GRollins said:
No, that's putting words in my mouth and distorting my position, something you do very well.
I heard the Halcros at RMAF last year and they sounded great. Nevertheless, I do not automatically assume that an amplifier sounds good because it has negative feedback and low THD. I do not make such generalizations. You are the one that makes such ridiculous generalizations, such as to the effect that negative feedback screws up the sound when you in fact don't know what screwed up the sound, or made it sound just different in a way that you happen to prefer.
At the same time, I probably would not go out and spend $40K for a pair of Halcros.
Cheers,
Bob
Charles Hansen said:
To correct the record, Mark B. was at National years before development of the new LME op. amps or the LME driver stages started. He got to National in late '99. But if you know him you can probably just call him for any info and then you know he loves to talk
What significance does his presence bring? -SL
Charles Hansen said:
Hi Charles,
I don't think you are competitor bashing here, and I am interested to hear your views of the Halcros.
I believe that Stereophile is one of the more credible review magazines out there. They have a fairly large stable of reviewers that have listened to the Halcros over many of the designs and many years. Although they may have a writer who has a negative opinion of the Halcros (I can probably imagine who it is), that is understandable in any such subjective situation. All of the published reviews of the Halcro power amplifiers were very, very good, over numerous reviewers. Stereophile has no reason to recant any of their reviews.
Are the Halcros perfect? Of course not. All of the data I am aware of suggests that they do not meet their published distortion specs by a wide margin. Their distortion is still very, very low, but I still hold it against them if they mislead in their specsmanship.
On average did they get better reviews than over all of the Ayre power amplifiers reviewed? I think yes (notwithstanding the really good reviews of the MX-R - I tip my hat to you on that amplifier).
Anyway, I am less familiar with the other magazines. I can only confine my comments to the way in which you used the information from that one particular review of one Halcro in Stereophile.
What is your view of other high-feedback amplifiers, such as the Boulder?
That there are some bad sounding amplifiers out there that use negative feedback is not surprizing, even in this day and age. Do they sound bad because of NFB or misuse of NFB? Who knows? There are also plenty of AWFUL no-NFB amplifiers out there designed by engineers who think that just because they eliminate NFB they will have a good-souning design. The landscape is pretty random as a result of the large variety of designer philosophy, skill and ignorance, not to mention the mostly un-disciplined subjective reviewing taking place.
Just as you asserted that there is no reliable correlation between good distortion measurements and good sound, perhaps one could also say that there is no such correlation that NFB makes the sound worse.
Indeed, as good as your MX-R apparently sounds (I have not heard it), what evidence is there that it sounds so good as a result of not using negative feedback? Maybe its just a darned well-executed amplifier with great quality parts.
Cheers,
Bob
There's no possible response but to laugh when people say silly things like "the reason it sounds more like real life is because of distortion products"...particularly when the circuits in question have distortion lower than the .1% THD level often quoted as the limit of human perception. As if that weren't enough, there's never any mechanism proposed as to exactly how those nasty, evil 'image enhancing distortion products' just happen to coincide so closely with what's heard when listening to real music in a real hall. This from people who always demand, in the most arrogant of tones, explanations as to how and why distortion causes certain effects--and if you don't provide a mechanism, then obviously the effect can't exist. (And gawds above, if you do propose a mechanism, they want to know why you haven't put every other facet of your life on hold for three years, applied for funding, designed a testing machine, tested fifty amps, done six different types of statistical analysis to prove the correlation between your new measurement and the listening room, submitted a paper or two for publication, yadda, yadda, yadda...)
Care to try again?
The review in question is the one mentioned by Bob Cordell himself in his letter--the Halcro dm58. Pull it up and read it...it's a scream. He keeps saying how much he loves it--all the while saying that it's too forward in the upper midrange and treble (absolutely can't be the fault of the amp...perish the thought).
