The disappearance of the traditional amplifier

K_ilpo_p, thanks for the excellent discussion . . . let me see if I can better refine and clarify my thoughts on the matter.

Also omnidirectional source directivity is constant, and in power response terms the driver type has no inherent effect.

The parameter of total, summed power response as I see it is most useful in trying to correlate the perceived timberal balance vs. measured frequency response for NON-constant-directivity systems, and for establishing the optimum placement and room treatment for a given loudspeaker system. It's pretty much irrelevant for the issue of establishing the best directivity characteristics of the driver(s) themselves. Consider that (for a single driver) electronic equalisation is supremely effective in altering the summed power response, but completely ineffective at solving directivity issues.

Altec made their Mantaray horns, using basically two flares and an abrupt joint between them. First expansion created the vertical pattern, opening into a narrow arc-like slit - being in itself a (horizontally) wide radiator- and the second flare controlled the horizontal pattern. The JBL Bi-radial horns use the same principle. In very simple terms a waveguide could be interpreted as the outer flare of a constant directivity horn, if you so wish.

While not all modern constant-directivity compression-driver waveguides use an abrupt change in expansion rate, I like your description, and find it a useful analogy. So I'll attempt to use it to illustrate my basic point in the whole matter - which is that to substitute a pistonic driver (cone or dome) for the compression driver and throat . . . brings out fundamentally different principles of operation in the waveguide as far as the directivity is concerned. Also, the difference between these two approaches is pretty much unrelated to the traditional view of the difference between compression drivers and direct-radiating drivers - which you accurately state as being efficiency, and acoustic impedance.

I do not understand what you actually mean with saying, that "for true constant-directivity performance to be possible, the wave-front propegation has to be constant with frequency."

Fair enough . . . my use of the term "true" implies a value judgement which I did not intend.

Instead, I'll refer to a compression-driver constant-directivity waveguide system (like the big JBL butt-cheek we've been discussing) as being a "wideband constant-directivity" system. In addition to the traditional points stated above (acoustic impedance and efficiency), a compression driver strives to transform the pistonic movement of the diaphragm into a pressure wave - a wave that has a shape that is (ideally) frequency independent. Early-20th-century practice viewed these as plane waves, examples being devices such as slant-plate acoustic lenses, and the driver measurement apparatus, a "plane-wave tube". And although the plane-wave as a useful, precise mathematical model may be completely outdated (I'll again reference Dr. Geddes' work), it is my understanding that in a "wideband constant-directivity system" (my terminology), the ultimate goal is for the driver to illuminate the waveguide in a manner that is constant with frequency. The result is a device where the useable constant-directivity frequency range is limited solely by the practical size of the waveguide, the mechanical performance of the compression driver, and the compression driver/phase plug/throat meeting the goal of frequency-independent waveguide illumination.

This is in (at least conceptual) contrast to the practice of using a pistonic driver to illuminate a waveguide, because the driver/waveguide relationship isn't (and cannot be) frequency-independent. Rather, (please correct me if I'm wrong) the idea is that the waveguide should dominate the directivity at the bottom of the driver's passband, and as the frequency increases, the directivity is decreasingly defined by the waveguide, and increasingly defined by the driver . . . this occurs because a pistonic driver will ALWAYS have an increase in directivity with an increase in frequency. Thus, in order for the driver/waveguide system to have smooth, predictable directivity performance . . . it is obviously of paramount importance that the driver itself have smooth, predictable directivity performance - in exactly the same manner as it should in a non-waveguide direct-radiating system.

My general conclusion is that while a piston-driver/waveguide combination can maybe acheive "a good practical approximation of constant directivity" (your description), its ability to do this will ALWAYS be limited to a much narrower frequency range than is possible with a wideband, compression-driver constant-directivity waveguide. It's also only effective over a specific range of desired radiation angles, which thankfully correspond to reasonably useful ones for studio monitoring. In the end, the directivity characteristics of the driver itself is the tail that wags the dog, and ultimately determines the extent of effectiveness in the waveguide.

As a final note . . . you make reference to the importance of matching the directivity of the bass driver(s) to the waveguide-loaded device(s) (something I very much agree with), and the effects of the crossover slope on the transition-band directivity. I'd be interested on how you view the common (recommended?) practice of turning i.e. the Genelecs sidewise and simply rotating the waveguide, which I feel makes a mess of these issues in both theory and practice.

Active loudspeakers are IMO long overdue for penetration into the consumer marketplace - there's virtually no aspect of amplifier, loudspeaker, or crossover performance that isn't improved by using a separate amp for each frequency range with the crossover being performed before amplification.

