Historical look at amps


The amplifier evolution thread reminded me of the history of amplifier circuits that has occured over the last 20 years. Lots of changes but the one that stuck in my mind was the change in feedback circuits. In the early 1980s a good amp like Crown, McIntosh, Phase Linear etc all had large amounts of feedback and distortion levels of 0.00001% IM and THD. These amps sounded bad and the question was raised (and still is) why objective measurement didn't jib with listening tests. A Finnish engineer (OTTELA) came up with a new measurement called Transient IM Distortion (TIM). I wont go into the details but it did show that large amounts of feedback which made static IM and THD measurements good, made music waveforms bad. The result has been today's amps with low levels of global and local feedback, and better sound but with IM distortion levels of only 0.01% (and of course tube amps with more even then odd distortion harmonics). Just recently Ayre, and probably other companys are offering zero feedback designs. Feedback circuits have been with us since the 1920s and we are now just elliminating this basic design feature in modern amps and preamps.
keis
Hey Ar_t,
I cut & paste an exerpt from your earlier post:-

"The crux: "current-feedback" amps have a bandwidth that remains constant, regardless of gain. Traditional feedback amps do not: as the gain increases, their bandwidth goes down."

This does *not* read correct to me! AFAIK, the gain-bandwidth product of a voltage-mode amp is fixed for a certain topology implemented by the designer & with a fixed set of devices. What I have seen tho is that as one increases the gain, the dominant pole frequency goes down & when the gain decreases, the dominant pole frequency increases. (you've got to have compensation somewhere in there to counter-act the phase shift of the music signal as it propagates thru the amp). However, the unity gain frequency, often used as the bandwidth of the amp, remains fixed. The unity gain freq can be increased by reducing the in-circuit parasitic poles (better devices), cascoding & by increasing the bias current.
Maybe I understood incorrectly. Perhaps you can clarify?

BTW, there is a company called Comlinear that makes current-mode opamps & other electronics. I don't think that it's audio grade stuff (but then it might be!) A lot of their circuits are based on the current conveyor (CC II) concept. I've never used them personally but their architectures sure look different than traditional voltage-mode amps. They also seem to be perhaps the lone star player in that field.
Muralman1, while I can understand how you came to believe that Ar_t is "Expounding on a trait that is most damning", I believe that he has qualified that assertion.
As far as "class D amps on my (your) speakers." and "There are plenty of speakers, of all types being run by class D." in my previous post I said the jury is still out on the qualified generalization that Ar_t offered. Ar_t is entitled to his opinoin as much as anyone else here. I think my mention of sodium intake adressed that. Never the less it would appear that Ar_t may have more insight in these matters than most. Untill other manufacturers are as generous with their wisdom, we are unfortunately weighted in one direction. For that matter it sure would be nice if more users posted their experience as well. I'll take what's available and appreciate it. I'm not saying that any one is right or any one is wrong.
As far as why Ar_t doesn't build amps like Henry's, well, is Henry unable to keep up with the demand? I mean why would he? That's like saying why doesn't Krell make amps like c-j and visa versa. Ar_t has already said that his amps were originaly intended for the Home Theater crowd. He is obvioulsy found his market segment target. I may be extrapolating here, but, I suspect that Ar_t is himself surprised at how well this technology in his products overlap into the pure audio market and that perhaps there is an audience there as well.
I don't see any problem with Ar_t sharing his experience and qualifiactions. I think that's a good thing. As far as "I don't see any other amp builder doing that." Well I don't see many other amp builders posting here and that's a shame. I think it's in all our best interest to encourage this kind of input. That's the reason I'm posting this. I don't own and never have to my knowledge owned any of Ar_t's products. I have never met Ar_t. Before these recent Audiogon threads I never knew he existed. BTW, Ar_t has offered many compliments to other designers.
I think that Ar_t's opinon is that without the above mentioned designers skills these amps might have a tendency to be bright due to some inherent attributes. What may be lost is the compliment he is offering to these designers and their implamentations. This is something like the discussion on feed back in tradtional solid state amps. I'm not trying to re kindle a tube vs. solid state argument, but, this is much like the way that tube amp designers have had to work with and with out transformers in order to address those concerns that are more of an issue with that technology. That the qualtiy of an amp is system dependent seems to be accepted wisdom. Again the jury is still out on whether digital amps have a propensity for brightness. There is obviously a diference of opinion. My own very limted experience with digital amps raised this very question. I don't have enough experience to determine the answer.
I think that your cut and paste may be mistaken. Either that, or I did not explain it fully.

