Bob Cordell's Power amplifier book

Some loudspeakers are more sensitive to variations in source impedance (damping factor) than others. Any speaker using passive 4th-order crossover filters, or the acoustical version, which is a vented box, is going to be more sensitive to source impedance than a speaker using low-Q alignments.

For the raw driver, damping factor doesn't matter that much, since voice coil resistance (Re) is in series with the inductors in the crossover, speaker wiring resistance, and the source impedance of the power amplifier. With a typical voice-coil DCR of 6 ohms, inductor resistance anywhere from 0.1 to 0.5 ohms, and speaker cable resistance anywhere from 0.05 (very unusual) to 0.3 ohms (for long runs), an amplifier source resistance much lower than 0.2 ohms (which is a damping factor of 80) is basically wasted. From the perspective of the driver, it is the ratio between DCR and all the source impedances that matter - and once this ratio is much beyond 40 or so, there's not much difference to the response or distortion of the driver.

4th-order crossovers (electrical) or 4th-order (vented) cabinets are a little more touchy, simply because they are 4th-order filters, and are more sensitive to small parameter variations. Nearly all the filter damping is coming from the power amplifier, not resistive terms in the filter. Unfortunately, in loudspeaker drivers, the Q is not that well controlled.

The cone mass is very predictable; 1% repeatability is common and expected. Likewise, the number of turns in the gap is well-controlled, since this is wound on a machine with a turns counter. Compliance is not well-controlled, with a 20~30% shift during the break-in process. But compliance shifts fortunately have little effect on cabinet alignment or crossover behavior.

It's the BL product that is troublesome. BL product are the turns multiplied by the Teslas of field strength in the gap; L stands for length of wire, which is well controlled, but the effective B field that the VC sees is quite another matter. Overhung voice coils, by far the most common, have a significant portion outside the gap, and the magnetic field lines are not well-controlled in the region (in the gap the field lines are parallel and reasonably equally spaced). Move the voice coil a bit (excursion) and the different portions of the VC are exposed to the nonlinear field structure above and below the gap (which are asymmetric). Various attempts have been made to linearize the out-of-gap field, but even the complex systems are far less linear than what's in the gap - and overhung VC always have mixture of turns exposed to the fields inside and outside the gap. Not only that, the VC is continually moving, so low frequency excursions intermodulate higher frequencies.

The old-school Altec and Lowther approach is an underhung VC, which later used by TAD for their Alnico-magnet woofers. This is inherently far more linear, basically the loudspeaker equivalent to Class A operation. But once the VC starts to leave the gap to get that boom-boom bass everyone loves, then linearity goes downhill quickly. Overhung VC speakers are cheaper to build, which is another reason they dominate 99% or more of the market. Guitar speakers are equal-hung, which gives the most efficiency, but also the most distortion, since any movement at all makes the VC leave the gap.

So unfortunately BL, and therefore Qts, is not a well-controlled, linear parameter. It is not an accident that the specified Qts is what driver manufacturers call a "small-signal" parameter, which is a cute way of saying it is level-dependent. What you get at 40 dB is not the same as what you get at 90 dB. This is inherent to the technology, sorry.

This variation in BL and Q under dynamic conditions has implications for cabinet and crossover design. Since BL and Q are not well-controlled, it calls into question the wisdom of high-order alignments that rely on precise control of the filter elements. Driver mass, cabinet volume, and the passive components in a crossover are well-defined, but cone compliance, BL, and Qts are not, and shift under dynamic conditions in a complex way, with significant stored energy in the spring-return of the spider and high-Q resonances and standing waves of the diaphragm and interior cabinet volume.