The main reason that it isn't more common is both cultural and economic. For starters . . . loudspeaker companies and electronics companies have different resources and abilities - speaker companies are usually mainly woodshops, and electronics companies stuff circuit boards into sheetmetal enclosures. Most of the companies that build active speakers rely on an outsourced "plate-amp" module (either off-the-shelf or custom) that's easily incorporated into their conventional design/manufacturing methods, and most electronics companies that have loudspeaker lines outsource the cabinetry from a woodshop.

In an economic sense, if you're going to integrate two things together, the receiver/passive-speaker has some clear advantages over the tuner-preamp/active-speaker combination -- putting like things together affords considerable savings. (Look at a cheap mini-system with biamplified speakers - the amps are always in the main unit, with two sets of speaker wires.) For custom home installation, it's also much cheaper to run low-voltage speaker wire everywhere, than to use a high-quality balanced line-level distribution system, plus AC power at every speaker location.

And there's also the cultural difference in the distribution side - it takes much more thought and effort for a salesperson to convince somebody to replace their amplifier(s) (that they may be attached to) when they're looking at making a speaker purchase. And in the high end, there are many symbiotic relationships between amp and speaker companies, that share resources at shows, and serve the same dealer network -- an active speaker product can upset these relationships.

But although I feel that the active speaker approach is in general a better way of doing things, I still think that there are a LOT of really poor products in the "professional" ranks, even some very high-priced ones . . . i.e. I feel passive ATC SCM20 or SCM50 (or even the old JBL 4435 running passive) absolutely smoke the now-ubiquitous active Genelecs. And don't get me started about the cheap "active monitors" (really overgrown computer speakers) that mail-order music stores ship by the truckload . . .

OTOH, field coils are making a huge comeback right now. Active speakers are not likely to see innovations like that since they are closed systems.

Historically, the vast majority of field coil speakers were indeed used in "closed" systems . . . such as radios, instrument amplifiers, organs and Leslie cabinets, etc. etc. This had the added advantage of the amplifier being able to use the field coil as a power-supply choke - effectively working to reduce the cost of two parts (choke and speaker magnet).

The main push for the adoption of permanent-magnet speakers came after World War II, with the demand for larger separate speakers for movie theaters and music production - many the very earliest examples of these at first had separate field-coil power supplies. The introduction of Alnico V as a magnetic material (itself developed during WWII) was the main reason that field-coil speakers were abandoned.

But as always . . . audiophiles have very fickle preferances, and while some may find it interesting and comforting to experiment continually with the amp/loudspeaker relationship, it seems to me that a significant number of posts here on the Audiogon forums is by people who are asking advice on trying to get this right -- maybe some of them would enjoy a product where this was already done for them as part of the product engineering.

I don't think that Genelec makes a bad speaker . . . it's just that as far as I can tell, the way they (and their knock-offs) implement the waveguide around a direct-radiating driver does not make really make it a constant-directivity system. Rather, it seems to simply to reduce the effects of the cabinet edge diffraction on the directivity. These are my rather informal observations based on hearing them in a decent handfull of studio control rooms, and measuring their response in two.

And while their idiosyncracies aren't really all that different from most direct-radiating studio monitors, Genelec specifically touts these features as making their monitors less sensitive to control-room acoustics and speaker placement, which I don't think holds up in practice. They also freely recommend most of their two-way nearfields and three-way mid-fields for horizontal configurations, which severely compromises the performance of virtually all speakers of this type. The result is that it's quite common to see Genelecs in a studio that give a very poor rendition of what the mix sounds like anywhere else.

I will concede that I do have some fairly strong opinions about both the environment and methodology of studio recording, and I feel that a true constant-directivity monitor (like the old JBL 4435) soffit-mounted in a competently-designed control room is the most neutral, consistent representation of what's actually on the master tape. Having a pair of good nearfields (NOT NS-10s) is a nice second perspective.

For a small home studio, relying solely on nearfields is frequently the only option, and the Genelecs aren't a bad choice . . . though I would personally prefer a pair of Meyers or ATCs. The main advice I would offer is to orient the monitors vertically, and keep your monitoring SPL as low as you're comfortable with. Also, pay close attention to your impressions when you take your mix to other systems, and adjust the monitor placement to get consistency between what you observe both inside and outside your studio.

I wasn't alive back then, but I think that field-coil speakers in the first part of the 20th century were actually significantly cheaper than permanent-magnet types of similar performance . . . they were NOT a "high end" design. This was an era when the labor and expertise for winding coils was cheap and plentiful -- even budget radios were chock full of inductors, chokes, and RF transformers that were typically wound in-house. On the other hand, high-permeability magnetic materials were VERY expensive or even unavailable - the infrastructure for securing the raw materials (especially cobalt) and making high-quality alnico alloys . . . this is all very high-capital-investment stuff.

