[Raj]

HI Guys,

Okay so we know that your're all trying to reproduce the music and or get to the emotion. At what point do you decide though how much you're prepared to spend on producing an amp giving the price you could sell it for, in terms of how much will you elaborate on design - adding extra power supplies to all requiring stages etc, the very best that the commercial world has to offer is usually presented in this way, do you think that most of this is not really required, or that the differences in quality would just be too small?

I like Hugh's idea of taking the GK-1 into multiple stages, is there any chance of one of you guys coming up with a no holds barred amp?

What about designing the sources also, if your policies are to provide people with musical enjoyment at reasonable cost, then you've got a world class amp with only say an average commercial cd player. Okay so it sounds good, but how can mr average really afford a source worthy of such an amp, unless you as the designers provide a complete setup? I do agree that sources maybe a riskier business venture for you, as we're dealing with formats, and these are subject to change. Certainly amps are a safer bet, giving that all they do is amplify the source, I've got a world class amp and pre, I would like to make the best of it, but I'm not going to spend £5000 on a cd player to do this. There really isn't anyone offering good kits for cd players/dacs, I think cd will be around for a while so how about it, unless ofcourse you're already a few ateps ahead of me.



Thanks
Raj

[Dan Banquer]

Well Raj; I don't know how to define the ultimate amp. I find a lot of this load dependent. Designing to power something like an OHM Walsh 5 or Apogee Diva is a lot different than designing an amp to run a Klipschorn. When I was doing the design work for the LNPA 150 I rapidly came to the conclusion that there were plenty of incredbly high power amps out there that could run things like the Apogee Diva. The great bulk of loudspeakers are no where near as difficult a load, so massive amounts of power just aren't used. I was looking for qualities not found in audio power amplifiers, things like low noise, uniformity, stability, & AC line immunity.
What's your idea of the ultimate power amp?
What's your idea of reasonable cost?
Dollars to Donuts (as we say on this side of the pond) if you asked those two questions to ten different audiophiles you would get ten different answers.

[Raj]

Hi Dan,

Well again a certain amount of my understanding is coloured by hype. What I would call a compromise would be say adding a single power supply per monoblock, when there are setions of the amp that require different rates of current at the same time, surely if one was going all out, each section would have it's own supply, yes one supply per monoblock is seen as optimum but were talking no holds barred here. Well reasonable cost, from a kit point of view the topology has to offer the ability to run from a single power source with the option to upgrade to multiple power sources easily, something along the lines of the GK-1 perhap - from basic, to full on (in certain terms). From a business point of view which one has to be sensative, it would make sense to appeal to both ends of the spectrum offering a kit for say £500 - £700, and also with the option of going all out  for those lunatics like me, albeit at extra cost. Dan you obviously know what you're doing in terms of design, and at some point you must make compromises, certainly on so called audiophile caps etc. I't's not really audiophile caps that make the sytem is it? Good caps help, but you don't have to spend the earth to get them. Oversized power supplies, power filtering, what about good earthing, how important is this to an amp. I'd like to see a fully upgradeable amp, allowing every concievable and viable upgrade to be made easily without having to cut tracks and the ability to understand schematics(obviously the benefits have to be worthwhile sonically for each upgrade), the work is all done by the designer, if you want to make the system better then you invest in the upgrades. And also what about the design and supply of sources to take advantage of the amps?

Thanks
Raj

[DVV]

HI Guys,

Okay so we know that your're all trying to reproduce the music and or get to the emotion. At what point do you decide though how much you're prepared to spend on producing an amp giving the price you could sell it for, in terms of how much will you elaborate on design - adding extra power supplies to all requiring stages etc, the very best that the commercial world has to offer is usually presented in this way, do you think that most of this is not really required, or that the differences in quality would just be too small?

Unlike Dan and Hugh, I do not manufacture and/or sell what I make. Therefore, I am free of any price constraints in that respect, which leaves me with only one price aspect I am extremely fond of - value for money.

How one determines it is very slippery ground, because by simply having different thresholds of price tolerance, one moves the bar up or down. But this is true of any man, not just us.