There are a lot of high-end speakers these days that have high-order crossovers in vented boxes; they will be much more sensitive to source Z of the amplifier than speakers that use closed-boxes, or resistive vents, and also have low-Q crossovers (2nd-order or lower). But relying on the amplifier to magically "straighten out" what's happening in the VC gap isn't going to happen; the entire amplifier-speaker interface rests on the field-line structure in the gap, and how linearly it moves the voice coil. If the field lines are curved and distorted, then the amp-speaker interface has to follow along, and behave the same way. The entire amplifier is always seen through the prism of the field lines that drive the voice coil. This is true of all magnetically driven loudspeakers, regardless whether they are planar, ribbon, or conventional.

This is where the great Class AB debate about bias stability under dynamic conditions has relevance for loudspeakers. A transistor amplifier is making its transition out of Class A (around a watt or so) at about the same level that a loudspeaker is leaving what is considered the "small-signal" region - in other words, anywhere from 85 to 92 dB SPL at one meter, depending on typical efficiencies. It also tells us that high-efficiency speakers in the 96+ dB region may be a lot less forgiving of dynamic bias shift than less-efficient speakers.
 
Last edited:
  • Like
Reactions: 1 users
Some loudspeakers are more sensitive to variations in source impedance (damping factor) than others. Any speaker using passive 4th-order crossover filters, or the acoustical version, which is a vented box, is going to be more sensitive to source impedance than a speaker using low-Q alignments.

For the raw driver, damping factor doesn't matter that much, since voice coil resistance (Re) is in series with the inductors in the crossover, speaker wiring resistance, and the source impedance of the power amplifier. With a typical voice-coil DCR of 6 ohms, inductor resistance anywhere from 0.1 to 0.5 ohms, and speaker cable resistance anywhere from 0.05 (very unusual) to 0.3 ohms (for long runs), an amplifier source resistance much lower than 0.2 ohms (which is a damping factor of 80) is basically wasted. From the perspective of the driver, it is the ratio between DCR and all the source impedances that matter - and once this ratio is much beyond 40 or so, there's not much difference to the response or distortion of the driver.

4th-order crossovers (electrical) or 4th-order (vented) cabinets are a little more touchy, simply because they are 4th-order filters, and are more sensitive to small parameter variations. Nearly all the filter damping is coming from the power amplifier, not resistive terms in the filter. Unfortunately, in loudspeaker drivers, the Q is not that well controlled.

The cone mass is very predictable; 1% repeatability is common and expected. Likewise, the number of turns in the gap is well-controlled, since this is wound on a machine with a turns counter. Compliance is not well-controlled, with a 20~30% shift during the break-in process. But compliance shifts fortunately have little effect on cabinet alignment or crossover behavior.

It's the BL product that is troublesome. BL product are the turns multiplied by the Teslas of field strength in the gap; L stands for length of wire, which is well controlled, but the effective B field that the VC sees is quite another matter. Overhung voice coils, by far the most common, have a significant portion outside the gap, and the magnetic field lines are not well-controlled in the region (in the gap the field lines are parallel and reasonably equally spaced). Move the voice coil a bit (excursion) and the different portions of the VC are exposed to the nonlinear field structure above and below the gap (which are asymmetric). Various attempts have been made to linearize the out-of-gap field, but even the complex systems are far less linear than what's in the gap - and overhung VC always have mixture of turns exposed to the fields inside and outside the gap. Not only that, the VC is continually moving, so low frequency excursions intermodulate higher frequencies.

The old-school Altec and Lowther approach is an underhung VC, which later used by TAD for their Alnico-magnet woofers. This is inherently far more linear, basically the loudspeaker equivalent to Class A operation. But once the VC starts to leave the gap to get that boom-boom bass everyone loves, then linearity goes downhill quickly. Overhung VC speakers are cheaper to build, which is another reason they dominate 99% or more of the market. Guitar speakers are equal-hung, which gives the most efficiency, but also the most distortion, since any movement at all makes the VC leave the gap.

So unfortunately BL, and therefore Qts, is not a well-controlled, linear parameter. It is not an accident that the specified Qts is what driver manufacturers call a "small-signal" parameter, which is a cute way of saying it is level-dependent. What you get at 40 dB is not the same as what you get at 90 dB. This is inherent to the technology, sorry.