The early Lansing alnico designs specified a flux in the magnetic gap of something like 13,000 gauss . . . I'm skeptical that any field-coil design of that era could even produce half of that. And as I understand it the move to ferrites in the 1970s was a reaction to geo-political events - the main deposits of cobalt in the world fell under the control of regimes sympathetic to the Soviets, and the price of alnico alloys shot through the roof in a very short time.

I am also quite skeptical of the claims for the superiority of ferrites vs. neo vs. alnico vs. field-coil arrangements . . . but all of these methods of making the magnetic flux allow very different approaches to the design of the motor structure itself, and this does have a huge, fundamental impact on the driver characteristics. The first generation of JBL professional drivers that used ferrites were very carefully designed to have the same magnetic characteristics as their alnico predecessors, and I've mixed and matched them in sound-reinforcement systems and couldn't tell a bit of difference (except for the odd alnico driver that's lost some of its flux).

But I will agree that the Alnico magnetic structures are so much more elegant in an engineering sense . . . with no stray field, and much easier on one's back when moving them around. And I can see some similar appeal to a modern field-coil speaker.

I haven't heard the Audiokinesis loudspeakers, but I have spent a little time with Dr. Geddes' "Summa" loudspeaker, on which (as I understand it) the Audiokinesis designs are based. The Summa's horn/waveguide is a true constant-directivity design, and this is immediately evident in their excellent imaging, and very consistent tonal balance. I actually feel that Dr. Geddes' research in this area is some of the most interesting, competent, and relevant work in loudspeaker design in recent years.

But the "waveguide" designs in the Genelec monitors bear very little resemblence to a true constant-directivity spherical or bi-radial "waveguide" horn, in both the theory and the way they behave. This is mainly because they don't use compression drivers - and the waveguides are so short that the directivity characteristics of the driver itself dominate the polar response of the loudspeaker. Genelec's "waveguides" do seem to clean up the directivity performance of their drivers at the more extreme realms of their off-axis response, but they are NOT constant-directivity.

I'm really not trying to slam Genelec in general, and when used in the vertical configuration, their directivity characteristics similar to many well-behaved direct-radiating monitors. I just feel that some of their recommended setup configurations give lackluster performance, and their marketing material seems to imply that they are truly constant-directivity (even though they don't actually make that claim).

Shadorne, great links! Thanks for the interesting reading.

One source of confusion is that there's a lot of ambiguity between in the terms "horn" and "waveguide" (add "lens" and it gets worse) - I tend to use them rather imprecisely as well. But to differ with AeroNET article (if I was to attempt to be precise), I think that the difference lies not in the efficiency or the type of driver used, but rather in the theoritical basis for its shape. A "horn" is usually based (at least loosely) on Webster's horn equations, which were derived in the early 1920s mainly to calculate the load the horn presents to the driver, for the purpose of maximizing efficiency -- this is the origin of the classic exponential shapes. However, there's very little theoritical basis here for understanding the horn's directivity characteristics, which is why horns for the first half of the 20th century used other techniques (multi-cellular construction, or a slant-plate lens) to control directivity without a good understanding of how the contour itself affects this.

A "waveguide" on the other hand is designed with mathematics that are derived from other fields, using techniques designed to accurately predict the directivity based on the waveguide's contours. The specifics of these maths are way over my head, but I think it's accurate to say that waveguide theory isn't limited to lower rates of expansion or lower acoustic gains.

The root of my skepticism with the direct-radiator/short-waveguide configuration (for which I've made the Genelec monitors the poster-boy) is that with my (admittedly VERY rudimentary and imprecise) understanding of both waveguide techniques and Webster's equations . . . all of it assumes a specific wave-front propegation for the horn/waveguide to work as intended. So for true constant-directivity performance to be possible, the wave-front propegation has to be constant with frequency . . . and the conventional direct-radiating cones and domes used in such configurations do NOT acheive this. Rather, they exhibit the classic increase in directivity with increase in frequency, just like all domes and cones.

The author of the AEROnet article does make mention of this, but then goes on to say that for wide-angle waveguides with direct-radiating drivers "that most of the mathematical detail can be side-stepped." Huh???? If you side-step the mathematical detail, then what you have is simply a random curvy recessed cabinet-front, and NOT a waveguide. He then goes on to make some measurements of some purely emperically-derived combinations . . . and while the final results look nice and smooth, this seems to be the obvious result of simply changing the way the waveguide is illuminated, thereby effectively altering its curve in a theoritical sense.