Off hand, I'd say a reasonably powerful power amp, prividing 50-100 W continuous into 8 ohms, and costing no more than $500 for a stereo version is good value for money (assuming it sounds good, of course). Same price thershold for the preamp -while it doesn't use bulky and expensive components like big power transformers, it would use some better quality components (e.g. ultramatched low noise transistors, etc). These apply for DIY kits, all told (inc. cases, PCBs, etc).

I like Hugh's idea of taking the GK-1 into multiple stages, is there any chance of one of you guys coming up with a no holds barred amp?

What about designing the sources also, if your policies are to provide people with musical enjoyment at reasonable cost, then you've got a world class amp with only say an average commercial cd player. Okay so it sounds good, but how can mr average really afford a source worthy of such an amp, unless you as the designers provide a complete setup? I do agree that sources maybe a riskier business venture for you, as we're dealing with formats, and these are subject to change. Certainly amps are a safer bet, giving that all they do is amplify the source, I've got a world class amp and pre, I would like to make the best of it, but I'm not going to spend £5000 on a cd player to do this. There really isn't anyone offering good kits for cd players/dacs, I think cd will be around for a while so how about it, unless ofcourse you're already a few ateps ahead of me.

Thanks
Raj

The problem here is that you are asking people with no large manufacturing facilities and capabilities to produce an essentially electro-mechanical product. I would rather go the other route - take a proven product (e.g. Marantz CD6000) and tune it up. This would let me concentrate on the essentials of the sound, since the mechanics would have been done for me already.

Cheers,
DVV

[Dan Banquer]

I just love these kind of questions. I would love them more if I was consulting because I could really run up the time here and make a small fortune.
The parts cost for a set of LNPA 150's is more than the 700 pound figure you mentioned, and we haven't started with the audiophilia yet. I will pick just one of the items you mentioned and take it from there.
   Audiophile caps: From my experience they love to come in different sizes than standard caps and different lead thickness too. So PC board layout now becomes very problematic. Given the present trends where new "audiophile" caps come out yearly, how do you manage to anticipate further changes in the PC board?
BTW I have tried Black Gate Caps once. I will not be recommending them in the future. Fully regulated power supplies negate the "sound" of capacitors in this application.
Raj; Can you understand where I am coming from here?

[Raj]

Dan, thanks for your patience with me

I can see what you're saying about caps etc, I know that as soon as the new 'wonder cap' comes out idiots like me will want them in my system, I won't fall prey to this kind of idiocy again! Given what I understand about the functions of an amp, isn't a large degree of it how well and cleanly the power is supplied, maybe what I'm talking about just seems to over the top to be viable, how many people would want it (individual dedicated supplies for each section etc). I''m not talking about having the best amp in the world but an amp with few compromises.
That's one of the reasons why I bought my current kit cause of the flexibility.


Coincidentally just so this doesnt go the wrong way Hugh Dan and Dejan, what's your favourite music/tracks, meaning tracks that have actually moved you emotionally maybe even to tears etc (not just sad or happy music, but something really touching)?

Thanks Raj

[Dan Banquer]

The list is to numerous to even begin to mention here. In times such as these a version of Amazing Grace will usually put a tear in my eye.
Raj; go back over the this thread and Tubes vs. Solid State for more info on power supplies.

[Raj]

Yes finally!!!


But just a few more (will help you dream of the money you could have made out of me as an advisor Dan)

A couple of questions for the aid of those who are about to spend a great deal of money on amps etc

How many commercial amps have you seen that you'd actually regard as value for money?

At which point in the dollar region do you think that hype is really taking over?

Dejan's already offered his view on cost etc.


Ahh What I want doesn't exist anyway!!!!