This variation in BL and Q under dynamic conditions has implications for cabinet and crossover design. Since BL and Q are not well-controlled, it calls into question the wisdom of high-order alignments that rely on precise control of the filter elements. Driver mass, cabinet volume, and the passive components in a crossover are well-defined, but cone compliance, BL, and Qts are not, and shift under dynamic conditions in a complex way, with significant stored energy in the spring-return of the spider and high-Q resonances and standing waves of the diaphragm and interior cabinet volume.

There are a lot of high-end speakers these days that have high-order crossovers in vented boxes; they will be much more sensitive to source Z of the amplifier than speakers that use closed-boxes, or resistive vents, and also have low-Q crossovers (2nd-order or lower). But relying on the amplifier to magically "straighten out" what's happening in the VC gap isn't going to happen; the entire amplifier-speaker interface rests on the field-line structure in the gap, and how linearly it moves the voice coil. If the field lines are curved and distorted, then the amp-speaker interface has to follow along, and behave the same way. The entire amplifier is always seen through the prism of the field lines that drive the voice coil. This is true of all magnetically driven loudspeakers, regardless whether they are planar, ribbon, or conventional.

This is where the great Class AB debate about bias stability under dynamic conditions has relevance for loudspeakers. A transistor amplifier is making its transition out of Class A (around a watt or so) at about the same level that a loudspeaker is leaving what is considered the "small-signal" region - in other words, anywhere from 85 to 92 dB SPL at one meter, depending on typical efficiencies. It also tells us that high-efficiency speakers in the 96+ dB region may be a lot less forgiving of dynamic bias shift than less-efficient speakers.

Wow! Great post, Lynn. This is one of the best, most succinct explanations of loudspeaker driver distortion I've seen. Your observations about concerns that may be related to high-order alignments and high-order crossovers are very interesting.

A very good read.

Thanks, Lynn.

Bob
 
Sorry for yanking the thread off-topic from Bob's wonderful book, so here's a question based on the book.

On pages 544 - 550, the Linear Technology LT1166 current-sensing bias controller is discussed at some length. It's an intriguing device that offers a fast response to dynamic bias shifts, and a good candidate for MOSFETs or BJT Triples. Has anyone - including Bob Cordell - tried the LT1166 with a high-speed MOSFET amplifier like the circuit on page 242 (fig. 11.7), or the wide open-loop-bandwidth MOSFET amplifier of page 525 (fig. 25.15).

On paper, at least, this would seem to combine the best of several worlds - near-complete removal of dynamic bias drift, good temperature control with no bias-setting pots, no nasty V-I limiters, as well as high slew rate, rapid settling time, and low 20 kHz distortion.

Instantaneous changes in output impedance - or what is depicted in the "wingspread" graphs in the book - is something I'm concerned about with many amplifiers. It was something I mentioned in passing in my 2004 European Triode Festival Presentation - sorry about the nasty dig at transistor amplifiers, but it was a triode crowd! Having read your book, I overstated the problem with Class AB transistor amplifiers - but I still see resonant back-EMFs arriving at shifting-impedance output sections as a speaker-interface problem that does not appear on any conventional measurement.

Any technique to smooth out the "gm-doubling" region would be an improvement - LT1166, the ThermalTrak transistors, or even the extreme solution of Class A operation. (Which is why I designed a Class A PP triode amplifier when SET's were all the fashion - the output characteristics were much closer to a true resistor than anything else.)

I see control of the "gm-doubling region" as a big deal, particularly if there are bias-drift issues. If many, if not most, Class AB BJT amplifiers have the "wrong" bias with real music and real loudspeakers, that's a major problem that needs to be addressed.
 
Last edited:
Administrator
Joined 2004
Paid Member
Hi Bob,
And yet the audio world is often proud of such speakers and hold them in high esteem. Remember the Infinity Kappa series as an example? Dumb.