Voltage feedback.......the "normal" kind. As you increase the gain, the bandwidth DOES drop. The product remains constant, but that is not the same. Conversely, as you lower the gain, the bandwidth increases.

One of the reasons firms like Comlinear came up with "current-feedback" is that (within reason) you can change the gain and NOT affect the bandwidth.

OK.......more damning inside info!!!!!

Any one here have preamp based on the Burr-Brown PGA2310?? Anyone???

Ever stop to think how they get such a large range of attenuation, all the way down from......I dunno.......-90 or more.....all the way up to a gain of.......30 dB or so. A lot of range, right. How do they do that? It must be some sort of miracle, because competing digital potentiometers have ranges less than half as much. Right?

Wrong! Here is what they do:

It uses some sort of digital pot, just like the others, to give a narrow range of adjustment. (I forget what....I measured it one day.) But at a certain point, as you slowly bring the level up or down, they suddenly switch the gain on the internal op-amp. Easy to do by changing the resistors in the feedback path.

OK.......what does this have to what the subject at hand?

If you measure the noise on the output, you will see that it jumps when they switch the resistors in the feedback loop.

And the bandwidth changes too! The -3 dB point changes. It changes so radically that at full gain, the response at 20 kHz is down quite a bit. I forget what, but it was quite a lot. It may have been -2 dB. Don't quote me, because I did not write it down. if you want to know the exact number, ask me and I will measure it.

As for my amps, would you rather that I lie and say that they are the greatest ever? Would that make you want to rush out and buy one? I bet it wouldn't. But if you did, and it sounded bright next to your SS amp, you might be mad. Look, as far as I know, no one who is posting in this thread has design experience comparable to me. So why do you feel threatened when I tell the dark secrets of the world of amp design. I would think that frank and open discussion would be welcome. When you are asking someone to fork over $3k or more on something that they are gong to put in their sound room, and stare at it for (hopefully) years to come, is trying to pull the wool over their eyes going to make them want to keep it?? I would rather be upfront about the strengths and weaknesses of my products. Much better than hyping them to death, getting a handful of quick sales, only to see them here in the "used amps for sale" section 6 months later. That kind of advertising does not make for a long-lived business.

But the truth is, most listeners "think" that a "digital" amp, with the same - 3 dB as a conventional SS amp, will "sound" bright.

Oh......I almost forgot........this is important......sorry.....but:

The guys who think that they sound bright all have cone transducer speakers. The ones who don't think that it sounds bright.........drum roll........have planar speakers! Sorry, I forgot to mention that. Probably because I had already sold amps to everyone that I could who owned planar speakers, and now I am trying to sell to the rest of the world. And they have cone loudspeakers. Now I have headaches.

Calm down, Vince.......it was an honest omission.

In fact, I recently sent out an evaluation unit. The guy spoke very highly of it. He liked it better than the "Brand N" amp that he was a dealer for. I knew that it was a winner because he and his girl friend were able to listen to it for hours on end. Then he complained about the binding posts. And said that I might want to "tone down" the HF a little, as he felt that most people with a cone speaker system would find it a bit bright. He said it, not me. Remember, he liked it.