[blizzard]

Hi All,
  Outstanding topics!  Unforunately, I only came upon it yesterday.  I tried to catch up.  I think I caught most everything, but this thread is huge.  Sorry if this question was answered (I didn't see it if it was).
   I know DVV touched on class A biasing earlier.  However, I have some more questions:
  1.  I never quite understood how biasing a transistor actually worked.  Let's use the final stage (a current output pair) as an example for my question.  When you set the bias, you're making the output transistor conduct a current (more curent for class A and less current for class A/B) at no signal.  My question is, doesn't this current translate to DC on the speaker?  Isn't this current goint right to the speaker load?  Obviously it doesn't and I'm missing something.  Could someone please explain.
  2.  My next question is about the difference between class A and class A/B.  If when you bias a transitor, you are making it draw current.  When would it ever be off in a class A/B.  Even if the bias current is ever so small, it is coducting.  So, it never seems to switch off.
  3.  Most (if not all) class A/B amps seem to have a class A region at very low outputs.  Again, this question is related to the questions above.  Specifically, how does this happen.
  These 3 questions are very similar.  I am obviously missing some basic point.  
I apologize if these questions were already answered.  I tried to read the entire thread before I asked.  But, I didn't find my answers.
        Thanks,
            Steve

[DVV]

Hi All,
  Outstanding topics!  Unforunately, I only came upon it yesterday.  I tried to catch up.  I think I caught most everything, but this thread is huge.  Sorry if this question was answered (I didn't see it if it was).
   I know DVV touched on class A biasing earlier.  However, I have some more questions:
  1.  I never quite understood how biasing a transistor actually worked.  Let's use the final stage (a current output pair) as an example for my question.  When you set the bias, you're making the output transistor conduct a current (more curent for class A and less current for class A/B) at no signal.  My question is, doesn't this current translate to DC on the speaker?  Isn't this current goint right to the speaker load?  Obviously it doesn't and I'm missing something.  Could someone please explain.

Essentially, the common way of doing is to turn up the wick on the simlated zener diode (the bias spreading transistor) and let the entire current stage work with higher bias current. While this does work, it has a drawback - bias is usually adjusted for least distortion, so by turning the wick up, you also increase distortion. Perhaps not much, perhaps even insignificantly, but you do do it.

A better way is to design for it. This can be done in many ways, my favorite being to use a predriver stage, then a driver stage and lastly the output stage proper. All I need do then is add two resistors from the opposite supply line to the emmittor of the driver, calculating the resistor so that it literally injects some current into the output stage. And presto, I have a "Current Injection System", or CIS, as I would print on the front plate of my product if I was a Japanese manufacturer. Just under my Magnasonic Hyperphase logo.

Nothing new, Otala did it way back in '72.

  2.  My next question is about the difference between class A and class A/B.  If when you bias a transitor, you are making it draw current.  When would it ever be off in a class A/B.  Even if the bias current is ever so small, it is coducting.  So, it never seems to switch off.

Depends on how you bias it. I prefer larger than standard, but far from full class A bias regimes.

  3.  Most (if not all) class A/B amps seem to have a class A region at very low outputs.  Again, this question is related to the questions above.  Specifically, how does this happen.

If you bias a circuit at say 150 mA, and have four output pairs, your total output current will be 600 mA. This is about 2.9/5.8 W into 4/8 ohms, so up to those points your amp will work in pure class A, and above that, it will operate as a standard class AB amp.

Cheers,
DVV

[Raj]

Haven't seen much on resistors yet, I've read something bout this on your website though Dan, the experience of bulk foil resitors and the 10db cut in noise, certainly interesting. How much do resitors affect the sound though, say a bulk metal foil resitor, which is certainly expensive, does it offer superior performance at all frequencies?



If you read sites like AA, you'll find contradictions in 'heard' sound, what's the truth in measurment terms of what's happening here?

Is it merley that an 'upgrade' like a superior resitor allow more info thru, and with say a system which still has 'weak' components in areas crucial to the supply of these resitors the percieved sound is not the whole truth of what that resistor is capable of?

Are these percieved differences cos some resistors are better at certain frequencies, even though they claim to be unconditionally superior to other types?

What about resitor tolerances, any effects of using tighter tolerences in key or all areas?

Certainly you three dudes are adept at reading between the lines on these posts and also implying what you deem needs to be implied!

Thanks
Raj

[DVV]

Haven't seen much on resistors yet, I've read something bout this on your website though Dan, the experience of bulk foil resitors and the 10db cut in noise, certainly interesting. How much do resitors affect the sound though, say a bulk metal foil resitor, which is certainly expensive, does it offer superior performance at all frequencies?

If you read sites like AA, you'll find contradictions in 'heard' sound, what's the truth in measurment terms of what's happening here?

