“Audio” means the machines that make music from recordings in the home. I’ve been working on a new amplifier, to multi-amp the Bill Woods-designed Yorkville U215s that I have around here. This has been a longstanding ambition. But, that is not done yet. Along the way, I looked around at the current state of affairs in the audio hobby.
Mostly, it hasn’t changed much. There is a lot of gear hype, and still a focus on little shoebox speakers that are very nice if you want something small and convenient that doesn’t cost too much, because you have better things to do than worrying about “audio,” but are not (in my opinion) where you want to focus if you have a little ambition.
I don’t bother following the chit-chat much, either in mags or review sites, or at forums like DIYAudio.com, since I already know what I want to do for the next several years.
Even in the shoebox-speaker category, I would focus on horn designs. I’ve mentioned that Jonathan Weiss of Oswald’s Mill Audio is a friend of mine going back to the “tube and speaker tasting” days around 2001. More than anyone, he creates what I think of as ambitious audio devices. I was there the first time he hooked up a 300B SET amplifier to his big RCA horns, replacing some kind of EL34 Williamson-type amplifier. (It was a lot better.) Heck, I bought those RCA horns, at least the basshorns, which we lowered out of the Mill with an electric winch. Of course he has gone 10x beyond me since then. I’ve heard some of Jonathan’s designs including the Mini, DeVille and Excelsior, which are all small, and they were astonishing — because of their horns.
Meanwhile, Magico released the M9, “only” $750,000 a pair, and they look like this:
An enormous amount of effort and optimization went into this, as you can imagine. And it is certainly “good,” and in fact very, very, good, within its genre. But, what I see here is: a six-inch cone, and a one-inch dome, which makes most of the music, aided by some woofers on the low end. This format has considerable inherent limitations. The one-inch dome is simply not capable of very much output, and begins to max out at quite low levels. The crossover to the 6″ woofer is typically around 2200hz, which is too low for the dome, and too high for the cone, already getting well out of its non-directional pistonic range. The DeVille shown above has a 1″ horn, which has practically unlimited power handling, at least in the home environment. Also, it is probably crossed around 1500hz, which is very comfortable for the compression driver, and also is low enough that you can push the 6″ cone very hard without worrying about breakup in the sensitive midrange.
For example, here is some data on the Scan-Speak Revelator 7″ midrange driver, the kind of high quality item that you often find in ambitious speakers like those from Magico.
You can see that there is a very nice range between about 100hz and 1000hz, where this cone is very well behaved. The red and green lines show off-axis response. As frequencies rise, the wavelength becomes smaller than the size of the cone. The result of this is increasing “directionality,” or high-frequency falloff when you are off-axis. We can see this falloff happening beginning around 1000hz. This is not too bad. But, from about 1200hz, the response is no longer so smooth and flat. There is a distinct rise, and then, beginning about 2500hz, a more jaggy response, and then a high-frequency falloff. The rising and then jaggy response is the cone going into breakup. It no longer moves as one rigid mass, but starts to resonate. The Scan Speak Revelator drivers have a lot of fancy technology to minimize this breakup, but you can see it here nevertheless. You can see that there is a region, between 1000hz and about 2200hz where we might cross over to a tweeter, where the response is … pretty good, this is one of the best drivers of its kind made today … but not really ideal. It gets worse, the more power you put through it. Also, the range beyond the crossover, from 2200hz to 6000hz let’s say, is still somewhat audible and does affect the overall impression of the speaker. Nevertheless, there is output, so you could use this driver up to about 6000hz if you wanted to.
But, all of this naturally makes you want to cross to the dome at a lower frequency, to avoid some of these higher-frequency difficulties of the 6″ driver. Since the dome is much smaller, it works better at higher frequencies. Here is a 1″ dome tweeter, also from Scan Speak, and also a very expensive and premier-quality item:
Here we have a range of pretty good performance from about 600hz. We see also the off-axis falloff from about 6000hz, or about 6x higher than the 6″ driver, which makes sense since this is 1/6th the diameter. It is a little jiggy but I wouldn’t worry about that much. What you don’t see here is power handling. A 1″ dome can’t produce much output, and distortion rises with output. The lower crossover you use, the more of the musical signal you are putting through the tweeter, and thus the more air you are asking it to move. It has to move back and forth to move this air; and it is hard for a little dome to move back and forth much, while also producing the very high frequencies above 5000hz in an appealing manner. In practice, we know that these kinds of speakers can sound pretty good, at the levels used in the home. They have made billions of people happy. But, there is a reason why you never see them in professional applications.
