the graph is from the mains' floor bounce, which from other measurements is centered at about 230Hz.)
S2 raw response, ~25% down the right wall, no EQ, no highpass:
An externally hosted image should be here but it was not working when we last tested it.
S3 raw response, ~70% down the left wall, practically nearfield, no EQ/ highpass:
An externally hosted image should be here but it was not working when we last tested it.
Summed response S1+S2+S3, after just optimizing levels and delays on the miniDSP, prior to application of any EQ
An externally hosted image should be here but it was not working when we last tested it.
Summed response S1+S2+S3, after Audyssey calibration (MultEQ XT on, DynamicEQ off) and three bands of fairly mild EQ cuts
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Conclusions
I don’t think one can argue with the measured results, honestly. There’s nothing at all magical about the subs I used. Dr. Geddes’ methods just work, even with less-than-elite gear.
As far as sound, it tracks the measurements. The bass just sounds locked-in. The bass is tuneful and in time, which makes the system sound like coherent music rather than a collection of speakers. (I am a little surprised that the 120Hz highpass on the mains hasn't brought up any localization issues. Yet, at least.) Until you run up to its output limits the system sounds way larger than the speakers would indicate. It won't exactly make me forget the big Tannoys and Aura NS18/NS15/NS12-based subs in storage, but it's certainly good enough sounding for a temporary abode.
The biggest improvements from running just S1 with Audyssey calibration, subjectively, are better time integration and less ringing. With one sub, the bass seemed to drone, and lag behind the rest of the music. Even after Audyssey calibration. (Pink Floyd's "Time" on DSOTM is a good track for quickly testing bass lag.)
Another important implication is that one does not need to use identical subwoofers in a multisub system. In fact, identical subs is actually rather inefficient, given the different roles of the subwoofers. While neither Dr. Geddes nor Dr. Toole make that explicit in their recommendations, it’s easy to infer. Geddes’ method by definition has most of the bass coming from the main sub, with the other two subs just filling in holes. That means the "BroadBand" subs, to use his nomenclature, do not need to be as capable. Toole impliedly agrees. As he wrote about the 4-sub system in his home, equalized via. Harman’s “SFM” method, “Interestingly, the algorithm found a need to reduce the output from three of the four subwoofers, two by 6 dB and one by 12 dB." Sound Reproduction, at 233. Obviously, if a sub is playing 6dB or more less than another, that does not need to have the same capability as the one playing on level. Arguably, having the same capability in all subs is wasteful, even! If nothing else, it makes placement more of a chore, because less capable subs are generally smaller and thus easier to place/hide.
I think the ability to use smaller subwoofers in the supplementary positions without an iota of sonic penalty is a very important point, often lost, that makes a good multisub system both more cost-effective than it appears when one first says "buy 3+ subwoofers!" and far easier to integrate into a room. In a system with extremely high ULF SPL capability, it is certainly easier to hide two more one cubic foot closed boxes with 10's or 12's than to hide two more five cubic foot boxes with 18's! Or, in a system the scale of this apartment system, I simply couldn't/wouldn't fit three 40L subs. But a 40L sub and two 15L subs, that was doable.
I don’t think one can argue with the measured results, honestly. There’s nothing at all magical about the subs I used. Dr. Geddes’ methods just work, even with less-than-elite gear.
As far as sound, it tracks the measurements. The bass just sounds locked-in. The bass is tuneful and in time, which makes the system sound like coherent music rather than a collection of speakers. (I am a little surprised that the 120Hz highpass on the mains hasn't brought up any localization issues. Yet, at least.) Until you run up to its output limits the system sounds way larger than the speakers would indicate. It won't exactly make me forget the big Tannoys and Aura NS18/NS15/NS12-based subs in storage, but it's certainly good enough sounding for a temporary abode.
The biggest improvements from running just S1 with Audyssey calibration, subjectively, are better time integration and less ringing. With one sub, the bass seemed to drone, and lag behind the rest of the music. Even after Audyssey calibration. (Pink Floyd's "Time" on DSOTM is a good track for quickly testing bass lag.)
