I've been in private communication of Alexander of RAAL ribbon tweeters, and he's auditioned the new double-ribbon with different sets of widerange drivers. His favorite midbass driver, the Celestion G12M "Greenback" (although good enough to match with the SA stretched-film ribbon or Electro-Voice T350 tweeters) isn't good enough to match with the new RAAL ribbon, with a noticeable "honky", or coarse PA-coloration when compared to the new ribbon.
It sounds like the new double-high ribbons have more resolving power than the standard models - not surprising considering the 6dB greater efficiency, headroom, and lowered distortion. 6dB improvements are usually immediately audible. It also sounds like it asks more of the widerange driver - less coloration, lower distortion, the whole shootin' match.
Alexander reported good results combining the double-ribbon with an array of Veravox 3 drivers - 3" square metal-coned drivers at 38 euros each. I e-mailed the specs for pair of 18Sound 8NMB420's, and in Alexander's response, he felt it was an "excellent" choice and had "no doubt that the performance of that combo would be fantastic in every way." The pair of 8NMB420's "get pretty much the same Sd, similar Mms, but twice the BL and much less diaphragm related problems than with any 12".
Sounds good to me! Thanks, Alexander - looks like we now have some serious choices for the project.
Health update: Just back from the doctor and the latest X-rays. The leg is now strong enough for full weight-bearing, and I've started Physical Training to improve the strength in my left leg. I tried walking with a cane a couple of weeks ago, and didn't get beyond the first step - in fact, didn't make the first step. I didn't know if it was over-caution, balance, or what. Well, after visiting the PT yesterday, it turns out the upper leg needs to rebuild its strength before I can walk again - the muscles were correctly reporting I couldn't do it.
It wasn't mental, it was physical. I'll be doing strengthening exercises for the next three weeks, then walking and stair-step exercises with the PT. So I'm on my way!
It sounds like the new double-high ribbons have more resolving power than the standard models - not surprising considering the 6dB greater efficiency, headroom, and lowered distortion. 6dB improvements are usually immediately audible. It also sounds like it asks more of the widerange driver - less coloration, lower distortion, the whole shootin' match.
Alexander reported good results combining the double-ribbon with an array of Veravox 3 drivers - 3" square metal-coned drivers at 38 euros each. I e-mailed the specs for pair of 18Sound 8NMB420's, and in Alexander's response, he felt it was an "excellent" choice and had "no doubt that the performance of that combo would be fantastic in every way." The pair of 8NMB420's "get pretty much the same Sd, similar Mms, but twice the BL and much less diaphragm related problems than with any 12".
Sounds good to me! Thanks, Alexander - looks like we now have some serious choices for the project.
Health update: Just back from the doctor and the latest X-rays. The leg is now strong enough for full weight-bearing, and I've started Physical Training to improve the strength in my left leg. I tried walking with a cane a couple of weeks ago, and didn't get beyond the first step - in fact, didn't make the first step. I didn't know if it was over-caution, balance, or what. Well, after visiting the PT yesterday, it turns out the upper leg needs to rebuild its strength before I can walk again - the muscles were correctly reporting I couldn't do it.
It wasn't mental, it was physical. I'll be doing strengthening exercises for the next three weeks, then walking and stair-step exercises with the PT. So I'm on my way!
Different/same Mids
Hi ... In choosing OB speakers we generally come to the conclusion that even though "X" brand will do FL to 20k that is not in the interest of doing what Lynn and you guys are kicking around and that is concert levels and quality.
Using 100hz as low and 4K as high which most near-fullrange speakers would do well, that's still 5+ octaves and a 40:1 ratio. A bit much for our purposes.
Brings us to two speakers. Three speakers perhaps, but for discussion two is easier and we can keep a tweeter out of the picture.
These two speakers, one lower mid and one upper mid would need to cross somewhere between 500hz and 800hz (notice this is traditionally where two-way and three-way speakers have been crossed for decades) so each would carry sufficient octaves. It is very difficult to find two speakers that are compatible and the crossover here is fairly important, but not critical. The purest want no crossover and they have a valid point.