The K&H monitor does indeed have very smooth directivity plots, but the directivity still increases quite steadily with frequency . . . as one expects with direct-radiating domes. I would say that the smoothness of these plots (compared to i.e. a Genelec 3-way) is a testament to the excellent performance of the drivers themselves, and well-implemented crossover design . . . and the cabinet contours help out at the extremes. But they're also endorsing my other pet-peeve -- horizontal placement, for which they give no directivity plots. Suffice it to say it will be worse, and its horizontal polar response will then exhibit some of the inconsistencies found in the vertical directivity plots. Maybe they can clean this up a bit in the DSP, by changing some of the crossover slope characteristics . . .

I have an old RCA FC speaker from the 30s that is purported to have about 18,000 gauss once energized. The magnet structure on the thing is immense- so large that it is used for mounting the speaker, not the basket.

Wow, holy crap! I have never seen a field-coil magnet structure that size. If you ever have pictures . . . it's be cool to see.

Shadorne, these are two great points, and I agree that they are very significant potential benefits of the "short open waveguide" approach. But they're not constant-directivity (which was my main point), and since as it does indeed very much depend on what "driver" you have to begin with . . . these behave fundamentally very much like a standard direct-radiating driver.

But as far as the cone vs. a dome to "maintain uniform spherical wavefronts" that's the whole problem, neither of them deliver any kind of wavefront that's consistent with frequency. Cones, domes, inverted domes, ring-radiators . . . they can all exhibit profound differences in their application and execution, but they are all of a similar ilk in their inability to deliver a consistent wavefront independent of frequency. The compression driver differs in the fact that it (at least aims to) acheive this goal.

I enjoyed Mr. White's article to which you kindly provided the link, but the main problem is . . .

The theory behind the waveguides to be described is that a dome driver produces what is fair approximation of a spherical wave over its piston range

I simply can't conceive of this as being valid . . . I wish my knowledge of physics and my mathematical skill was sufficient to expound on this further, but I think it reasonable to say that it would be hard to build a consenus on this among those who do have competencies in these areas. Further, his calculations are based on the idea that the dome behaves as a point source . . . which is certainly impossible except perhaps for an extremely narrow range of frequencies.

After all, if a dome behaved as a point source, then simply screwing it into a baffle of appropriate size would produce absolutely perfect directivity characteristics, and we wouldn't need waveguides at all.

Thanks to all for the interesting discussion on waveguides in active monitors; it's given me much to think about and listen for.

However, I see no practical problem in rotating the waveguide in 3-way Genelecs as the MF/HF section still remains vertical. Because the LF/MF crossover is somewhere around 400 Hz (i.e. pretty low) there will be no practical difference whether vertical or horizontal, but of course the room reflection pattern will change due to different height of the woofer. The woofer directivity remains as it is and so does the MF, the only changing parameter is their relative position.

I find myself taking over the producer's role on a small-label classical production, and was thinking about these comments a couple days ago during a tracking session . . . the mains in this studio are horizontal soffit-mounted Genelec 1038s. I still feel that they're problematic in this configuration, and much of what I don't like is in this lower-midrange area that's likely related to the low/mid crossover region. It's bad enough where for the mixing sessions we'll either have to come up with a near-field arrangement that I'm happy with, or move to another facility. Either way, I was relieved that I was only making performance decisions through those 1038s . . .

Vertical mounting of course doesn't eliminate the crossover-related off-axis lobing, but it places it entirely on the vertical axis . . . and the vertical-axis listening position in the control room varies far less than the horizontal. Perhaps the use of brick-wall filters would reduce this, but I have no experience with monitors that use them.

Atmasphere, thanks for the recommendation - but for a number of reasons I'll probably be limited the BBS options (beg, borrow, or steal) . . . at least the nearfields in this control room are stand-mounted so they're easily swappable and moveable. Console meter bridge is also nice and low.

Shadorne, you make the excellent point that active designs can allow much more flexibility in the crossover design as well as more idealized response. But I think that the vast majority of both active and passive designs use fourth-order Lindquitz-Reilly alignments - and with analog filters this is usually an excellent overall choice. I believe that the 1038 also uses a fourth-order slope, and rotating the waveguide assembly eliminates the horizontal-axis lobing for the mid-high transition only . . . the low-mid transition is that of a horizontally-placed design.

Yeah, your logic makes complete sense . . . and there are of course innumerable differences in the environments and installations - including those where I've been able to compare vertical vs. horizontal installations of this type.

But 400Hz does lie in a region below where most types of room treatment are effective, and yet above the region where a "control-room-sized" (whatever that is) room is exhibiting primarily modal behavior. In larger spaces (medium-to-large-venue sound reinforcement) it's pretty much a given that well-controlled directivity in the 400Hz region is very important, but in smaller rooms . . . there are many opinions.

For some reason I also have this association with other horizontally-configured monitors, namely Westlakes. But it seems like all the people I've known who have Westlakes also monitor at ridiculously high levels, and maybe that's why I have a bias against monitors that look anything like them . . .