One thing I’ve noticed a lot of, in audio, is imitation and optimization. The speaker above basically began as a simple two-way shoebox or “monitor” speaker, like the LS 3/5A from the 1970s, or the original Advents. This basic format, a 6″-ish driver and a 1″-ish dome, was imitated and optimized, resulting in the Magico M9. But, along the way, very few people stop to ask the questions that I ask, such as: Is a 1″ dome really the best we can do? Anyway, after hearing big horns, or, for the top end, bullet tweeters or ribbons, combined with large-format compression drivers, I don’t have so much interest in cones and domes anymore.
To take another example: We’ve had thirty years now of 300B single-ended amplifiers. But, is single-ended really the best topology? Should we use push-pull instead? There are a lot of good reasons to use push-pull. Here, Lynn Olson thought about it, and went with a 300B push-pull design.
So you would think that, after several decades now, there would be an interesting variety of 300B push-pull amplifiers. But, there are almost none. 99.9% of 300B amplifiers are still single-ended. Mostly, this is just imitation.
For $1300 (and less if you look), you can try a pair of Klipsch The Sevens, which also have a 1″ compression driver and a 6″ cone.
Also, this is a fully multi-amped speaker, with all the digital bits and amplification built in. You just bring a smartphone and $1300, and you are listening to music, which, in some important ways, will be better than you will get from the Magico M9.
Oswald’s Mill Audio has a YouTube channel, which I have been enjoying recently. Here is a video on the Excelsior system, which is basically a “convenience/lifestyle system.” That’s how small it is. I was amazed at the “big sound” that these produced — basically because of their conical horns. I think the constant directivity characteristic of conicals helps create some of this “bigness,” as it fully energizes the room in the upper frequencies, avoiding the unnatural “beaminess” of typical horns.
Jonathan likes to complain about the state of “high end audio.” I think this is a bit of a posture, but I actually agree with most everything he says. He gets warmed up around 5:00 here:
Meanwhile, you can see Jonathan Weiss’ “reference system” or main home system, here:
Basically, big horns. But, you could build something like this in your own home for probably under $5000. The RCA MI-1428B field coil compression drivers, from the 1930s, are serious unobtanium today. But, you can also buy the B&C DCM50, based on the 1428B (I think Bill Woods, who designed the speakers for OMA and also the horn in the picture above, had a main role in their design), and I think it would be almost as good, and they are $488 each at Parts Express. Make some horns like you see (it’s a 40 degree conical, simplest thing in the world), and you are in business. These 12-sided horns involve tricky woodworking, but you can make a simple four-sided square and it would be almost as good. These horns are around 110db, compared to about 90db for the Magico M9. This means that they are 20db or 100x more efficient. The amount of sound pressure that you would make with 100 watts on the M9, takes one watt on these horns. But, at one watt, these horns are loping along gently, like a V12 Mercedes at 45 mph, while that 6″ cone and 1″ dome would be rattling themselves to death if you put 100 watts through them.
Bill Woods, who spent his whole career designing horn speakers, was a regular attendee at the “Tube and Speaker Tastings.” After decades playing with this and that horn design, he became a fan of the straight conical horn, which is why OMA uses them and why they have also become common throughout the prosound sector. Besides their constant-directivity characteristics, the straight horn sides lead to less coloration of the sound. Woods would demonstrate with a typical vintage exponential-type horn, which would produce familiar horn colorations. Then he would take a conical horn, and show that these colorations disappeared. This is something that is almost never discussed, but it is very obvious when you hear it in person. I am not sure that you can even measure these “colorations,” using a typical frequency analyzer. Both horns would measure flat. Typically, we would use vintage horns despite their colorations, because they sound good; and, over time, we would get used to these colorations, along with their beaminess and all the other vintage-y things. But, it is nice to now be able to not have them.
Bill Woods, in the 1970s. Yeah that guy definitely doesn’t know anything about horns.