My idea is to simply use two of the same near-fullrange speakers and cross them at their optimum point. What could be more compatible that two identical speakers, and the crossover point should be fairly seamless? Pseudo 100hz to 4K fullrange is possible.
Thinking of two Nirvana Super 12's for starters.
Zene
Hi ... In choosing OB speakers we generally come to the conclusion that even though "X" brand will do FL to 20k that is not in the interest of doing what Lynn and you guys are kicking around and that is concert levels and quality.
Using 100hz as low and 4K as high which most near-fullrange speakers would do well, that's still 5+ octaves and a 40:1 ratio. A bit much for our purposes.
Brings us to two speakers. Three speakers perhaps, but for discussion two is easier and we can keep a tweeter out of the picture.
These two speakers, one lower mid and one upper mid would need to cross somewhere between 500hz and 800hz (notice this is traditionally where two-way and three-way speakers have been crossed for decades) so each would carry sufficient octaves. It is very difficult to find two speakers that are compatible and the crossover here is fairly important, but not critical. The purest want no crossover and they have a valid point.
My idea is to simply use two of the same near-fullrange speakers and cross them at their optimum point. What could be more compatible that two identical speakers, and the crossover point should be fairly seamless? Pseudo 100hz to 4K fullrange is possible.
Thinking of two Nirvana Super 12's for starters.
Zene
Lynn Olson said:
Lynn,
Have you discussed the construction of a dipole double-ribbon with Alexander at RAAL? My experience is that a dipole tweeter is mandatory for the best dipole speaker. Linkwitz's recent epiphany with a rear tweeter on the Orion would be another vote.
I think the RAAL Flat Foil(tm) construction technique will allow their longer double-ribbons to mate better with a midrange speaker than the very loose short foils in most ribbon tweeters. I've been building a type of embossed flat foil ribbon for 10 years to avoid the "light" and "sizzle" ribbon sounds that I heard in my Apogees. People following this thread would have the option to DIY a dipole ribbon at modest cost.
LineSource said:
I've chatted with Alexander over a sizzly Internet connection (What? Pay the going rates for an international call? You must be kidding!) and I vaguely remember a comment from Alexander the existing construction of the RAAL tweeter doesn't allow an open-back structure - it's all magnets back there. I think. I'm not sure of this, since the conversation was several weeks ago.
An interesting angle about the Linkwitz epiphany is SL is recommended a rear tweeter that is 6 dB down from the front tweeter. Well, isn't that convenient - a single RAAL will of course be half the efficiency of a twin RAAL, so a single RAAL on the rear surface should do the job just fine.
So I dunno about the construction of the existing RAAL ribbon tweeters. Alexander uses an interesting and rather complex pattern embossed on the foil to spread out all the resonances so they have an extremely fine grain structure - and in fact cannot be measured more than a fraction of an inch away from the surface of the ribbon.
It's great if people build their own ribbons - more power to them!
I hope it's no problem if I bring back an older topic about TAD TL1102 drivers. The problem when it was first discused was the lack of informations. This thread here has two links with FR, distorsion, CSD plots (post 6).
Although the speakers are not in production anymore (maybe they still can be sourced though) I've mentioned them again because I find them apropiate in the eventuality of a 2 units line array. The distorsions in the 100Hz-1Khz region I find it to be at least great... I just hope I'm not miss-interpretating them.
Although the speakers are not in production anymore (maybe they still can be sourced though) I've mentioned them again because I find them apropiate in the eventuality of a 2 units line array. The distorsions in the 100Hz-1Khz region I find it to be at least great... I just hope I'm not miss-interpretating them.
Well, in addition to being out of production (something I'm trying to avoid), the TADs are something like US$800 each - four to five times the price of an 18Sound or comparable prosound driver. Now, if they had field-coil magnets and the smoothest response I'd ever seen, well, maybe ... but the per-pair price is in the Lowther/AER range, right up in the stratosphere.