Why the devil do some educated engineers even consider creating a speaker system that presents a difficult load to drive? It's not a stretch to realize that an unhappy amplifier will not perform at it's best. Therefore the loudspeaker system stands a poor chance of performing to it's possible potential.

I can only think of one comment to apply here that isn't coarse.

"Unclear on the concept" :)

-Chris
 
Administrator
Joined 2004
Paid Member
Hi Lynn,
I wasn't impressed with the LT1166. Could be me, but I did use it as John Curl had suggested as it seemed to "want" to be used as a Vbe multiplier. This seemed obvious at the time.

As for solid state equipment, I use mostly solid state, but I also use tube equipment. To be honest with you, a Marantz 300DC that I modified satisfies all my friends who use tube equipment. You can't level the charge that all solid state products are bad, just as you can't say that all tube gear sounds the best. I have been very fortunate to have exposure to excellent examples of each technology. I enjoy both, but neither is a perfect solution.

Another thing you have to accept is that no matter what brand of equipment you own, it will not perform at it's best unless it's torn down and rebuilt with some care and attention to details. You can often see huge gains in sound quality (that you can also measure by the way ;) ). I know there are some members here who know precisely how to improve most any amplifier, although the approaches are as different as the amp in question. Also, some designs are simply unhappy no matter how they look on paper. I'll blame the circuit layout for those cases.

-Chris
 
I'm chipping away some of my anti-transistor prejudices, partly because a 5-channel system with all-triode Class A PP amps wasn't going to be happening, and partly because Bob's book is a superb overview of the transistor world. Still lean heavily towards high-speed (70V/uSec or faster) amplifiers, though.

Part of what drew me into this was auditioning a lot of HT receivers, as well as Anthem, NAD, Rotel and similar high-end HT products. Urgh! I had no idea these things sounded so bad. Eventually ended up buying Marantz AV8003 & MM8003 combo, but this is still well short of what transistors can do. Now I want to know what's going on in these gizmos, and why they sound the way they do.
 
Last edited:
This is where the great Class AB debate about bias stability under dynamic conditions has relevance for loudspeakers. A transistor amplifier is making its transition out of Class A (around a watt or so) at about the same level that a loudspeaker is leaving what is considered the "small-signal" region - in other words, anywhere from 85 to 92 dB SPL at one meter, depending on typical efficiencies. It also tells us that high-efficiency speakers in the 96+ dB region may be a lot less forgiving of dynamic bias shift than less-efficient speakers.

Sorry, I'm missing something in your last sentence. It seems counter intuitive to me. Why does it tell us that HE speakers are less forgiving?
 
Next time I encounter one of those witless "all amps sound the same" people I'm going to challenge them to read this book, cover to cover, and see if they still feel that way after reading it. For more open-minder readers, particularly if they've never designed an amplifier before, this book will be a revelation. This book belongs on the "must-read" list for every serious audiophile or DIYer. If I had the power, I would force every reviewer to read it, and write about the product sitting in front of them, instead of a rehash of marketing talking-points.

Witless???

You need to attend one of Bob's workshops where he demonstrates an A/B switch between a tube and solid state amplifier - 3rd picture from the bottom:
Stereophile: HE 2007 Show Report

They sound very similar even though they had widely different power ratings, and were driven hard. I correctly identified the tube amp numerous times, but it was difficult, and many there got it backwards. I've done this sort of level matched test myself and most good SS amps sound the same when not driven into clipping.

Witless, is not a nice word, you should know that, LOL!

Bob, do you recall if many people correctly identified tube vs. SS in your A/B test?
 
Sorry for yanking the thread off-topic from Bob's wonderful book, so here's a question based on the book.

