I was trying to come up with a method of moving my woofers back to time align them to my tweeters as my crossover isn't easy to deal with. I started thinking about corner horns.
That last plot was supposed to indicate spectral balance and I thought that was the better way. Point noted for future measurements. By the way I sorted that one.
First off, I am going to confess I have not read the previous 139 pages of this discussion. I got to about page ten before my eyes glazed over. A summary sure would be nice.
Here's a data-point, for what it's worth. I have finally gotten control of the bass in my listening room. Using multiple subs only made matters worse, perhaps because the extras, left over from before my enlightenment, are of inferior quality to the main sub and the satellites.
The "answer" turned out to be simple because of lucky coincidences, and the fact that I am satisfied with good "length of sofa" response.
The room is 18 feet square. Eighteen feet is 63 Hz. Once I got 63Hz tamed, the rest went without a fight. I am using both an old Rane 15 band stereo equalizer and the five-band mono equalizer in an old Richter Scale III. I just happen to have both. I have the Rane in the tape-loop of the preamp, and the Richter Scale between the pre-amp and the sub's plate amp. Thus the sub gets the effect of both. The satellites would be down over 4dB at 63Hz in a neutral room, but they boom at 63 in this room. Luckily, 63Hz coincides with sliders on both the Rane and the Richter Scale, and the pre-amp provides 60Hz as a subwoofer crossover point. So I cross over at 60Hz in order to give the system three effective drivers at the problem frequency, and slide both of the 63Hz sliders all the way down, killing off 18dB in (partial) addition to the satellites' natural 4dB deficit! Any less is not enough.
Ahhh. It's good to be flat.
Here's a data-point, for what it's worth. I have finally gotten control of the bass in my listening room. Using multiple subs only made matters worse, perhaps because the extras, left over from before my enlightenment, are of inferior quality to the main sub and the satellites.
The "answer" turned out to be simple because of lucky coincidences, and the fact that I am satisfied with good "length of sofa" response.
The room is 18 feet square. Eighteen feet is 63 Hz. Once I got 63Hz tamed, the rest went without a fight. I am using both an old Rane 15 band stereo equalizer and the five-band mono equalizer in an old Richter Scale III. I just happen to have both. I have the Rane in the tape-loop of the preamp, and the Richter Scale between the pre-amp and the sub's plate amp. Thus the sub gets the effect of both. The satellites would be down over 4dB at 63Hz in a neutral room, but they boom at 63 in this room. Luckily, 63Hz coincides with sliders on both the Rane and the Richter Scale, and the pre-amp provides 60Hz as a subwoofer crossover point. So I cross over at 60Hz in order to give the system three effective drivers at the problem frequency, and slide both of the 63Hz sliders all the way down, killing off 18dB in (partial) addition to the satellites' natural 4dB deficit! Any less is not enough.
Ahhh. It's good to be flat.
I think a summary is in order. I really mean what I said in my previous post: When done by Dr' Gedddes method, additional subs will never make matters worse.
The method has very little to do with sub quality, and everything to do with a particular manner of implementation.
Dr. Geddes approach to multisub is not simply adding subs. I had a nice writeup of the method somewhere on my other PC . . . hopefully someone here (or Dr. G himself) can provide it. Here's my gotta-get-back-to-work summary:
Essentially each sub is carefully optimized before adding the next one. The #1 sub should be optimized using phase, volume, and XO adjustments. Then turn the #1 down a bit, and proceed to #2. The added sub is used to fill-in response holes while taming peaks at the same time, again via phase, volume, XO. Always start soft and fade in the added sub & watch for changes. If things get worse, turn down, adjust phase and try again, etc. If the addition doesn't improve matters, then move the added sub. Bear in mind that can include a vertical move. There are likely modal issues in that dimension that won't be as well addressed by more floor subs.
If it doesn't work, then something has been overlooked or skipped in the process. Really! In the voodoo world of hi-fi, it is scientifically sound, which means it requires patience and sticking to the process. None of the variables involved have anything to do with the grade of equipment, and everything to do with the airborne interaction of seperate sound sources.
I'm glad you are happy with your results. I would encourage you to not writeoff the multisub approach.
-- Mark
Here you go:
mehlau.net > audio > multisub
But, from all I know now, this isn't the end of the road.