These horns are big — very big — but I think that is mostly just an aesthetic decision, to make them look good with the very big basshorns. Basshorns have to be big for acoustic reasons — wavelengths in that range are very big. At 100hz, the wavelength is about 10 feet. But, the horns don’t have to be so big, except to look cool. They do work better when they are bigger, but not that much better. The minimum size of a horn mouth is a circle whose circumference is one wavelength at the cutoff frequency. At 500hz, the typical cutoff for a large-format 2″-exit compression driver like a JBL 2440, the wavelength is about two feet long; this means a circular horn eight inches across. Basically, this:
(It’s rated for 350hz because it would be mounted on a flat speaker baffle, thus operating in effective half-space. In practice, you would cross it around 500hz, but it would still have effective loading down to 350hz, making it an easy load for the driver.)
In the Magico M9, the 6″ driver is probably crossed to the 12″ woofer around 400hz. This horn would cross around 500hz. Almost the same. Add a bullet tweeter like the B&C DE35, crossed around 5khz, and you would be covering the same range as that 6″ cone/1″ dome combo — but with big horn sound. It wouldn’t cost much. And it would be almost the same size, an 8″ horn exit instead of a 6″ cone. (The bullet tweeter is about three inches wide.)
To be fair to the Magico-type people, one advantage of using cones and domes, in a project like the M9, is that it meshes well with the cone woofers. We aren’t blending two very different devices, cones and horns. Also, horns tend to big, which means that multiple horns tend to be very far apart, which introduces integration problems. Horns also have very different shapes. If you put a long midrange horn and a short bullet tweeter on the same flat plane (the front of a speaker box), the drivers will be significantly out of line, introducing timing issues. Plus, the horns and bullet tweeters have significantly different dispersion patterns. All of these things introduce integration difficulties. A simple 6″ cone and 1″ dome blend together well, since they are similar devices on the same flat plane, and close together. But, these are the kinds of things you have to struggle with, with an ambitious project.
Magico itself is no stranger to big horns. Here is the “Ultimate III”:
Note how Magico itself calls these the “Ultimate.” And I am sure they sound good, because it is almost impossible for them not to sound good. It is super-expensive, although, not much different than the M9. Maybe Magico built the M9 simply because that’s what people wanted, even though they should be buying the Ultimate III, which Magico developed several years earlier, but maybe it just didn’t sell. Maybe people with $750,000 to spend on a speaker don’t want something like this in their living room. Kinda sensible, actually.
Side view, in all their ugliness. “$600K+ to put that in my living room? Naaaaah, I’m good thanks.”
You could literally run these big horns, in a multi-amp setup, on a one-watt amplifier. The one-watt amplifier I would use here is the MicroZOTL, based on the ZOTL technology from David Berning. This has been around a while, and you can find them on the used market for not too much.
Basically, you end up with OMA’s Museum Speaker:
This is actually quite a lot like the Magico M9. Instead of two 12″ woofers, there is a single 15″ sealed woofer for the lower midrange and upper bass (around 80-500hz). Then, there is another 15″ sealed subwoofer for the bottom, 20hz-80hz. (The Magico M9 has two 15″ subwoofers.) It is 50″ high and 22″ wide, so not real big. The horn you see is maybe about 10″ high and 15″ wide, not huge. It uses a coaxial compression driver, where the large format compression driver and the equivalent of the B&C DE35 bullet tweeter are combined in one unit. It has a rather dramatic grille. I like this very much. It is very expensive. These are ambitious, Ferrari-like devices, at Ferrari-like prices and Ferrari-like diseconomies of small scale. But, it is a basic format that other companies can produce for not too much money, let’s say under $10,000. This is the kind of thing I would like to see as a typical audiophile aspiration, that affordably fits in a typical living room, not another shoebox speaker with 6″ cones and 1″ domes. Basically, it would be the Chevy C8 Corvette of speakers. Great execution, great value and more than enough even for ambitious people.
A similar thing today, that you can buy for not too much money, is the Klipsch Cornwall.