Even though this is hardly going to be a cheap loudspeaker, I want every part of the speaker to justify its price, in terms of subjective and measured performance. That's why I have a bias towards items I've heard for myself, or measure particularly well - although the TAD has nothing to apologize for, with superb distortion measurements.
Speaking of items I need to audition (maybe at this year's RMAF show), Alexander had good things to say about PHL as well as 18Sound. PHL has pretty steep pricing the last I checked, maybe I need to look deeper.
Readers of this thread have probably picked up on my preference for spending money on the HF driver, compared to the rest of the spectrum. That's because the HF is most difficult to get right, and most annoying and artificial-sounding when it's wrong. I'm just not very tolerant of poor HF reproduction, since it so grossly degrades the sound of symphonic music.
Thus, the interest in studio-monitor-quality compression drivers, LeCleac'h horn profiles, modern versions of AMTs, and true ribbon tweeters - and with excellent power handling down to 2 kHz (or a bit lower). These are HF drivers that are very different than generic dome tweeters - with more serious technology and higher costs associated with small production runs and custom machining.
Of course, if you've already got a stash of vintage AlNiCo-magnet Altec 515's, 288's, Vitavox S2's, and the accompanying multicell theater horns, well, go for it! But if you're at that level, you probably have a multiway horn system already, and don't need to mess with this project.
Even though this is hardly going to be a cheap loudspeaker, I want every part of the speaker to justify its price, in terms of subjective and measured performance. That's why I have a bias towards items I've heard for myself, or measure particularly well - although the TAD has nothing to apologize for, with superb distortion measurements.
Speaking of items I need to audition (maybe at this year's RMAF show), Alexander had good things to say about PHL as well as 18Sound. PHL has pretty steep pricing the last I checked, maybe I need to look deeper.
Readers of this thread have probably picked up on my preference for spending money on the HF driver, compared to the rest of the spectrum. That's because the HF is most difficult to get right, and most annoying and artificial-sounding when it's wrong. I'm just not very tolerant of poor HF reproduction, since it so grossly degrades the sound of symphonic music.
Thus, the interest in studio-monitor-quality compression drivers, LeCleac'h horn profiles, modern versions of AMTs, and true ribbon tweeters - and with excellent power handling down to 2 kHz (or a bit lower). These are HF drivers that are very different than generic dome tweeters - with more serious technology and higher costs associated with small production runs and custom machining.
Of course, if you've already got a stash of vintage AlNiCo-magnet Altec 515's, 288's, Vitavox S2's, and the accompanying multicell theater horns, well, go for it! But if you're at that level, you probably have a multiway horn system already, and don't need to mess with this project.
New RAAL Tweeter Pricing
Direct from Aleksandar Radisavljevic of RAAL:
"Price is 680 Euro per piece of "doublet", Ex Works.
Can sell directly, except Canada, Germany and Hungary."
Not bad - just a little more than the 1.4" Radian 745PB and matching AH-550 Azurahorn, and about the same as a 2" Radian 850PB/950PB and Azurahorn. As mentioned before, we have some excellent choices for the project!
Direct from Aleksandar Radisavljevic of RAAL:
"Price is 680 Euro per piece of "doublet", Ex Works.
Can sell directly, except Canada, Germany and Hungary."
Not bad - just a little more than the 1.4" Radian 745PB and matching AH-550 Azurahorn, and about the same as a 2" Radian 850PB/950PB and Azurahorn. As mentioned before, we have some excellent choices for the project!
This is not just fair, it's great common-sense
I have to say this is probably the best thread I've read on loudspeaker design. Not because it changed my opinions on most of the decisions someone has to deal with when designing a loudspeaker, but rather for geting an in-depth understanding of some of the problems I didn't even now how to tackle.
Thank you Mr. Olson!