The entire THD-tells-all position requires that its adherents strain at gnats and swallow camels. It's intellectually dishonest, hypocritical, and easily disproved if only you can bring yourself to sit and listen to two amps with below-perception distortion levels that nevertheless manage to sound different. This immediately places the listener in the uncomfortable position of trying to explain which premise is wrong: The one that says THD is sufficient as an indicator or the one that says that humans can't perceive differences below X%. Given the inability to fact those simple facts, the THD apologist then must fall back on either 'listener bias,' in which case he, himself, is guilty of the same thing (note that the dm58 reviewer is very impressed with the looks of the amp and with the specifications). Then he trots out ABX testing, which inherently discriminates against subtlety and favors anything which makes a favorable first impression...long term listener fatigue be damned. And so it goes--one by one the excuses fall away, leaving only the raw, insistent belief that THD must somehow be sufficient because it seems so straightforward and reasonable. Part of the problem here is a lack of perspective on the history of audio, wherein you find hitherto "impossible" things being proven again and again. Various other distortion mechanisms have been proposed. Some fell by the wayside. Some are difficult to test, compared to THD. Some may yet pan out. Other mechanisms will, no doubt, be advanced in the years to come. But in the rush to grab hold of something, anything that gives results, THD won because it's conceptually simple and relatively inexpensive in practice. There's just this one little, bitty annoying problem...real music isn't steady state. And that leaves the back door wide open to all sorts of other possibilities.
None of which are acceptable to those who believe so fervently in THD as a global solution to measurement problems.
Oh, and SY? Perhaps you're forgetting dither, jitter, LSB problems, too-low sampling rates, etc.? Digital was not ready for prime time when it was introduced. Period. I would argue that it should have been held until it reached level that it finally attained in the mid-'90s, but with the telecommunications industry going digital and storage heading towards digital, there's not a chance of a snowball in Hades that would have actually happen. There's no doubt that recording engineers played their part in the debacle, but they were not the sole problem. Not by a long shot.
Grey
P.S.: I find it ironic that the prompters of Red Book CD standards have now been hoisted on their own petard. They introduced the CD format to boost flagging audio sales. It worked. Only now they find themselves unable to pull the same trick again with SACD or DVD-A because the consumer already bought the "Pure, Perfect Sound, Forever" line the first time around. Swallowed it hook, line, and sinker. So why should they update? They've already got the convenience, and have diminished expectations for quality, so both formats are failing to ignite the sales the industry hoped for.
Care to try again?
The review in question is the one mentioned by Bob Cordell himself in his letter--the Halcro dm58. Pull it up and read it...it's a scream. He keeps saying how much he loves it--all the while saying that it's too forward in the upper midrange and treble (absolutely can't be the fault of the amp...perish the thought).
The entire THD-tells-all position requires that its adherents strain at gnats and swallow camels. It's intellectually dishonest, hypocritical, and easily disproved if only you can bring yourself to sit and listen to two amps with below-perception distortion levels that nevertheless manage to sound different. This immediately places the listener in the uncomfortable position of trying to explain which premise is wrong: The one that says THD is sufficient as an indicator or the one that says that humans can't perceive differences below X%. Given the inability to fact those simple facts, the THD apologist then must fall back on either 'listener bias,' in which case he, himself, is guilty of the same thing (note that the dm58 reviewer is very impressed with the looks of the amp and with the specifications). Then he trots out ABX testing, which inherently discriminates against subtlety and favors anything which makes a favorable first impression...long term listener fatigue be damned. And so it goes--one by one the excuses fall away, leaving only the raw, insistent belief that THD must somehow be sufficient because it seems so straightforward and reasonable. Part of the problem here is a lack of perspective on the history of audio, wherein you find hitherto "impossible" things being proven again and again. Various other distortion mechanisms have been proposed. Some fell by the wayside. Some are difficult to test, compared to THD. Some may yet pan out. Other mechanisms will, no doubt, be advanced in the years to come. But in the rush to grab hold of something, anything that gives results, THD won because it's conceptually simple and relatively inexpensive in practice. There's just this one little, bitty annoying problem...real music isn't steady state. And that leaves the back door wide open to all sorts of other possibilities.
None of which are acceptable to those who believe so fervently in THD as a global solution to measurement problems.