This guy, Steve Guttenberg, has been an audiophile forever. He used to work at Sound By Singer, one of New York City’s biggest and most famous audio retailers. So he has heard everything, over a period of decades. Recently, he had some Cornwalls in his home:
https://www.youtube.com/watch?v=EIHTMGIxLxc&t=813s
The Cornwall is basically the same as the Museum Speaker, with a large compression driver, a small tweeter horn, and a 15″ woofer handling the bass. It uses a simple bass-reflex port for low-end extension instead of a subwoofer. They are $6,600 a pair at Crutchfield — an excellent price. The Klipsch stuff is always a great value, but typically a rather budget offering with less-than-great drivers or horns. For example, the coaxial compression drivers in the Museum Speaker are probably from B&C or BMS, and cost about $700 each, or $1400 for the pair — just for the wholesale driver cost, not including the woofers, which are at least $1000 more. This is what I would use, but you can’t do that and sell it for $6,600. Nevertheless, if you want something that you can buy today with your Amex, the Cornwall is a great place to start. These have been made continuously since the 1960s (in various iterations, this is the Cornwall IV), so used speakers are available in every metropolitan area. Older speakers often have maintenance issues, but you might get a pair for less than $3000 — something even a high schooler can afford.
Basically it is the Ford Mustang GT of speakers. A good speaker at a great value, and a format that hasn’t changed much over the years — for good reason.
One thing about these older, classic speakers is, you can buy them used and use them for a decade, and then sell them for more than you bought them for. Most of my speakers — the Fertin field coils, the Altec A5, the Altec 755C, the Western 720As, the Altec Model 14s, the RCA basshorns, individual drivers from Diatone, Altec, Supravox — I later sold for more than I bought them for.
We’ve been waiting for the speaker genius of our generation, Tom Danley, to come out with some products aimed at the home market. He did — not one, but four speakers, all of them quite innovative by the looks of it.
In the little box speaker category, we have this:
What is it? Danley says that it is a “Synergy Horn,” which looks like it covers everything down to about maybe 120hz — horn loaded — with this little pepper shaker thing.
How is that possible?
Here’s my guess: The Tapped Horn design, which is part of the Synergy Horn, covers about a 3x wavelength range. In other words, a range from 150hz to 450hz (3x). Then, you can have a regular front horn operating in half-space (the front baffle of the cabinet), which would be about 6″ wide at 450hz, or maybe even quarter-space (4″ wide) as we see in some of Danley’s other designs. The port comes in around 150hz and probably has a peak around 100hz. With active DSP you can correct some of the oddities, and maybe get a pretty nice result.
For the home, a different version, with a woofer on the bass:
Here the “pepper shaker thing” has a fabric cover. Unlike the similar-looking Klipsch speaker, which is probably crossed around 1500hz, the horn here probably goes below 200hz.
Basically it is a pair of 15″ subwoofers, and a horn going down to about 100hz I would guess. It looks like this is a Unity horn, not a Synergy horn. It does not use the “tapped horn” technology to boost the bottom end. This speaker is active, digitally processed and multi-amped, which means that Danley can use EQ and other means to raise low-end output. Basically it is the 2×15 version of the Klipsch The Sevens.
Lastly, we have this:
That slot in the front is actually a kind of horn, basically a diffraction horn I think (the literature says it has a 140 degrees horizontal dispersion), and it goes down to 80hz I would guess. For example, here is a similar speaker from their prosound lineup, using the “Paraline” technology to create “actually a large horn with an effective path length of over 25′!”, that is rated to 77hz at -3db; and, Maximum Output of 127db continuous, which is the kind of thing you can only do with horns, so it must be real.
The whole idea that a little slot provides horn loading down to 80hz is freakish. That little slot basically does the same thing as the Magico Ultimate III. But, Danley has delivered one freakish thing after another over the years. He is in a class by himself.
This speaker is also “active,” which means, like the Klipsh The Sevens, you just plug in a digital source like a iPhone and you are in business. This can be irritating to “audiophiles,” because audiophiles want to play with stuff. They want to plug this thing into that thing, and experiment. No experimenting here, you get it just as Danley intended, and no changes are possible, not even a speaker cable. Getting it just as Danley intended is not a bad thing, however. Personally I also like to experiment with things, but I do it in a DIY fashion. One reason for this is, to get what I want (basically big multiamped horns), you can’t buy it in stores. You have to make it yourself.
OK, there are a few products out there that are multiamped horns, including the Danley or Klipsch speakers above. Or, the AvantGarde Trio system. But, they are mostly very, very expensive. Also, I like making it myself.