Well, the views in the thread are my own, the preferences (and biases) of an old-time designer. There are other folks out there doing good work - Zaph Audio, Siegfried Linkwitz, etc. etc. Speakers reflect the skill and tastes of the designer, who are aiming for a certain kind of sound when they start the design.
This is why I think the entire notion of "best speaker" is kind of silly - it means the designer doesn't even know where they want to go! In the physical world, if you don't know which direction you're going, you probably won't get there!
Now, if there was a direction in mind, and the design falls short, that's educational too, since the reasons for the shortfall can usually be discovered after the fact. Aiming too high, technology doesn't exist just yet, flaws in the underlying theory, perception doesn't line up with measurements, etc. etc. All discoverable with a little work.
I always have to warn people that there's a reasonable chance one of my designs will fail, since I like to work in areas where there are little or no commercial equivalents - fresh, empty territory, as I see it. Since it is unexplored, though, all kinds of unexpected surprises can lurk in the shadows, and these can trip up the whole concept.
I went pretty far out into unknown territory with the Amity, Aurora, and Karna amplifiers, particularly since I was teaching myself electronic design as I went. I was very lucky the first prototype of the Amity had all the voltages and currents measure within 5% of calculations the first time it was built! It's nice that vacuum tubes are such predictable little creatures - more than you can say of bipolar transistors or MOSFETs.
My instincts - that's the only word I can use right now - is the basic concept is sound, and the only reason there aren't lots of speakers like this on the market right now is the political divide between the audiophile and prosound community.
This is why I think the entire notion of "best speaker" is kind of silly - it means the designer doesn't even know where they want to go! In the physical world, if you don't know which direction you're going, you probably won't get there!
Now, if there was a direction in mind, and the design falls short, that's educational too, since the reasons for the shortfall can usually be discovered after the fact. Aiming too high, technology doesn't exist just yet, flaws in the underlying theory, perception doesn't line up with measurements, etc. etc. All discoverable with a little work.
I always have to warn people that there's a reasonable chance one of my designs will fail, since I like to work in areas where there are little or no commercial equivalents - fresh, empty territory, as I see it. Since it is unexplored, though, all kinds of unexpected surprises can lurk in the shadows, and these can trip up the whole concept.
I went pretty far out into unknown territory with the Amity, Aurora, and Karna amplifiers, particularly since I was teaching myself electronic design as I went. I was very lucky the first prototype of the Amity had all the voltages and currents measure within 5% of calculations the first time it was built! It's nice that vacuum tubes are such predictable little creatures - more than you can say of bipolar transistors or MOSFETs.
My instincts - that's the only word I can use right now - is the basic concept is sound, and the only reason there aren't lots of speakers like this on the market right now is the political divide between the audiophile and prosound community.
Lynn Olson said:
The divide is real, but why does it exist? Certainly different goals and objectives.
- Pro stuff has to take a beating and still work - day after day. Audiophile gear does not.
- Pro gear has to fill huge spaces with high SPLs, Audiophile gear does not.
- Pro stuff has to sound good, Audiophile stuff is supposed to sound great.
- Pro gear is big and heavy, Audiophile is not.
- Etc.
Funny thing is, a lot of the pro drivers sound more like real music, even purely acoustic music, than the home audiophile drivers.
Does it all come down to size? And maybe expense? Market forces? Why the difference?
I suspect it's because pro driver companies also make musical instrument drivers and associated equipment. From there they develop engineering traditions and rules of thumb of the kind of stuff that musicians like, and like to buy.
Imagine a fashion designer growing up in Milan or New York - he will already have quietly sucked in a lot of information from the sophisticated style that surrounds him every day before he even starts studying fashion, and later have a huge headstart in his intuitions compared to a poor fella coming from say a coal and steel industrial town... (no disprespect to the coal and steel industry. but the way of life *is* different).
So the pro audio designers make trade offs like everyone else, but by tradition I suspect they make the right ones because they deal with professional musicians and associated personnel every day.