On pages 544 - 550, the Linear Technology LT1166 current-sensing bias controller is discussed at some length. It's an intriguing device that offers a fast response to dynamic bias shifts, and a good candidate for MOSFETs or BJT Triples. Has anyone - including Bob Cordell - tried the LT1166 with a high-speed MOSFET amplifier like the circuit on page 242 (fig. 11.7), or the wide open-loop-bandwidth MOSFET amplifier of page 525 (fig. 25.15).

On paper, at least, this would seem to combine the best of several worlds - near-complete removal of dynamic bias drift, good temperature control with no bias-setting pots, no nasty V-I limiters, as well as high slew rate, rapid settling time, and low 20 kHz distortion.

Instantaneous changes in output impedance - or what is depicted in the "wingspread" graphs in the book - is something I'm concerned about with many amplifiers. It was something I mentioned in passing in my 2004 European Triode Festival Presentation - sorry about the nasty dig at transistor amplifiers, but it was a triode crowd! Having read your book, I overstated the problem with Class AB transistor amplifiers - but I still see resonant back-EMFs arriving at shifting-impedance output sections as a speaker-interface problem that does not appear on any conventional measurement.

Any technique to smooth out the "gm-doubling" region would be an improvement - LT1166, the ThermalTrak transistors, or even the extreme solution of Class A operation. (Which is why I designed a Class A PP triode amplifier when SET's were all the fashion - the output characteristics were much closer to a true resistor than anything else.)

I see control of the "gm-doubling region" as a big deal, particularly if there are bias-drift issues. If many, if not most, Class AB BJT amplifiers have the "wrong" bias with real music and real loudspeakers, that's a major problem that needs to be addressed.

Lots of good points here, Lynn. About four years ago I built a MOSFET power amplifier using the LT1166. I was fascinated by the potential of the part, yet disappointed in the app notes that came with it. I concluded that it needed to be used as a shunt bias spreader only, essentially as a replacement for the Vbe multiplier. I also learned that the application of Miller compensation has to be a bit different to avoid increased distortion from the dynamic bias spreading action.

However, it was a struggle to stabilize it, and I had discussions with the LT people and got a SPICE file for it that I literally used to draw a schematic of it to see what was going on. Once I got it stabilized, it worked quite well. There were a lot of other design variations I was trying in the amplifier at the same time (poor engineering approach), so I never did come to a conclusion about whether the dynamic way in which the LT1166 controls the bias might also reduce crossover distortion, but I have long suspected it. Just too many things to do.

You are absolutely right about the need for correct output stage biasing and the need to minimize dyamic thermal bias mis-tracking. This is an area where many BJT amplifiers have suffered, at least until the use of ThermalTrak transistors reduced the problem. BTW, although MOSFET output stages usually start out with more static distortion than BJTs, it is virtually impossible for gm doubling to occur in a MOSFET output stage, and they like to run hot, yielding a larger class A region. The crossover transition is also far less abrupt that with BJTs.

Cheers,
Bob
 
Hi Bob,
And yet the audio world is often proud of such speakers and hold them in high esteem. Remember the Infinity Kappa series as an example? Dumb.

Why the devil do some educated engineers even consider creating a speaker system that presents a difficult load to drive? It's not a stretch to realize that an unhappy amplifier will not perform at it's best. Therefore the loudspeaker system stands a poor chance of performing to it's possible potential.

I can only think of one comment to apply here that isn't coarse.

"Unclear on the concept" :)

-Chris

Hi Chris,

I guess the loudspeaker designers take a "not my problem" view when it comes to what it takes to drive the loudspeaker. Most design the speakers as if they will be driven by a true voltage source. Not worrying about the gyrations of the load the loudspeaker presents to the amplifier is one less constraint that they have to meet. It is certainly true that the loudspeaker designers have enough to worry about in the first place.

However, I don't think they should be totally let off the hook. Making the best sounding speaker in a given box size usually means some sacrifice in efficiency, both in terms of low-frequency extension and flatness, and in terms of distortion (as Lynn pointed out). So many speaker designs take the easy way out and let the impedance dip to 2 ohms or even less - sometimes even if the loudspeaker is rated at a nominal 8 ohms. They'll sound good (and will produce reasonable SPL) with big amps that can produce a lot of current, but maybe not so good with ordinary amps.