Another important implication is that one does not need to use identical subwoofers in a multisub system. In fact, identical subs is actually rather inefficient, given the different roles of the subwoofers. While neither Dr. Geddes nor Dr. Toole make that explicit in their recommendations, it’s easy to infer. Geddes’ method by definition has most of the bass coming from the main sub, with the other two subs just filling in holes. That means the "BroadBand" subs, to use his nomenclature, do not need to be as capable. Toole impliedly agrees. As he wrote about the 4-sub system in his home, equalized via. Harman’s “SFM” method, “Interestingly, the algorithm found a need to reduce the output from three of the four subwoofers, two by 6 dB and one by 12 dB." Sound Reproduction, at 233. Obviously, if a sub is playing 6dB or more less than another, that does not need to have the same capability as the one playing on level. Arguably, having the same capability in all subs is wasteful, even! If nothing else, it makes placement more of a chore, because less capable subs are generally smaller and thus easier to place/hide.
I think the ability to use smaller subwoofers in the supplementary positions without an iota of sonic penalty is a very important point, often lost, that makes a good multisub system both more cost-effective than it appears when one first says "buy 3+ subwoofers!" and far easier to integrate into a room. In a system with extremely high ULF SPL capability, it is certainly easier to hide two more one cubic foot closed boxes with 10's or 12's than to hide two more five cubic foot boxes with 18's! Or, in a system the scale of this apartment system, I simply couldn't/wouldn't fit three 40L subs. But a 40L sub and two 15L subs, that was doable.
I haven't got the capability to try this out yet, but I am thinking of something something similar myself.
One thought I've had is while you use one sub as the main provider, with fillers applied as necessary, it should be possible to use all the subs together below the rooms modal regions.
In a small room you could use three/four small identical subs. 1 or 2 of them would be used as the mains and then 2 would be used as filler subs. The filler subs would obviously contribute far less then the mains. Below the rooms modal region however you should be able to bring those 2 filler subs up in level to match the level of the main subs too, giving the potential for more overall output at 20hz for HT and the like. Of course, as you bring up the level of the filler subs you'd have to reduce the level of the main subs for things to remain flat, but it seems like it should be doable, especially with a DSP.
Of course you might just argue, then why not use a really big main sub, as modal interaction isn't important down low, just have one capable main sub. Maybe I don't want a large sub in my small room!
Speaking along the lines of having one main large sub, it does really show that the filler subs don't actually need to be subs at all. In my small room the lowest room mode occurs at around 40hz, so it stands to reason that I'd only really need useful output down to around 40hz from the fillers. As the fillers are used far less then the main subs, perhaps you could get away with well implemented 8" or even 6" drivers. It depends how loud you really want the end system to go.
One thought I've had is while you use one sub as the main provider, with fillers applied as necessary, it should be possible to use all the subs together below the rooms modal regions.
In a small room you could use three/four small identical subs. 1 or 2 of them would be used as the mains and then 2 would be used as filler subs. The filler subs would obviously contribute far less then the mains. Below the rooms modal region however you should be able to bring those 2 filler subs up in level to match the level of the main subs too, giving the potential for more overall output at 20hz for HT and the like. Of course, as you bring up the level of the filler subs you'd have to reduce the level of the main subs for things to remain flat, but it seems like it should be doable, especially with a DSP.
Of course you might just argue, then why not use a really big main sub, as modal interaction isn't important down low, just have one capable main sub. Maybe I don't want a large sub in my small room!
Speaking along the lines of having one main large sub, it does really show that the filler subs don't actually need to be subs at all. In my small room the lowest room mode occurs at around 40hz, so it stands to reason that I'd only really need useful output down to around 40hz from the fillers. As the fillers are used far less then the main subs, perhaps you could get away with well implemented 8" or even 6" drivers. It depends how loud you really want the end system to go.
I haven't measured it, to be honest. I would guess the listening "zone" is maybe 16-18 feet wide, maybe 13-15 feet deep, with 9ishfoot ceilings. But there's also a kitchen separated by a bar, and an open dining room. The primary listening spot is 11' from the center channel speaker, which is about 3.5' from the wall behind it.
Even still, the limiting factor for bass output is unwillingness to needlessly antagonize the neighbors, not the system itself.