Oh, and SY? Perhaps you're forgetting dither, jitter, LSB problems, too-low sampling rates, etc.? Digital was not ready for prime time when it was introduced. Period. I would argue that it should have been held until it reached level that it finally attained in the mid-'90s, but with the telecommunications industry going digital and storage heading towards digital, there's not a chance of a snowball in Hades that would have actually happen. There's no doubt that recording engineers played their part in the debacle, but they were not the sole problem. Not by a long shot.
Grey
P.S.: I find it ironic that the prompters of Red Book CD standards have now been hoisted on their own petard. They introduced the CD format to boost flagging audio sales. It worked. Only now they find themselves unable to pull the same trick again with SACD or DVD-A because the consumer already bought the "Pure, Perfect Sound, Forever" line the first time around. Swallowed it hook, line, and sinker. So why should they update? They've already got the convenience, and have diminished expectations for quality, so both formats are failing to ignite the sales the industry hoped for.
Charles Hansen said:
Charles, no offence, but you definitely could use some notions of analog IC design.
The analog IC design process is different from what you do today. You optimize biases, they stabilize biases and trying to design bias independent circuits, you use the best discrete devices on the market, they rely on device matching. You scrap your wallet for a few 2SK389 they have them practically for free. You may get 0.1% precision resistors from Caddock at $1 a pop, they are using only +/-20% resistors but always matched to 0.1% (fro free). The set of design trade-offs are different. And so on...
Today, there are a very few limits on what can be integrated on a chip (one of these limits is integrating very high power devices) and within these limits even the best discrete circuits can be redesigned as "integrable" with the same or better performances. It is not happening every day and for every circuit for economics reasons only. The IC industry is true mass production and they need a very good (market) reason in deciding to integrate, say, something with the equivalent performance of Blowtorch, in one or more chips.
Optimizations of a certain design is almost never done at the (number of) masks level. The manufacturing process has to be a common denominator for a whole class of analog circuits. It's the design that has to be optimized, in the frame of the same manufacturing process.
One to another, although certain basic principles and building blocks are shared among the discrete and linear IC designers (e.g. the current mirror, the differential stage, etc...) their jobs are very different.
If you feel like diving into this world I could suggest a few very good and gradual lectures. I am sure this would change your perspective on the analog IC place in audio and you would fully appreciate what National (and a very few others) are doing for us. But if you choose not to, there's nothing wrong with that as well.
May I play Dr. Phil and guess that your repulsion for ICs is ultimately because you somehow feel threatened by their evolution? If so, then rest assured that there's place for everybody. The flexibility and cost effectiveness (for small scale production) of discrete designs are always going to remain major advantages for certain applications, in particular high end audio.
john curl said:
I'm not quite sure what that has to do with the recordings in your music collection, which if I were to guess, would be large. Were they all made with studio boards of your own design? Are they all, or are the majority recorded without any NFB, op-amps, cheapo caps, inductors & resistors?? Is the "grain, harshness, etc., etc." heard by people with Grey's condition embedded or absent? Remember, until I pointed it out, he may not have had prior knowledge of this and therefore not heard it in his recordings (he certainly would not have had direct sight of the equipment making all his recordings). Do you hear it in yours?
Bonsai said:
The issue as I see it, is that there are those whose conditions cause them to hear things like "grain & harshness", when the sight or prior knowledge of NFB, op-amps (insert audiophile anti-purity item here), etc, etc. enters the equation.
It reminds me a bit of those who have seisures at the sight of flashing red lights. Except that those folks most likely realize that not everyone has the same reaction to such stimuli.
For audiophiles OTOH, it is very real to them and they cannot see or understand why others don't have the same reaction. They perform continuous self-analysis of their own hearing (without any outside distractions/intervention) and determine it to be superbly faultless.
Most people don't hear "it" embedded in recordings....or during playback, because "it" does not reside in the soundwave impinging upon the ear any more than a flashing red light does.
cheers,
AJ
Charles Hansen said:
Charles, now how do you know that?