Cheers,
Bob
 
Here I go again

Hi Chris,

I guess the loudspeaker designers take a "not my problem" view when it comes to what it takes to drive the loudspeaker. Most design the speakers as if they will be driven by a true voltage source. Not worrying about the gyrations of the load the loudspeaker presents to the amplifier is one less constraint that they have to meet. It is certainly true that the loudspeaker designers have enough to worry about in the first place.
.................................

Cheers,
Bob

Hi Bod,

That's precisely my issue with those so called 'high-end' speaker manufacturers. The development of a good speaker doesn't stop when the frequency response is flat when driven by a pure voltage source. A really well designed loudspeaker has not only has a flat frequency response (and low distortion, of course), but also a flat impedance curve throughout the audio spectrum AND above. IOW, a flat response WRT to input power, not just only voltage. Speakers that dip below 1 Ohm or so, should be banned, forbidden, boycotted, burned, etc. Such misbehavior lays an unnecessary burden on amplifiers and amplifier designers. It's time to revolt against this kind of bad practice.

Cheers,
E.
 
Witless???

You need to attend one of Bob's workshops where he demonstrates an A/B switch between a tube and solid state amplifier - 3rd picture from the bottom:
Stereophile: HE 2007 Show Report

They sound very similar even though they had widely different power ratings, and were driven hard. I correctly identified the tube amp numerous times, but it was difficult, and many there got it backwards. I've done this sort of level matched test myself and most good SS amps sound the same when not driven into clipping.

Witless, is not a nice word, you should know that, LOL!

Bob, do you recall if many people correctly identified tube vs. SS in your A/B test?

As applied to the "all amps sound the same" people, witless is a kind word. It implies that they are not thinking about how indefensible their absolute statement is. It is very likely that they are saying it to bait and insult amp designers; in which case, a much darker adjective would apply.

If one removes the absolute part and simply says "high quality SS amps sound the same". I don't think many would take offense and a productive discussion might follow. Personally, I hope that as amps develop they will sound alike; it would demonstrate that the amps were doing their function correctly.

However to disprove the statement "all amps sound the same"; I just need to find one amp in the whole world that sounds worse than the very best amp.
 
Witless???

You need to attend one of Bob's workshops where he demonstrates an A/B switch between a tube and solid state amplifier - 3rd picture from the bottom:
Stereophile: HE 2007 Show Report

They sound very similar even though they had widely different power ratings, and were driven hard. I correctly identified the tube amp numerous times, but it was difficult, and many there got it backwards. I've done this sort of level matched test myself and most good SS amps sound the same when not driven into clipping.

Witless, is not a nice word, you should know that, LOL!

Bob, do you recall if many people correctly identified tube vs. SS in your A/B test?

Hi Pete,

That A/B test was one of those tests where many people were surprized in several ways, including myself. The tube amp in question was one that I had designed myself solely for the purpose of hearing the difference between a tube amplifier and a SS amplifier where the listening conidtions were well-controlled AND I was familiar and trusting of both designs. Up to that point, I had not designed a tube amplifier since I was in high school. So I designed the tube amp with more modern knowledge and some updated techniques. For example the tube amp used transistor current sources for the tube differential pairs and screen voltage supplies that were regulated with MOSFET pass transistors. Some may call this cheating :). The tube amp is briefly described on my web site (CordellAudio.com - Home).

Anyway, I was surprized at how good the 35 wpc tube amp sounded.

I subsequently took the amp to a tube-head meeting where it was compared with other tube amps in the usual uncontrolled way. It was well-liked, so I guess it had at least some legitimate tube pedigree despite some of the modern circuit techniques. (those push-pull KT88's, loafing at 35 wpc, really stepped on the SETs).