Is it merley that an 'upgrade' like a superior resitor allow more info thru, and with say a system which still has 'weak' components in areas crucial to the supply of these resitors the percieved sound is not the whole truth of what that resistor is capable of?

Are these percieved differences cos some resistors are better at certain frequencies, even though they claim to be unconditionally superior to other types?

What about resitor tolerances, any effects of using tighter tolerences in key or all areas?

Certainly you three dudes are adept at reading between the lines on these posts and also implying what you deem needs to be implied!

Thanks
Raj

Raj, you need to relax a bit. Look at your questions above, and it turns out you want a cracs course in electronics which requires a staff answering your questions. But the problem is, too many questions in one go implies quick'n'disrty answers, so you end up with superficial understanding.

First of all, don't ask us here about other sites. We do what we do here, we do it our way, and they do it their way. I don't think too many of us have the time to read out all the sites out there, especially the more voluminous ones.

Secondly, yes, it is important that you use tight tolerance resistors at least in some critical places. Typical examples would be the negative feedback network (where this will reduce gain mismatch between channels, of CRITICAL importance in multichannel audio, from stereo upwards), or differential transistor pair resistors, in particular the emmittor degeneration resistors (a mismatch here will produce DC drift and offset).

Third, using better quality resistors ALWAYS results in improved performance, but how much and in which way, depends on several factors, from circuit topology, to the resistors themselves. I have not tried bulk types, I plan on trying them out after Dan's firm recommendation, and at this time, I use Roe/Vishay metal film types, 1% in less critical places, 0.1% in critical places.

Four, a resistor, any resistor, is a part of the circuit. As such, it will by default influence the sound of the whole. This simple fact of life is now being grossly misused, as is just about everything else, to tell tales of myth and magic to the unsuspecting public. I have seen it written on Internet that all this metal film, foil, etc is just bunk, the REAL sound comes only from carbon types, which have a "more relaxed", "warmer" sound. Not to even mention stories how products from company A are vastly superior to those from companies B and C.

The danger from this sort of Chinese whispers launched over the Internet is not in that they are lies - if they were outright lies, they would be easy to debunk. No, the problem is that they all use half-truths and then build on them for desired effect. Very much along the ABC of special warfare, and make no mistake, it IS warfare because the audio/video industry is worth well over 300 BILLION dollars per annum. Since the cake is big, so are its slices, and therefore the trench warfare is ugly as hell.

What can you do about it? Simple - THINK. Do not accept anything, from anyone (and by all means, do start with me), at face value, instead ask around, go to totally different places and ask again, or better yet, try it yourself on a small scale. Take it from me, I have been let down so many times so badly by so many hyped up ultra, super-duper, drastic plastic products, I trust no-one at face value any more. The ONLY real way is to try it out yourself.

And you will notice that, for example, Elna filter capacitors will give you bass like few, if any, others, but will shortchange you elsewhere. Nichicon I find much better balanced, but somewhat lacking in force (so to speak) somewhat. Panasonic I find so mediocre I wouldn't dream of sticking it in any better product - though they are very reliable. In my view, Germany's Fischer&Tausche produce the best commercial quality caps, and some which sit on the fence between medium and high class. For true high class, I go for Siemens Sikorel series - very expensive, but as good as any I have ever seen, with balance, power and speed all rolled into one as nowhere else. On the other hand, there are a few I haven't tried, so my findings could change.

Cheers,
DVV

[Raj]

Hi DVV,

I didn't use the name of AA as a negative reference, I was merley implying that there's loads of different opinoins on the same product, and trying to show whats happening from a designers point of view. Sorry about posting too many questions though. Certainly what youve said here though DVV will help people not to fall victim to all this hype, which is of course one of the keys to this thread.