This would seem to be a ridiculous assertion on its face. This is the kind of irresponsible, baseless assertion that gets you into trouble. Have you taken individual IC resistors and capacitors and put them in your circuit instead of quality discretes? I seriously doubt it. Even if you did, how would you know that it was the IC resistor or capacitor as opposed to the use of Aluminum in the bond wire?
What on earth is your basis for this assertion??
Cheers,
Bob
Bob Cordell said:
Excellent! Glad you got to hear them as I'm sure that was probably a matter of great interest to you.
Now, let me be sure I've got this right...
You listened to an amplifier once, briefly, in a hotel room, with what were likely unfamiliar associated components, with all the distractions and noise and commotion, and decided that it 'sounded great.' And from that one listening session under poor conditions you blasted Charles for his letter based (in part) on your assessment of the 'very high sonic quality' of the Halcro.
Did I get that right?
That's pretty good hearing you've got there. I'm envious.
Even seasoned reviewers usually stop at "It sounded promising at the show, but can I have one for extended review in my own system?"
Or am I putting words in your mouth again?
ShinOBIWAN said:
Precisely.
THD ain't the only game in town...or at least it shouldn't be. I don't imagine that any one distortion test will be adequate, regardless of how clever it is. Granted, there's IM, but I'm assuming that it will take a minimum of three or four tests to even begin to get a grip on what something really sounds like. Ten tests? Probably better still, though a potential nightmare for the consumer to wade through.
Let's see...amplifier A did better on tests #3 and #6. Amplifier B did better on #7. Amplifier C did great on #1 and #9, but flunked #10 completely.
At which point the poor consumer throws up his hands and buys a boom box at Wal-Mart.
Can't say that I blame him.
On the other hand, given the grotesque non-correlation between current tests and sound quality (despite apologists' rear guard defense of THD), anything would be an improvement.
I've proposed several tests over the years, most recently an admittedly brutal signal comprised of two different frequencies of square wave to see if a more complicated signal would accomplish anything more than standard sine vs. sine IM. Why squares? Because I'm curious about rapid rise times interacting. Nulling out the original signals and making sense of the results would be an interesting way to spend the next ten years of your life.
I'd like to see a test that gave a good indication of how dynamic an amplifier is. At first blush, you'd think that something like a pulse or square wave at, say, 75% of rated power would be useful. Only it isn't. Perhaps monitoring the rail voltage instead of the output would be in order.
Tightness of low end. There's no correlation between damping factor and perceived low end quality. Why? What are we missing? Life would be so much simpler if a really low damping factor actually told the tale.
I suspect that part of the answer to both of these riddles is current delivery on demand. So you...what?...stick a low resistance current sensor in series with the load and see what happens? Heisenberg rears his head. You just changed the damping factor. And it may not show anything useful, anyway.
What's the answer? I don't know. But it would be nice to shake things up a bit.
Grey
john curl said:
For a good reason, they could be integrated. But, you are in the same trap... If you are looking to directly integrate your designs, the results will be certainly dissapointing. But re-designing to get the same functionality/performance by using linear IC principles and building blocks may lead to the conclusion that complementatary JFETs are actually not required.
True, this means that you have to define precise and measurable performance metrics. Designing linear ICs based on the subjective "sound quality" is way to expensive to be afforded by IC manufacturers. That's another good reason why Charles is safe.
Everything you guys say (GR specifically) in the negative about high negative feedback amps, I tend to feel about, and hear in 'balanced' signal transmission.
So a high negative feedback, balanced signal amp (which is also actually single inside), is an amplifier that I avoid like the plague.
As for negative feedback itself, I'm in the 'local, low level feedback at most' camp. Which I think that most reasonable designers who actually know how to listen to what they create..end up being at, in the long run.
edit:
Oh yes. My little bugaboo: The measurement tests that are used in audio have almost nothing in common with how the ear hears.
Meaning: What you are measuring is almost wholly irrelevant to the ear.
Please study, very closely, how the ear actually works. Then weight the measurments the same way, and the results of the measurements will finally correlate exactly to what the ear hears.