Home audio and audiophile designers by comparison have feeble *specific* customer feedback, other than sales, and whatever feedback there is is comparatively "uneducated", music production wise. So they have to go by textbook engineering, data, and of course, by the customer's technology fad du jour.
Corrections
Mr. Olsen
It was suggested that I view this thread so I did. I certainly agree with much of what you say, but I certainly disagree with much of it. The following statement by you, however, is incorrect, unfair and not very polite.
"Well, this is where individual perceptions enter in. Earl Geddes, ... , believes Costco-quality electronics are good enough to exhibit at the RMAF, and anyone who thinks otherwise is fooling themselves, or worse, trying to sell snake oil to the gullible."
First the electronics that I used did come from Costco but they where non-the-less high quality pieces, Pioneer receiver and DVD player. The digital signal from the player was delivered to the receiver digitally thus basically negating any effect of the player on the sound quality (bits are bits and the clock is in the receiver not the player). The signl path in the receiver is digital and thus there is only a single D/A just before the amps.
What I most object to about your position at RMAF was the fact that you concluded "bad sound" with absolutely no data to support your view other than my claim that the equipment came from Costco. You basically had no idea if it was good or bad, only that it was inexpensive. This offended you and thus you "heard" poor quality electronics. (Do you believe that price is a good predictor of quality? I should hope not!)
The fact of the matter is that I had tested this particular piece of equipment (Pioneer receiver) and its performance is actually exemplary. So, in fact, I had data to support my claim that the electronics that I used here were "good enough", while you, in fact, did not. Unless your "perception" is infallible (you would be the first) you were jumping at conclusions that you could not support.
I have never said that SS is great or that tubes were bad, only that good design is good design no matter what materials are used. And I wholeheartedly agree with SL that it is crossover distortion and various other amp features not shown by THD or "common" measurements that make the biggest differences in amp quality - not the devices themselves. The Pioneer amp that I used at RMAF was as good a design in this regard as I have ever tested (I don't used standard THD tests, I use my own proprietary tests).
Quite simply your attributing things to me that I don't believe and have never claimed does not bode well for you credibility in a forum like this.
Mr. Olsen
It was suggested that I view this thread so I did. I certainly agree with much of what you say, but I certainly disagree with much of it. The following statement by you, however, is incorrect, unfair and not very polite.
"Well, this is where individual perceptions enter in. Earl Geddes, ... , believes Costco-quality electronics are good enough to exhibit at the RMAF, and anyone who thinks otherwise is fooling themselves, or worse, trying to sell snake oil to the gullible."
First the electronics that I used did come from Costco but they where non-the-less high quality pieces, Pioneer receiver and DVD player. The digital signal from the player was delivered to the receiver digitally thus basically negating any effect of the player on the sound quality (bits are bits and the clock is in the receiver not the player). The signl path in the receiver is digital and thus there is only a single D/A just before the amps.
What I most object to about your position at RMAF was the fact that you concluded "bad sound" with absolutely no data to support your view other than my claim that the equipment came from Costco. You basically had no idea if it was good or bad, only that it was inexpensive. This offended you and thus you "heard" poor quality electronics. (Do you believe that price is a good predictor of quality? I should hope not!)
The fact of the matter is that I had tested this particular piece of equipment (Pioneer receiver) and its performance is actually exemplary. So, in fact, I had data to support my claim that the electronics that I used here were "good enough", while you, in fact, did not. Unless your "perception" is infallible (you would be the first) you were jumping at conclusions that you could not support.
I have never said that SS is great or that tubes were bad, only that good design is good design no matter what materials are used. And I wholeheartedly agree with SL that it is crossover distortion and various other amp features not shown by THD or "common" measurements that make the biggest differences in amp quality - not the devices themselves. The Pioneer amp that I used at RMAF was as good a design in this regard as I have ever tested (I don't used standard THD tests, I use my own proprietary tests).
Quite simply your attributing things to me that I don't believe and have never claimed does not bode well for you credibility in a forum like this.