Thanks
Raj

[blizzard]

Hi DVV,
  Thanks for the detailed response.  But, my question was a little more basic.  Let me ask it as simply as possible.  The transistors are electronic valves which allow current to pass from the collector to the emitter as a function of the initial bias and the signal voltage at the base.  
  1.  At idle (no signal) there is DC current flowing.  Where is this current going?  How does it not reach the speakers as DC?
  2.  Later, when there is an input signal, the base gets an AC voltage (the signal) and this translates to an audio signal (AC current) to the speakers.  How can the initial bias current not reach the speakers, and the current generated when a signal is present does?
  3.  Class A vs Class A/B:  My question here really is more about crossover distortion.  To keep the same example, lets say bias is set to 150mA.  At the zero point in the audio ac signal, when the one transistor supposedly turns off and hands off to the second device, is it ever really off?  Being that there is a bias current flowing at the zero point, isn't the transistor on.  So, where is the crossover distortion/switching noise coming from?
     Thanks again,
           Steve

[DVV]

Hi DVV,

I didn't use the name of AA as a negative reference, I was merley ...

Didn't take it as such, never fear. Each forum has its "customers".

implying that there's loads of different opinoins on the same product, and trying to show whats happening from a designers point of view. Sorry about posting too many questions though. Certainly what youve said here

Nothing to be sorry about, just take it a little easier, if you want some real answers. Also, I assume you also need time to assimilate what you have just read, and possibly enlrage or clarify this or that.

though DVV will help people not to fall victim to all this hype, which is of course one of the keys to this thread.

Thanks
Raj

I hope so. All I'm saying is THINK, don't take anything at face value, because there are far too many vested interests, all of which see you as the sheep to fleece. So, walk softly, ask, and above all, think.

Cheers,
DVV

[Dan Banquer]

Hi Raj;
         I'm a liitle pressed for time today so bare with me. First is that DVV made a lot of good points. Tolerance and Temperature Coefficents can make a difference in critical areas. I find for most applications I will use mil standard RN55C, 1% tolerance from -55C to 125C, temp coefficent at 50 ppm. Matching for diff amps as DVV pointed out is helpful. The Vishay Bulk Metal Foil resistors have noise figures close to an order of magnitude better than metal films. The measurements confirm that. The vishays also have advantages at higher speeds due to their very low inductance. This is not a factor for metal films at audio band. It is a factor for metal films at RF. The Vishay's if I remember correctly have a lower temp coefficent but I don't remeber the numbers off hand. They are also damned expensive.
Hope this helps.

[DVV]

Hi DVV,
  Thanks for the detailed response.  But, my question was a little more basic.  Let me ask it as simply as possible.  The transistors are electronic valves which allow current to pass from the collector to the emitter as a function of the initial bias and the signal voltage at the base.  
  1.  At idle (no signal) there is DC current flowing.  Where is this current going?  How does it not reach the speakers as DC?
  2.  Later, when there is an input signal, the base gets an AC voltage (the signal) and this translates to an audio signal (AC current) to the speakers.  How can the initial bias current not reach the speakers, and the current generated when a signal is present does?
  3.  Class A vs Class A/B:  My question here really is more about crossover distortion.  To keep the same example, lets say bias is set to 150mA.  At the zero point in the audio ac signal, when the one transistor supposedly turns off and hands off to the second device, is it ever really off?  Being that there is a bias current flowing at the zero point, isn't the transistor on.  So, where is the crossover distortion/switching noise coming from?
     Thanks again,
           Steve

Steve, you just asked for a complete theory of operation for semiconductrs in a few short sentences. This is something one usually needs years to understand, and after 34 years, I'm still not sure I understand it completely.

But I love challenges. Let's give it go.

Ad 1. The DC current is being turned into heat, and as for speakers, it's being cancelled by the negative feedback loop or loops, with or without a DC servo. This heat is essentially useless in terms of zero signal situation, but is highly desirable precisely because of the heat. With zero signal, practically all of it is turned into heat, which of course heats up the output devices and the associated heat sink. If it weren't there, they'd be cool. When the signal does come, and with very little bias, output stages will start to heat up and change their operating parameters. If biased high, this will also happen, but at a considerably greater signal level.

In other words, a high bias heats up the output stage presumably to a point where there is a greatly reduced thermal effect problem, since the devices are already heated up and will operate more smoothly, with less change.

Ad 2. Again, the AC voltage signal is superimposed on the DC signal, which is thus turned into an AC signal of greater overall magnitude. The residual DC is cancelled out by the feedback (local and global, if any) and the DC servo, if present.