Oh......btw......I do know how mine stack up against Rowland, Ear, and H2O. Wanna guess how??????? I'll give you a hint: guess which one they bought. Frankly, I'm not sure why. Maybe they wanted something that looked almost as cool as the Rowland, but less expensive. Or they did not want to have to fight trying to get big fat speaker cables up inside some crazy looking triangular gizmo with Speakon connectors.
Well, it has been very informative but I think my time spent on this topic has come to an end. I still think Ayre is being deceptive by using the term "zero feedback," but unless we get some more information on their amplifier design I'll never know for sure. Ayre never responded to my email asking them to clarify and I did not expect they would.

I sure don't expect Hansen to jump in at this point and clear it up even though you can be sure he is watching this thread (I'm suprised he answered Keis' question at all even though it didn't tell us anything), and Ar_t evidently knows something that he can't or doesn't wish to share about the design. I understand that.

In any case, thanks for the history lesson Ar_t. I know more about amplifier design and the business of bringing them to market now than I did a few days ago so I suppose it was time well spent.
Ar_t you won't find the H2O at a dealership, nor the eAR. Being they are rare, the odds they both are in the same room with your amp, and a Rowland, sounds like the stuff of fiction. Who, and how, did this shootout come about?

Let's settle this. Send your amps to California for review. I will pay shipping both ways. What have you got to lose?

BTW, The Gallo is not a planar, neither is the Meadowlark, nor German Physics... etc. etc. Like I said, do a search.
Seriously, I only saw one model Ayre amp. Can't even remember the model number. Heard several of them, in more than one location. They were the hot ticket to go with a certain Thiel speaker, that badly needed "taming". Last time I saw Charlie Hansen, he was at Avalon. If I knew more, I would say so. The only reason that I know as much as I do, they did silk screen the schematic on the inside of the lid.

OK.....tweeter level and amp brightness......as promised.

A buddy who designs speakers calls me one day, fretting over what level to set the tweeter on his latest product. If he used the one made by a mutual friend, it was one setting. If he used one of the "East Coast Big Boys", he had to set it 0.25 dB different. He asked for my advice.

My advice was: "Well, who is the target user? Since we all share dealers together, it may be a safe bet that they will have our CD player, and an amp/preamp combo by "our buddy". I would set it up to sound right with it."

"Yeah, but......if I do, it won't sound right on the other one."

"Who cares? None of your customers can afford that stuff."

"Yeah, but......if I send it to Stereophile, and they do a review on it, you know what kind of amp they are going to hook up to it."

"You're screwed, bub. You lose either way."

I know some of you are going; "A quarter of a dB. One quarter of a dB makes that much difference? C'mon, bub, you're pulling our leg."

Nope, 0.25 dB is a LARGE difference. You figure that level change on this speaker is from around 5 kHz to 20 kHz. Two octaves. Definitely the difference between "just right" and "too bright/too dark".

Back to amps........you-know-who frets about 0.2, maybe 0.3 dB, on his "digital" amps. From around 10 kHz to 20 kHz. One octave. Yes, not as wide a range, but I can assure you that much level difference in an octave is audible.

Just like the 0.25 dB over one octave in an RIAA network. In absolute numbers, taken at one point, not much. Added all together, over one or two octaves, a lot of energy.

OK.....more food for thought on amp design.

EVERY amp designer that I know will tell you, if they are honest, that if there is "too much" going on in the HF region (like overshoot in a cascode stage, very easy to do), that the bass will sound wrong. Getting the bass "right" on a conventional SS amp is not as easy as you might think. A lot of things can creep in that cause too much HF energy. Everything from circuit topology to power transformer and filter cap selection.

The worst sounding amp that I ever made had some fancy caps, intended for SMPS. Low ESR, low ESL, put several in parallel to lower those numbers even more.

Absolutely no bass at all. None. Zero. Nada. Useless for anything other than PA use.

Put in some regular ol' filter caps, only one per rail. Product ready to ship. Go figure.
agree, Herman, time to exit this thread.