Welcome to the forum, Earl
My hope is that we can learn more from your participation. It would be interesting to know more about your proprietary tests...and any improvement in correlation with "percieved" sound to the ear.
I'd also be interested in your comparison of the waveguide with something along the lines of an Azurahorn...I have your book, "Audio Transducers"
TIA
My hope is that we can learn more from your participation. It would be interesting to know more about your proprietary tests...and any improvement in correlation with "percieved" sound to the ear.
I'd also be interested in your comparison of the waveguide with something along the lines of an Azurahorn...I have your book, "Audio Transducers"
TIA
Ed
First, I'd like to recommend to all audiophiles that they read the book "Black Swans" by Nassim Taleb. While not specifically about audio (its about economic forcasting) its subject matter is more than relavent to the problem of determining truth in audio. We humans are so prone to elevate our illusions to reality that it threatens all that we think that we know.
Lets suppose that in an amplifier the most important criteria for perception is crossover distortion (there is a lot of data to support this). Now this means that looking at the typical THD versus output level is kind of irrelavent because the rise at the higher power is not subjectively important and the rise at lower power level is due to the fact that virtually all measurements of this type measure "THD + noise", meaning that it is impossible to determine how much crossover distortion exists from one of these plots because the noise will mask it. THD versus frequency is likewise irrelavent because this test is done at a mid level and crossover distortion is masked again.
I measure the spectrum of a sine wave as the level is dropped down into the noise floor. But I do this using a sychonous averaging technique with a window that is locked to the sine wave thus giving me data well below the noise floor. I look at the spectrum - the harmonics of the tone. This measurement shows serious problems with many amplifiers (alas NOT the Pioneer that I used at RMAF). Does it correlate with a subjective impression - I have too little interest in amplifiers to do this test so "I don't know". But IF crossover distortion is significant THEN its not currently being assesed in the marketplace. This, to me, is a big issue, but I'll let the amplifiers guys worry about it.
As to the waveguide comparisons, I don't know what an Azurahorn is so I can't comment.
First, I'd like to recommend to all audiophiles that they read the book "Black Swans" by Nassim Taleb. While not specifically about audio (its about economic forcasting) its subject matter is more than relavent to the problem of determining truth in audio. We humans are so prone to elevate our illusions to reality that it threatens all that we think that we know.
Lets suppose that in an amplifier the most important criteria for perception is crossover distortion (there is a lot of data to support this). Now this means that looking at the typical THD versus output level is kind of irrelavent because the rise at the higher power is not subjectively important and the rise at lower power level is due to the fact that virtually all measurements of this type measure "THD + noise", meaning that it is impossible to determine how much crossover distortion exists from one of these plots because the noise will mask it. THD versus frequency is likewise irrelavent because this test is done at a mid level and crossover distortion is masked again.
I measure the spectrum of a sine wave as the level is dropped down into the noise floor. But I do this using a sychonous averaging technique with a window that is locked to the sine wave thus giving me data well below the noise floor. I look at the spectrum - the harmonics of the tone. This measurement shows serious problems with many amplifiers (alas NOT the Pioneer that I used at RMAF). Does it correlate with a subjective impression - I have too little interest in amplifiers to do this test so "I don't know". But IF crossover distortion is significant THEN its not currently being assesed in the marketplace. This, to me, is a big issue, but I'll let the amplifiers guys worry about it.
As to the waveguide comparisons, I don't know what an Azurahorn is so I can't comment.
Hi Earl, your comments are appreciated!
My experience with electronics goes back to my days at Audionics, from 1973 through 1979, followed by the Tektronix years, from 1979 through 1988. I learned about amplifiers from Bob Sickler, designer of the Audionics CC-2, an early complementary-symmetry low-TIM amplifier - one of the first on the US market. The first in the world was probably the Electrocompaniet designed by Matti Otala, if I recall right - and we were about 6 months behind them.
I have been consistently disappointed by the design and sonics of most amplifiers on the market, particularly the mass-market products - this is an area where we appear to disagree.