The capacitor after the shunt resistor in the feedback voltage divider really simply increases the amount of negative feedback at very low frequencies, making the amplifier less susceptibe to DC offset. However, no matter what you use and how, some DC will always be there. In my experience, with a properly implemented DC servo, I can keep this to within +/- 1 mV, which is low enough to be negligible.

Ad 3. One of main advantages of high bias is that the crossover distortion is greatly reduced. As you quite rightly stated, the word is "reduced", not eliminated. Some residuals will still be there once the amp leaves the region of pure class A operation, i.e. when it goes into the classic AB mode. In the example I used, once the power demanded of it exceeds 2.9/5.8 W into 4/8 ohms.

However, please bear in mind that this is NOT a linear function. Crossover distortion is different at different power points, despite fundamental similarity of its very nature. It will not be the same at say 0.1W and say 10W and say 50W. For a start, at higher power levels, it will be better masked by the signal. Also, since high bias will produce smoother temperature operating conditions, the nature of crossover distortion will also change somewhat.

On the other hand, the advantage of pure class A operation is in the fact that crossover distortion is reduced to a practically speaking zero value, however at what I feel is a tremendous cost (heat sink requirements increase by a factor of 4 or more, power supply requirements also increase tremendously, etc, all of which translates into astronomical prices soon enough).

Cheers,
DVV

[DVV]

Hi DVV,

I didn't use the name of AA as a negative reference, I was merley implying that there's loads of different opinoins on the same product, and trying to show whats happening from a designers point of view. Sorry about posting too many questions though. Certainly what youve said here though DVV will help people not to fall victim to all this hype, which is of course one of the keys to this thread.

Thanks
Raj

Like I said, the problem is not in your asking questions, the problem is in your asking too many questions too quickly. To quote G.B. Shaw: "Questions are never indiscrete, answers sometimes are."

Quite simply, it becomes impossible to answer them meaningfully after a point, and this presents the danger of dragging this forum into precisely those waters we want to avoid.

So, a little patience, and we'll get to the bottom of it, never fear.

Cheers,
DVV

[blizzard]

Steve, you just asked for a complete theory of operation for semiconductrs in a few short sentences. This is something one usually needs years to understand, and after 34 years, I'm still not sure I understand it completely.

Hi Dejan,
  See, that wasn't so bad was it?  
  So, just to emphasize (or clarify):
  1.  If you set some initial bias on the output transistors and didn't use any feedback or DC servo, you would have speaker damaging DC getting through.  Am I correct in making this statement?  Aren't some amps designed without feedback?
  2.  

Ad 3. One of main advantages of high bias is that the crossover distortion is greatly reduced. As you quite rightly stated, the word is "reduced", not eliminated. Some residuals will still be there once the amp leaves the region of pure class A operation, i.e. when it goes into the classic AB mode. In the example I used, once the power demanded of it exceeds 2.9/5.8 W into 4/8 ohms.

  So, exactly what is crossover distortion? I had always heard that it was the distortion of a device turning on and off.  But because of a bias setting, the device is never really off (in either class A or class A/B).  Knowing this, I would then assume that crossover distortion doesn't have to do with a device going from conducting to not conducting.  But rather has to do with a device going from hadling its portion of the audio signal to not handling its portion of the signal.
  If I am correct in my previous statements.  I am still confused as to how class A operation produces such small amount of crossover distorion as compared with class A/B?  In your example above

once the power demanded of it exceeds 2.9/5.8 W into 4/8 ohms

in the low power range (class A), why are the lower power outputs yielding less crossover distortion?  The devices are still handing off to one another.
  Sorry for the somewhat repetative nature of my questions.  This is a subject that I have been contemplating for some time, and have never really been able to get a handle on.  Your fantastic explanations have gotten me closer than ever to full understanding.  One more little step, and I think I got it.
    Thanls a lot Dejan -- so far awsome,
           Steve



[/quote]

[DVV]

Steve, you just asked for a complete theory of operation for semiconductrs in a few short sentences. This is something one usually needs years to understand, and after 34 years, I'm still not sure I understand it completely.

Hi Dejan,
  See, that wasn't so bad was it?  

You think so? Look below ...