Ar_t has been very helpful in revealing the dark secrets of amp design. I don't know if this "So why do you feel threatened when I tell the dark secrets of the world of amp design." was directed at me of Muralman1? IFFF it was directed at me, I'll say that I am not threatened in the very least. Au contraire, I stand to gain a lot more knowing the info shared in the thread. Your having 20 years of product design doesn't diminish my capabilities one bit. In fact, we can talk at the same level, which might be enjoyable.

Anyway, what should have been learnt from this thread is that manuf will do the bare minimum to disclose aspects of their design. Even good/very good manuf held in high(er) esteem by consumers. There is no standards committee in this audio consumer market that'll keep them honest. The only people who can do this is, us, the consumer. This will happen ONLY if we consumers are educated enough to ask them pointed questions that'll give us an idea of what the manuf might have done inside the product we are about to spend a ton of cash on. Like Ar_t said, if someone is putting down $3K on audio gear, expect some serious questions to be asked! Not unreasonable.
Also, don't simply eat hook, line & sinker what the manuf has to say about his product. Ask some serious questions. Try to understand their implementation.
By keeping manuf-consumer loop closed, both parties will benefit - we will supply feedback about their products, keep them honest & let them know where their performance is sub-par. Hopefully this will not fall on deaf ears & they'll make products that we are seeking using the best possible methods to implement them. You want audio to be around, we'd better help each other. No other way out. (you do this at your place of work with your suppliers, correct? why is audio any different?)
I did not say that I heard them at a dealer.

I sent them to guys who had aleady brought one of those models in to audition. And sent them back, unsatisfied. There are a few times when I found someone who actually owned one of those brands. There is a big difference when doing a serious side-by-side comparison, as opposed to a "shootout", where there has to be a winner and a loser.

Actually, you may recall, I didn't buy their explanation why they preferred ours over the others. Some guys get carried away exaggerating minor warts. The guys that I placed the most faith in their reasoning was the ones that said they all sounded very close, but chose ours for other reasons. Features........ease of use and interfacing.......looks, whatever. As grateful as I am when someone thinks ours sounds better, I tend to dismiss their enthusiasm when it overflows with superlatives.

I know the Gallo isn't planar. Strangest looking thing I have ever seen.

No, I am not doing a Google search, because I know the truth. I have to.......I build this stuff.

Honestly, I do not see why you seem so defensive. I have not maligned Henry's, or anyone else's, efforts on this genre of amps. I may have a different opinion wrt their pros and cons. They may be among the best we have to work with at present, but I seriously expect that we will have many other options in the near future. And you can rest assured that I will be just as honest and forthcoming about them, at that time.
Guys,
Oh no! You all have got Muralman started again!
To Muralman the H20's are the perfect amp, and can do no wrong...let him continue to live in his fantasy world.
Let's not jump all over Muralman without the consideration of his rather unique situation. How many of us are driving one of the most difficult loads (Apogee Scintilla 1 ohm) in all of home audio? Very very few I'll bet. I had Apogee Calipers and although they are easier to drive than Scintillas they are still a bitch. OTOH I am not endorsing the H2O as I've never heard it but I suspect there are only a limited number of amplifiers that will drive the Scintillas properly. He found one. So what? Let him have some fun.
Audio Girl, we all live inside our heads, don't you know? There isn't a thing you feel, touch, hear, that isn't a product of your brain activity. :)

Ar_t has said, repeatedly, tucked into his long winded lectures, ICE highs are wrong. I say, the ICE module is a chameleon. Actually, I'm borrowing that from another thrilled H2O owner. It can be made to sound like just about anything the designer wants.

Henry's amp has a full kilowatt high quality transformer, and over 70,000 mƒ of capacitance. These figures far exceed that of competing amps. There are other devices that I'll wager is unique in class D amp design.