The specific issues with nearly all commercial amplifiers are:
1) Class AB switching transitions that are not corrected by feedback, and are most visible when the amplifier is driving a reactive load (which decreases phase margin, which in turn decreases the effectiveness of feedback). Back when I was building and testing the CC-2, I would amuse myself by moving the wires for the power supply rails around and watching the distortion residue wriggle around on the scope. We were using Radford distortion analyzers with a 0.0008% residue, so the switching artifacts were well above the noise+distortion floor.
The problem is made worse by the Vbe multiplier that sets the output device bias always being a few seconds behind the temperature of the output transistor die, thanks to the thermal transmission time from the die to the Vbe multiplier transistor sensor mounted on the heat sink. This has the malign effect of the tempco compensation always being behind the pulse of the music. It is only in the last year we finally have transistors with temp sensors directly on the die. Better thirty years late than never, I guess.
2) When the amp clips, the recovery time is stretched out by poor phase margin which is worsened by the reactive load of the speaker + connecting cable. The overload performance on the test bench and real loads are two completely different things - test-bench overload is an ideal case that is never met in practice. A similar problem happens when the overcurrent protection circuits activate - they can get confused by reactive loads and go into brief periods of oscillation, toggling the amplifier in and out of current-limit mode.
3) The bridge rectifiers for the split-rail power supply have a small amount of capacitance that resonates with the stray inductance of the power transformer. When examined on a fast scope, several cycles of oscillation are visible in the 3~10 Khz region. Soft-recovery or HEXFRED diodes can help here, but these are never used in cost-sensitive mass-market products. Custom RC snubber circuits can help as well, but these must be tuned to the specific model of power transformer.
4) There is an interrelationship between slew rate, phase margin, reactivity of the load in the 20 ~ 200 kHz region, and speed of recovery from overload - either voltage or current-domain. It is possible to design amplifiers with adequate slew rate and phase margin with a resistive load, but when exposed to real-world loads, the recovery time is stretched out 4 to 5 times compared to the bench-test load. In the op-amp world, this is known as "settling time", and is a key figure-of-merit between different devices.
This is so common that amps that don't exhibit this behaviour are quite unusual. The lack of phase margin is also part of the reason that the feedback loop has ineffective correction of the Class AB transition or the switch-noise pulses of the bridge rectifiers. Very few amplifiers have adequate phase margin with real-world loads in the ultrasonic region.
5) There are other problems - large shifts in beta and Hfe with device temperature, grossly nonlinear capacitance at the base/gate of bipolar/MOSFET power devices, nonlinear electrolytic caps in the feedback loop, marginal HF stability of active regulators for the low-level stages, but the four mentioned above are the most serious problems. The others aren't trivial, though, and affect the sound as well.
I would be frankly astonished if any moderate-price amplifier addresses any of these problems, since they are almost universal in transistor amplifiers. When the price approaches US$5000 on up, then you start to see a handful of designs that start to tackle the problems mentioned above, but it's not common by any means.
Can I hear these things? Not in a 1,2,3 way, but more generally as low-fi sound. It sounds "dirty", with obscured low-level detail. When the problems are addressed one by one, the sonic problems go away too. Bob Sickler and I did this back in 1979. It's not that different than removing a resonance from a loudspeaker - when it's gone, the speaker sounds better. The same applies for amplifiers.
The audibility - or non-audibility - of the above-mentioned defects is an area where we appear to disagree. Rest assured everything I've mentioned is measurable if you know where to look - they were problems back when I was Audionics in 1979, and they remain so today.
One of the best things about the DIY world is the most important sonic problems can be addressed, instead of placing marketing first and foremost. That Nelson Pass is a participant here at diyAudio speaks volumes for the forum.
My experience with electronics goes back to my days at Audionics, from 1973 through 1979, followed by the Tektronix years, from 1979 through 1988. I learned about amplifiers from Bob Sickler, designer of the Audionics CC-2, an early complementary-symmetry low-TIM amplifier - one of the first on the US market. The first in the world was probably the Electrocompaniet designed by Matti Otala, if I recall right - and we were about 6 months behind them.