I believe bass trapping is a fairly important part of it. The way I understand it, modes don't cover all frequencies. Exciting them all is about avoiding nulls and being otherwise evenhanded. The bass trapping action then primarily reduces the influence of those modes. I don't doubt that a good response could be had without it. The trapping would not only improve it but may increase the suitable listening area and reduce the dependence on precise variable phase adjustments, making it easier for most people to get it right.
The number of modes increases with frequency:
An externally hosted image should be here but it was not working when we last tested it.
Exiting all modes creates peaks and dips at the same time. Absorption (lots of it) reduces peaks and dips and makes modes "wider".
A so called double bass array (or CABS) creates a low frequency sound field that is nearly free of detrimental modal effects: Double Bass Array (DBA) - The modern bass concept! - AVS Forum
I've always supposed that the benefits of corner-loading as you get lower and lower in frequency were far more significant than any EQ-fudging you could do mid-wall.
In which case, it is moot whether those benefits are more important or the room-mode effects of having two woofers in corners?
I've become very interested in the wild gyrations room interaction causes. But being very wild and changing inch by inch also means that the effect of placement in one corner might be quite different than the effects of placement in another corner.
Also, under 120 Hz or thereabouts, there's little sense of localization with regular music playing (using sharp cross-overs). So any corner in a room can be used.
In other words, I'd use any corner I could grab.
In which case, it is moot whether those benefits are more important or the room-mode effects of having two woofers in corners?
I've become very interested in the wild gyrations room interaction causes. But being very wild and changing inch by inch also means that the effect of placement in one corner might be quite different than the effects of placement in another corner.
Also, under 120 Hz or thereabouts, there's little sense of localization with regular music playing (using sharp cross-overs). So any corner in a room can be used.
In other words, I'd use any corner I could grab.
Not necessarily. Looking at theoretical mode cancellation scenarios is a broad generalization and rarely useful. Room properties, location and number of low frequency sources and listener location have an effect on the low frequency response.
My experience is that at the very lowest frequencies placement has almost no effect. Certainly small movements have none. I have a problematic mode in my room at about 30 Hz. There were no feasible locations that I tried that had any real effect. But as you allude to, all rooms and setups are different.
My experience is exactly the opposite. Perceived bass response in the lowest audible octave is very sensitive to speaker placement especially for the dual 12" side firing AR9. A few inches can make a lot of difference. Floyd Toole did a lot of research and his results are summarized on Harman International's web site. They recommend 4 subwoofers all located either in the room corners or at the center of the room walls. They claim this produces the most uniform bass response.
My experience is that not only do most loudspeakers begin to falloff in output at a fairly high frequency but the relatively short dimensions of most rooms in homes causes the room cutoff frequency to be fairly high. Substantial boost to the electrical signal within the capabilites of the driver are required to compensate for that or the lowest octave just won't be heard. When the system is flat to the limit of the lowest octave, gain is so high that much very low frequency noise on many CDs that would be inaudible on lesser systems become very annoying. It is surprising how many CDs have this problem. A LF filter is very useful in these circumstances. Also acoustic feedback becomes a serious problem for every turntable I've tried including the well isolated Empire 698.
In the attached acoustic measurement, the orange trace is with two subs playing: a corner Klipschorn and a giant OB midwall (Stephens 150W, around 1960, has a resonance at 20 Hz).
The darker trace is with the OB off.
The addition of the OB, perhaps just by luck, results in flatter levels esp. at 20, 40, 55, and 110 Hz.
Taken with REW software, 1/6 8ave smoothing, Radio Shack meter correction, one mic location at listening location.
My impression is that diverse mix-and-match of auxiliary woofers is good policy.
The darker trace is with the OB off.
The addition of the OB, perhaps just by luck, results in flatter levels esp. at 20, 40, 55, and 110 Hz.
Taken with REW software, 1/6 8ave smoothing, Radio Shack meter correction, one mic location at listening location.
My impression is that diverse mix-and-match of auxiliary woofers is good policy.
Attachments
Much too coarse. If you want to show performance at low frequencies then frequency resolution should be 1 Hz, i.e. smoothing 1/24 octave or better. Showing the response up to 150-200 Hz is sufficient.