Imagine that.
So a high negative feedback, balanced signal amp (which is also actually single inside), is an amplifier that I avoid like the plague.
As for negative feedback itself, I'm in the 'local, low level feedback at most' camp. Which I think that most reasonable designers who actually know how to listen to what they create..end up being at, in the long run.
edit:
Oh yes. My little bugaboo: The measurement tests that are used in audio have almost nothing in common with how the ear hears.
Meaning: What you are measuring is almost wholly irrelevant to the ear.
Please study, very closely, how the ear actually works. Then weight the measurments the same way, and the results of the measurements will finally correlate exactly to what the ear hears.
Imagine that.
SpittinLLama said:To correct the record, Mark B. was at National years before development of the new LME op. amps or the LME driver stages started. He got to National in late '99. But if you know him you can probably just call him for any info and then you know he loves to talk
If Mark Brasfield was at National since '99, I stand corrected. I was basing my information from the National press release. They made it sound as if they had just introduced a brand-new program with a new director (Mark B.) and a brand new listening room (and surely it is new since the speakers mentioned were not introduced until many years after '99). If Mark Brasfield had been working on these new IC's for years, I wonder why they didn't trumpet that fact.
The significance of his presence is that he founded a small high-end manufacturing company called MSB (his initials). While he was there they focused on high end digital products. See reviews of some MSB products (in chronological order) at:
http://stereophile.com/digitalprocessors/234/
http://stereophile.com/digitalprocessors/276/
http://stereophile.com/digitalprocessors/799/
At some point Mark Brasfield left (or was forced out by outside investors, I don't really know -- that type of situation often happens in high-end audio for some reason). Reading the reviews it would appear the Mark B. designed the first two products but had left the company by the time of the third review -- there is a Manufacturer's Comment signed by "The MSB Team" for that product.
Presumably the idea is that Mark B. has some "audiophile credibility" and that is why National is using his name to promote their products. My recollection is that while Mark B. was at MSB they produced products that were very good value for the money. I don't think they were considered to be contenders for "state-of-the-art". You can read the reviews and see what you think.
SY said:
I could do that very easily. I haven't tried, but I would bet I could get the second down to -80 or -100 dB open loop. The only caveat is that it gets harder and harder to do so as the load impedance drops. So to get those numbers for a power amp driving an 8 ohm load would require a hefty output stage dissipating several hundred watts at idle. But for a preamp that's driving a typical load of 10 kohms or more, this would be pretty trivial with the right circuit and the right parts.
The only question I have is why in the world you need to swing 100 V?
But you are absolutely right. It is *much* easier to design good sounding equipment with tubes. In fact, I would go so far as to say that as long as you don't do something stupid (liking using high-feedback solid-state voltage regulators) it's almost impossible to design bad-sounding tube gear. You can even get away with solid-state rectifiers, although tube rectifiers will generally sound better.
There are only three reasons I don't design with tubes:
a) There is no longer any reliable suppliers of truly high quality tubes. (There are however, fairly reliable supplies of fairly good sounding tubes.)
b) Tubes generate a lot of heat and waste a lot of energy in power amps.
c) Tubes wear out. I'm much too neurotic to be wondering about my tubes. (Should I replace them yet? Will the replacements sound as good as the current tubes? Et cetera, et cetera.)
But for the person willing to put up with the hassles of tubes (I did in my younger days), more power to them. I'm not shy about saying that our goal at Ayre is to make gear that is as listenable and enjoyable as great tube gear, but without the hassle factor.
Quote
My annoyance begins when they proclaim that it sounds better...only they claim they you can't trust your ears because your ears are too easily fooled by listener bias. Huh? I must have missed something--how do you know it sounds better if you can't trust your ears?
Unquote
YOU are the guy claiming NFB makes amps sound BAD. this is a blanket statement.
None of the NFB protaganists on this thread claims it makes amps SOUND BETTER. I don't recall anyone on the NFB side saying you cannot trust your ears. What they have asked YOU to do is provide a rational explanation, and some way of measuring the zero NFB assertion.