At the HE2007 workshops, my recollection is that a decent number of people could hear a difference, but about half were wrong in identifying which was the tube and which was the solid state. Similarly, I think that about half preferred the one that was solid state and about half preferred the one that was tubed.

Cheers,
Bob
 
As applied to the "all amps sound the same" people, witless is a kind word. It implies that they are not thinking about how indefensible their absolute statement is. It is very likely that they are saying it to bait and insult amp designers; in which case, a much darker adjective would apply.

If one removes the absolute part and simply says "high quality SS amps sound the same". I don't think many would take offense and a productive discussion might follow. Personally, I hope that as amps develop they will sound alike; it would demonstrate that the amps were doing their function correctly.

However to disprove the statement "all amps sound the same"; I just need to find one amp in the whole world that sounds worse than the very best amp.

Try reading what I wrote rather than offer more offensive insults:
"most good SS amps sound the same when not driven into clipping"
I might offer my definition of "good" one of these days, but I'm not really interested in
your offensive bait.

I would also say that the absolute part is most often a misquote or deliberate
misinterpretation by those who wish to spew insults.
 
Last edited:
"most good SS amps sound the same when not driven into clipping"

I've asked this question on the 'do you agree with Doug Self?' thread - but I wonder if the readers of this thread will have a view. Given that Bob's demonstrated how easy it is to clip amps (I'm still trying to track down Rickie Lee Jones myself :)), why would we exclude clipping behaviour from comparisons between amps? In the real world of audio, amps do clip and so their clipping behaviour is most certainly going to be part of how they sound.
 
Some loudspeakers are more sensitive to variations in source impedance (damping factor) than others. Any speaker using passive 4th-order crossover filters, or the acoustical version, which is a vented box, is going to be more sensitive to source impedance than a speaker using low-Q alignments.

... snip ...

There are a lot of high-end speakers these days that have high-order crossovers in vented boxes; they will be much more sensitive to source Z of the amplifier than speakers that use closed-boxes, or resistive vents, and also have low-Q crossovers (2nd-order or lower). But relying on the amplifier to magically "straighten out" what's happening in the VC gap isn't going to happen; the entire amplifier-speaker interface rests on the field-line structure in the gap, and how linearly it moves the voice coil. If the field lines are curved and distorted, then the amp-speaker interface has to follow along, and behave the same way. The entire amplifier is always seen through the prism of the field lines that drive the voice coil. This is true of all magnetically driven loudspeakers, regardless whether they are planar, ribbon, or conventional.

This completely ignores the well known technique of conjugate load matching, however it is costly and therefore most simply do not use it. KEF has, and B&W also in selected models as I recall.

This is nothing new, it became well known in the 70s that if a speaker load is nominally 4 ohms and reactive then it should be paired with a high current, low output Z amp capable of driving roughly half the rated impedance. High current output was widely recognized as important in those days and the ability to drive 2 ohms was a common design goal. Has everyone forgotten this history?

There are also many examples of heavily damped vented designs that have been marketed. The Vandersteen 2C and Genesis II come to mind as just a few examples.
Not all 4th order alignments are necessarily high Q. I'm not suggesting that those are good designs, simply examples that counter your claim.

It is unfortunate that more is not done to provide more linear driver motors, but you sweeping generalization ignores the modern XBL and split coil designs.
 
Last edited:
I've asked this question on the 'do you agree with Doug Self?' thread - but I wonder if the readers of this thread will have a view. Given that Bob's demonstrated how easy it is to clip amps (I'm still trying to track down Rickie Lee Jones myself :)), why would we exclude clipping behaviour from comparisons between amps? In the real world of audio, amps do clip and so their clipping behaviour is most certainly going to be part of how they sound.

I have a total of over 1000W in my main rig ... you don't have to clip.
I discovered the importance of head room about 40 years ago as a kid with a scope on the output of a 60W amp.