Whether that is bad or good, one can debate. One thing is for sure, Henry builds amps his own way. No one should assume what they sound like.
It might be a little late to comment on this, but here at Atma-Sphere we use an old school means of defining feedback:

A cathode or emitter resistor is a form of negative feedback called *degeneration*. This form of feedback occurs in real time and also has the effect of increasing the output impedance of the circuit. A secondary form of degeneration is the type where a push-pull amplifier can cancel even-ordered harmonics in its output. In any case the feedback is characterized by being in exact real-time opposition to the output signal.

The second type, the 'bad' kind IMO, is *loop* or *global* feedback and is the more common variety referred to loosely as negative feedback. In this form, a portion of the output signal of an amplifier circuit is applied back to the input of the circuit (and can be tube, transistor or opamp). Due to propagation delays in the amplifier circuit itself, the weakness of this approach is to be increasingly less correct as the frequency increases, and/or less accurate as the propogation delay through the circuit is increased. IOW it is not in real-time opposition to the signal. Generally the output impedance is said to be lowered with this type of feedback.

I personally do not like loop feedback as it increases odd-ordered harmonics (depending to certain degree on the amplifier circuit too) in the range at or beyond *about* the 9th harmonic up the 17th- the area that the human ear (over millions of years) has evolved to use to detect loudness. Obviously we cannot change our ears! IOW, these harmonics make an amplifier sound louder and harsher, and they only need be in vanishingly small amounts to be noticed. Humans are *very* sensitive to this type of distortion.

I feel that the appearance of low and 'zero' feedback amplifiers ('zero' meaning zero loop feedback) in the last two decades is a good thing. Negative feedack is a character of the ruling Voltage test and measurement paradigm, which has been in place for some 50-60 years. It has been responsible for a lot of foolishness over that time, negative feedback's acceptance being one of them.

The opposing paradigm, the Power paradigm, uses slightly different rules of test and measurement, which conform more closely to the rules that the human ear has developed over the millennia. Horn speakers, ESLs and magnetic planar speakers all operate according to the power paradigm. So these zero feedback amplifiers that have appeared have some venerable company...

Back to our regularly scheduled flames, already in progress :)

Horn speakers, ESLs and magnetic planar speakers all operate according to the power paradigm
... another manufacturer (Mr Pass) also expressed an interest in this and actually played with a number of amp-spkr applications. The spkr units used were wide-rangers, with high magnetic fields (and correspondingly low q -- Lowthers, etc) and the results quite convincing.

A simple point of departure in considering feedback could be the very low output impedance achievable for the amp using (loop) f/back... Now, if we look at an amp alone and we concur that low output impedance is very impressive, then why not be impressed? But then we can likewise be imporessed by its horsepower, looks, engineering, etc.

If OTOH we consider the amp as ONE component of the amp-spkr interface -- i.e. we see it as an operational tool doing a specific job rather than a finality -- then the importance (or not) of its output impedance approaching zero may not be as impressive per se... we would be more impressed with 1st how the combo SOUNDS, and THEN (2nd), what the electrical and design attributes of the amp that sounds "good" are.
For example, we may conclude it's preferable to have an amp with absolute steady impedance fm 5Hz-1MHz, even if we only achieve say 1,5 ohms invariable o-impedance... (actually that's not bad). We might even decide that the keeping phase and minimising signal propagation delays due to loops may be useful: THEN, these (measurable) characteristics, rather than the previous ones, will be "impressive".

In my longwinded way, I'm suggesting that:
*quoted design, standard performance and measurement characteristics currently offered are mostly true and real -- BUT do not necessarily influence the resulting SOUND in a positive manner;
*other design..etc..characteristics may have a more immediate relation to the sound resulting fm amp-spkr interface, and those too are available -- but less standard;
*In older times, when amp+spkrs were part of a single package, there was very little problematic regarding this amp or that, etc...
*Ultimately, as atmasphere has already indicated, it's VERY difficult to design the amp that will do the best with ALL full-range (or otherwise) passive speakers. Indeed, IMO, this item will either be a statistical aberration or ridiculously expensive and experimental, or all of the above.
Cheers