I have been consistently disappointed by the design and sonics of most amplifiers on the market, particularly the mass-market products - this is an area where we appear to disagree.
The specific issues with nearly all commercial amplifiers are:
1) Class AB switching transitions that are not corrected by feedback, and are most visible when the amplifier is driving a reactive load (which decreases phase margin, which in turn decreases the effectiveness of feedback). Back when I was building and testing the CC-2, I would amuse myself by moving the wires for the power supply rails around and watching the distortion residue wriggle around on the scope. We were using Radford distortion analyzers with a 0.0008% residue, so the switching artifacts were well above the noise+distortion floor.
The problem is made worse by the Vbe multiplier that sets the output device bias always being a few seconds behind the temperature of the output transistor die, thanks to the thermal transmission time from the die to the Vbe multiplier transistor sensor mounted on the heat sink. This has the malign effect of the tempco compensation always being behind the pulse of the music. It is only in the last year we finally have transistors with temp sensors directly on the die. Better thirty years late than never, I guess.
2) When the amp clips, the recovery time is stretched out by poor phase margin which is worsened by the reactive load of the speaker + connecting cable. The overload performance on the test bench and real loads are two completely different things - test-bench overload is an ideal case that is never met in practice. A similar problem happens when the overcurrent protection circuits activate - they can get confused by reactive loads and go into brief periods of oscillation, toggling the amplifier in and out of current-limit mode.
3) The bridge rectifiers for the split-rail power supply have a small amount of capacitance that resonates with the stray inductance of the power transformer. When examined on a fast scope, several cycles of oscillation are visible in the 3~10 Khz region. Soft-recovery or HEXFRED diodes can help here, but these are never used in cost-sensitive mass-market products. Custom RC snubber circuits can help as well, but these must be tuned to the specific model of power transformer.
4) There is an interrelationship between slew rate, phase margin, reactivity of the load in the 20 ~ 200 kHz region, and speed of recovery from overload - either voltage or current-domain. It is possible to design amplifiers with adequate slew rate and phase margin with a resistive load, but when exposed to real-world loads, the recovery time is stretched out 4 to 5 times compared to the bench-test load. In the op-amp world, this is known as "settling time", and is a key figure-of-merit between different devices.
This is so common that amps that don't exhibit this behaviour are quite unusual. The lack of phase margin is also part of the reason that the feedback loop has ineffective correction of the Class AB transition or the switch-noise pulses of the bridge rectifiers. Very few amplifiers have adequate phase margin with real-world loads in the ultrasonic region.
5) There are other problems - large shifts in beta and Hfe with device temperature, grossly nonlinear capacitance at the base/gate of bipolar/MOSFET power devices, nonlinear electrolytic caps in the feedback loop, marginal HF stability of active regulators for the low-level stages, but the four mentioned above are the most serious problems. The others aren't trivial, though, and affect the sound as well.
I would be frankly astonished if any moderate-price amplifier addresses any of these problems, since they are almost universal in transistor amplifiers. When the price approaches US$5000 on up, then you start to see a handful of designs that start to tackle the problems mentioned above, but it's not common by any means.
Can I hear these things? Not in a 1,2,3 way, but more generally as low-fi sound. It sounds "dirty", with obscured low-level detail. When the problems are addressed one by one, the sonic problems go away too. Bob Sickler and I did this back in 1979. It's not that different than removing a resonance from a loudspeaker - when it's gone, the speaker sounds better. The same applies for amplifiers.
The audibility - or non-audibility - of the above-mentioned defects is an area where we appear to disagree. Rest assured everything I've mentioned is measurable if you know where to look - they were problems back when I was Audionics in 1979, and they remain so today.
One of the best things about the DIY world is the most important sonic problems can be addressed, instead of placing marketing first and foremost. That Nelson Pass is a participant here at diyAudio speaks volumes for the forum.