Yes. That's one of the big benefits of using subs with 2d order rolloffs (sealed/4th-order BP). They still contribute quite a bit down low. Especially when there's a bunch of them.
I think that it's probably best to use one main sub. Two subs playing on level are just a lot harder to get "playing nice" together.
I think that's a better situation than the one I've created, more at least more efficient. Though it doesn't much matter in my case. S1 is capable of 105dB at 25Hz (on paper, not tested), and at that level the mains are the limiting factor.
Typical relatively small sealed boxes with 8's or 10's will roll off above 40Hz.
But overall I think there's great merit to your idea. If I were doing a small multisub system like this one from scratch, I'd probably use two 12's for the sub in a sealed or 4th-order BP cabinet (I still might replace the PR with a second XLS12), and probably use good 8" woofers in closed boxes for the filler subs. MAYBE a 10 for S2 and an 8 for S3. Keep 'em all sealed so they sum for greater ULF output.
Last edited:
I've currently got 4 XLS10s in small 1' cubes. Combined with room gain and boundary re-enforcement they'll hit 105dB @ 20hz @ rated xmax (theoretically!). I've haven't yet had the time to play with them as the DSP I am building isn't complete. Depending on how things go I'll have to see what the best way to integrate them is.
The mains will go down as far as I want to EQ them to go, but an optimal extension that matches the amplifier power and driver xmax is around 60hz with a 4th order roll off. The idea was to have two of the XLS cubes going full whack, with the other two used as fillers, of course I can always put two cubes on top of one another if that works out better.
The mains will go down as far as I want to EQ them to go, but an optimal extension that matches the amplifier power and driver xmax is around 60hz with a 4th order roll off. The idea was to have two of the XLS cubes going full whack, with the other two used as fillers, of course I can always put two cubes on top of one another if that works out better.
It's worth taking a look at the slimline car subs that are available now. I'm using 2 x Pioneer 10" TS-SW1041D in 17ltr (0.6cu ft) sealed boxes, with DSP/ EQ pushing up the low end they perform really well (loads of Xmax) and don't take up much space.
I think that's a very good point. Especially when the ability to hide a subwoofer is more important than its actual output.
My only caveat would be to make sure they're otherwise well-designed. Most car subwoofers are designed for boom rather than fidelity, which in practice means no Faraday rings, and poor performance in the 60-120Hz octave. Some of them are very well designed though. The Alpine Type R's propaganda says the right things (shorting rings, normalized inductance of under 1mH/Ω, attention to thermal issues, etc.) but they're also quite expensive.
If it ships as a regular production sub, the Stereo Integrity BM is another one that looks really good. And because it's an internet-direct line it should cost less than subs from the "big guys" sold primarily through car-fi shops and venders such as Crutchfield.
Dayton also has a neat line of shallow woofers that are quite inexpensive. Peerless also now has a line of shallow woofers with cool-looking inverted motors and prices lower than typical car-fi woofers. All of them have potential for the "BroadBand" subs, though for now at least I would stick to a conventional driver for the "ULF" position. The shallow ones just don't have the volume displacement of "regular" woofers. Yet, at least.
Building your own DSP? You're clearly way more advanced than I am! I think four XLS10's could be a great system with plenty of clean output for most people's needs. Peerless really did a great job on that series of woofers, and outside of some of the pro stuff with much bigger voicecoils (and much higher cost) there doesn't seem to be anything else clearly better at near the price.
IMO, I would start by putting two of them in the same corner, one on the opposite wall far away from those two, and the third high up and as far away from the other two as possible. That may require making some of the subs a different shape. I find shallow and taller is often easier to hide than a cube. Fortunately the XLS doesn't have a pole vent (the vc is vented radially from holes in the cone under the dustcap), so you can make the cabinet just deep enough to fit the driver.
I would seal the mains. It's just easier to integrate things that way, and the extra bass won't be missed with four XLS10's underneath. If the mains have enough cone area (considering the size of your subs, I would say two 7" woofers or better), I would highly recommend running them wide open, to provide more bass sources to smooth out room modes, and bringing in the subs at maybe 120-150Hz. The high corner depends on the size of your room more than anything. That way, every big cone contributes to ULF AND to smoothing the room response.
Last edited:
- Status
- Not open for further replies.