One problem with this type of measurement is how much of the output is doubling and how much is actually fundimental bass. Despite Klipschorn's huge size, I'd expect that below about 35 hz, its response would fall off a cliff. AR1W by contrast is designed for a system F3 of 42hz with a 12 db per octave slope. At 30 hz, it's about 5% THD, still very respectable. Its relatively gradual slope makes it equalizable. How does AR1W compare to the Klipschorn within AR1W's power handling capacity? I'd expect it would do better especially when equalized. What's your opinion?
Too coarse? Smoothing is there to create a picture that illuminates some issue not to "hear by eye."* But, attached are unsmoothed curves. I patiently await learning from Markus what the earlier, 1/6 smoothed curves may have concealed (and what insights are now revealed to him) by unsmoothing the curves, apropos this discussion.
The longer curves were presented to foster (and illustrate) the matching of the with-Stephens and without-Stephens curves.
Soundminded points out,
Traces were made at pretty low levels like 70-80 dB and so no doubling likely. But as the level goes up, chotchkas in the room do start to jiggle. On the other hand, you can see the levels are well above the noise floor. I don't want to over-sell the enduring significance of these quick and crude little curves. They illustrate one point suitably and that is all. That is what good experiments do.
Yes, the Klipschorn (the bass unit of which is not terrible "huge") does "fall off a cliff"... exactly as my curve shows. But I was quite surprised at how much oomph there is low down. And even more surprised at the low bass of the OB!!! I sure don't know what is going on when I look at the response below 20 Hz.
Only Klipsch and Stephens in light trace. The dark trace is only Klipschorn. No AR-1. I haven't done measurements with the AR-1 and my new REW software. But the AR-1 with a bump around 42 Hz is a great, full-sounding woofer. I've been arguing that except for fans of earthquake movies, there's not much on recordings much lower and few speakers make organ music (esp. Saint-Saens 3rd Sym) sound as good as the AR-1 with the 12-Hz driver.
The curves have EQ so they don't represent any kind of natural response of those speakers. Also, as mentioned previously, the RS meter correction (LF only) is present. They do represent the very salutary effect of adding a second, profoundly different woofer to the ear-location results.
*Evaluating by eyeball the terrible visual effects (but trivial aural effects) of comb filters is an example of that fallacy.
Attachments
Last edited:
What's the frequency resolution of the unsmoothed data?
Modes can have a very high Q so our eyes need to be able to see what our ears hear, hence 1/24 resolution is needed.
In your graph there's a huge 25 Hz wide "hole" around 50 Hz which I would fix. These kind of holes are tricky because something that cannot be heard can't sound bad. Beside of that, this another example why measurements are vital in order to improve sound reproduction.
As you are probably aware, the history of Acoustic Research and the AR1W woofer is about Edgar Villchur who could not sell his idea to anyone, other manufacturers of the time said it wouldn't work. He went into production himself and around 1956 the NY Audio League, the predecessor of AES took 4 AR1s to Riverside Church in NYC where they held a live versus recorded demo against an Aolean Skinner pipe organ. I think they used 150 watt Western Electric amplifiers. This very successful demo proved the superiority of the AR woofer design leaving no doubt. It was the king of the hill for a long time in LF reproduction. The late 1970s/early 1980s Teledyne AR9 (there is a later speaker called AR9 which is not the same) uses a later variant (different materials but the same design and nearly identical performance) of the 12 in AR woofer. Two per 4 cubic foot tower with an F3 of 28 hz. And it does do earthquakes. It can also shatter windows, rattle floors, walls, and loosen fillings in your teeth. It is interesting and surprising to me that all these years later, restored speakers of this type will still give the best on the market a run for their money. The new Dayton dual 12" side firing drivers looks promising. With a built in 1000 watt amplifier for $1000 or less it seems like a bargain. A pair of those would probably do well compared to anyone's speakers.
I share Soundminded's appreciation for this 65 year old concept. I remain puzzled that you can't buy a 12-Hz woofer. How come? The AR-1s seem durable, ahem, ahem. There's no woofer enclosure concept that makes much sense to me or is without profound (I repeat, "profound") shortcomings except a sealed box or a large horn with a sealed box behind the driver.