I think it has been repeated quite a few times, that NFB is used to IMPROVE ELECTRICAL SPECS.
For the record, I agree (I guess so would everyone on the NFB side) with the idea that amps need to go through a 'voicing' design cycle once the electric design is complete. It may have escaped some people, but amps with feedback can also get voiced - compensation schemes, LT pair standing current, Cdom, power supply tweakling and if one is into that sort of thing, the dropping of a coil or two from the output. But, like the other NFB guys, I don't buy the general assertion that NFB is bad for sound - too much and badly used - yes, no will argue that point.
One other thing: 'feedback' keeps getting hi-jacked into 'high feedback' to give the zero NFB argument more weight by its proponents. Again, few users of NFB would argue for high feedback in itself.
Finally, the Halcro's might sound crap to SOME PEOPLE (because some reviewers still rate them and some people have still kept them). But what about the Krells - 8db of NFB with good reviews (remember: Feedback is feedback).
My annoyance begins when they proclaim that it sounds better...only they claim they you can't trust your ears because your ears are too easily fooled by listener bias. Huh? I must have missed something--how do you know it sounds better if you can't trust your ears?
Unquote
YOU are the guy claiming NFB makes amps sound BAD. this is a blanket statement.
None of the NFB protaganists on this thread claims it makes amps SOUND BETTER. I don't recall anyone on the NFB side saying you cannot trust your ears. What they have asked YOU to do is provide a rational explanation, and some way of measuring the zero NFB assertion.
I think it has been repeated quite a few times, that NFB is used to IMPROVE ELECTRICAL SPECS.
For the record, I agree (I guess so would everyone on the NFB side) with the idea that amps need to go through a 'voicing' design cycle once the electric design is complete. It may have escaped some people, but amps with feedback can also get voiced - compensation schemes, LT pair standing current, Cdom, power supply tweakling and if one is into that sort of thing, the dropping of a coil or two from the output. But, like the other NFB guys, I don't buy the general assertion that NFB is bad for sound - too much and badly used - yes, no will argue that point.
One other thing: 'feedback' keeps getting hi-jacked into 'high feedback' to give the zero NFB argument more weight by its proponents. Again, few users of NFB would argue for high feedback in itself.
Finally, the Halcro's might sound crap to SOME PEOPLE (because some reviewers still rate them and some people have still kept them). But what about the Krells - 8db of NFB with good reviews (remember: Feedback is feedback).
Bob Cordell said:
Well on the one hand, I don't "know" that as I have never listened to a silicon resistor. But on the other hand you are going over the top trying to defend an essentially indefensible position. Two indicators (admittedly not proof, but life's too short to chase this kind of silliness):
1) If silicon resistors are so wonderful, why doesn't anybody make and sell them?
2) Without special processing, silicon resistors will exhibit both non-linear capacitances to the substrate and non-linear resistance values. When the Crystal (Cirrus) CS3310 was released, they bragged about how they solved this problem. This has been dropped from their current datasheet, but the details of this can be found at the following website:
http://www.jeffrowland.com/Technology/DigitalVolumeControls.htm
An excerpt follows:
"The CS3310 uses a special physical and electrical structure to cancel the nonlinearity of its polysilicon resistors eliminating distortion. The depletion zone (the boundary) between the P type silicon and the N type that make up the resistor and its substrate (the supporting surroundings) grows greater in thickness as the junction is reverse biased with a signal. As the depletion zone grows thicker, two undesirable things happen: The capacitance between the resistor and substrate changes (decreases.) Of greater concern, the cross sectional area of the resistor shrinks, thereby increasing the resistance in proportion to the applied signal. This is the principle cause of silicon resistor nonlinearity."
Are we to assume that this non-linearity found in typical silicon resistors is somehow a good thing? Or should we instead assume that *all* modern IC's are now made with this special technology that minimizes the problem and that nobody bothers to mention it in their data sheets and/or promotional materials? (Either proposition seems unlikely to me.)
- Status
- Not open for further replies.