Among amplifiers, I like this:
I am using the Behringer A500, the Class AB transistor-based predecessor, and it is a heck of a nice amplifier for the money. For the newest version, the A800, they went with a Class D technology. I don’t think it is all that great, but it puts a lot of watts down for the price. Since I do a lot of multiamping, I tend to have a need for big power amplifiers that are refined enough to drive woofers up to about 300hz (that is, not very refined); and then a need for small, highly refined <10 watt amplifiers for horns. I don’t really need both refinement and power, which is always a tall order. A lot of the time I am listening literally to the “first watt” of power, or maybe first 100 milliwatts. Often, the first watt on a 400 watt amplifier is not that great. You have to use low-efficiency speakers just to get the amplifier into its “sweet spot.” Maybe I should just add some resistors to the output — har!
For example, here are specs from Hypex’s 500 Watt NCore (Class D) unit:
The “sweet spot” for measured distortion is between about 5 watts and 100 watts. But, 5 watts is ridiculously loud on a 110db horn — that would be about 117db, which would cause hearing loss in a short amount of time. At 100mW, which is still 100db (a transient musical peak like a drum strike) on a horn, the measured distortion is about 10x higher. The same 100db musical transient would require about 10 watts on the Magico M9, putting it right in the middle of the “sweet spot” of this amplifier.
Here we have distortion vs. frequency:
Here we see that most of the distortion is at higher frequencies, in the 2000-10000hz range. This is a very sensitive area for the human ear, for distortion. Distortion at 5khz is — let’s say — 10x more objectionable than distortion at 100hz. If you take a fundamental tone of 200hz (about a male voice), then the “tenth harmonic” would be 200hz X 10, or 2000hz. Often these measures of perceived objectionableness are expressed in terms of harmonic number, instead of simple frequency, but they end up in the same place — 1000 to 20000hz. These amplifiers are best below 500hz, and at higher power — basically, driving woofers. Measured distortion isn’t everything. There are amplifiers which measure very well, but sound terrible (very expensive gear from Halcro was notorious for this). But, it gives you an idea.
Feedback (and Class D amplifiers have breathtaking amounts of feedback) tends to cause “harmonic” distortion to translate into “non-harmonic distortion,” which you can think of as white noise, or “static”. This white noise rises and falls with the signal, so you don’t hear it when silent. When there is a lot of signal, there is a lot of white noise. This might be expected to produce a “bleached out” effect on the sound, which is in fact what we tend to hear with amplifiers with feedback, either tube or transistor. This is just hypothesizing, but it is one example where measurements of “harmonic” distortion don’t match what you hear.
Among things I’m looking at, but haven’t tried yet, in the big power category is the newest offering from Hypex, the Nilai. The current NCore was a big improvement from the prior (but still good) UcD series. The Nilai, Hypex claims, achieves 10x less distortion even than the NCore. I haven’t found that these kinds of Class D amplifiers achieve the highest class, but they can be very good for the money, and provide big power. Maybe the Nilai does deserve to be in the highest class.
Also, there’s this GANFet-based “Power DAC” amplifier. Basically the amplifier itself is a digital-analog conversion device, of the PWM variety (like a DSD DAC). There is no “analog” amplifier. You just feed it a digital signal. The response it has gotten seems similar — it provides a heck of a lot of refinement and power for the money, but it is not quite in the top class.
Cables do make a difference. But, that doesn’t mean you have to spend big money on them. I would start with “good wire.” “Good wire” means something perhaps from Mogami or Canare, which are popular sources for wire for professional recording studios. These cables use wire from Mogami, and good connectors from Neutrik or Amphenol, and don’t cost much.
Yes, another, more expensive cable will sound different. Better? Maybe not.
In the bigger-dollar category, there’s Analysis Plus, which would be my first stop if I had ambitions beyond Mogami. Their Oval One RCA interconnects, at $120/pair in 1 meter length, have the basic technology.
I am mentioning this because I think you can read the magazines and the online forums all day long and still never know about any of this stuff — which, to me, is the most interesting stuff, and the stuff that I personally want to play with.
As I always say: Have fun with it.
Here’s a guy who had a lot of fun with it. He’s a retired policeman.