  So, just to emphasize (or clarify):
  1.  If you set some initial bias on the output transistors and didn't use any feedback or DC servo, you would have speaker damaging DC getting through.  Am I correct in making this statement?  Aren't some amps designed without feedback?  

No, you are not. As far as I know, it is not possible to make any amplifier, of any sort, without feedback. It's just a matter where and how you apply it. "Zero feedback" does not exist, it is a flippant and incorrect way of saying "no overall feedback", but to have it work, you had to apply local feedback.

Let me put it this way - a typical active device (tube, transistor, etc) has a certain gain (amplification) factor under certain circumstances. This can be anything from say 20 to well over 10,000. Now, a typical say 50W amp will need say 1V of input signal to make it to 50W, which is 20V continuous, and therefore an overall amplification factor of 20:1. Tie up any two transistors or tubes, and you will have an amplification factor of at least (20x20) 400:1, even up to several million to one. Furthermore, your distortion would be record beating at this level. So you have to reduce both gain and distortion. Reducing gain will also reduce distortion, but only up to a point, so after that point, you either feedback it, or you end up with unholy muck on your hands, good for nothing. Less gain, but still good for nothing.

So you apply feedback. It can be local or global, but the best way is to use both - judiciously. But we're moving into the feedback theory now ...

Anyway, however you do it, DC drift will also be reduced. With proper design and execution care, DC offset will be low, typically below 50 mV with the least measures taken. Even in the dark days of audio, it was rarely over this value, except on the junkiest of equipment.

  So, exactly what is crossover distortion? I had always heard that it was the distortion of a device turning on and off.  But because of a bias setting, the device is never really off (in either class A or class A/B).  Knowing this, I would then assume that crossover distortion doesn't have to do with a device going from conducting to not conducting.  But rather has to do with a device going from hadling its portion of the audio signal to not handling its portion of the signal.

You almost got it. Crossover distortion is really caused by a miniscule lag, a time gap which appears when the NPN device stops conductiong and the PNP device has not yet started conducting (or vice versa). What you refer to above is called switching noise, which is broadly speaking the random noise created by a device when it changes states, from conducting to non-conducting and vice versa.

Crossover distortion is a necessary and an unavoidable evil of any push-pull configuration. However, it can be reduced by various techniques to really academic levels, the only problem being the price of doing it.

It depends on many things, from mode of operation, to the speed of the devices you are using.

  If I am correct in my previous statements.  I am still confused as to how class A operation produces such small amount of crossover distorion as compared with class A/B?  In your example above

once the power demanded of it exceeds 2.9/5.8 W into 4/8 ohms

in the low power range (class A), why are the lower power outputs yielding less crossover distortion?  The devices are still handing off to one another.
  Sorry for the somewhat repetative nature of my questions.  This is a subject that I have been contemplating for some time, and have never really been able to get a handle on.  Your fantastic explanations have gotten me closer than ever to full understanding.  One more little step, and I think I got it.
    Thanls a lot Dejan -- so far awsome,
           Steve

Very simple, Steve. In class A, the devices are always in their ON state, they conduct their full current all of the time irrespective of whether there's any signal at all or not. Since they are operating on a constantly open basis, and do not switch on and off, their switching distortion is naturally reduced to almost nothing.

If both devices are in their full ON state, there will be no time lag when one switches off and before the other switches on. This is the key benefit of full class A operation, everything else are its downsides.

Remember that both switching and crossover distortions are compunded in amplifiers which use multiple pairs of output devices, because one simply cannot match perfectly even several NPN devices among themselves, let alone each of them with a corresponding PNP device. Of course, there are techniques to overcome this problem as well, believe me, I've been using no less than 2 pairs of output devices per channel for over 20 years now, because this approach has its tremendous virtues as well as downsides.

Lastly, both problems are really appearent only at low level listening, because both tend to be masked by music. In other words, they don't go away at high power levels, they are just much less obvious as such, but can still cause the loud sound to be smeared and muddy, for example. This is why I use the high bias approach - incidentally, the 600 mA figure I quote is "my" sweet number, that's what I always go for - this all but kills the problem dead at critical low levels, but without the extreme penalties of full class A operation.

A year and a half ago, I wrote a text published on TNT (http://www.tnt-audio.com ) titled "Put A Tiger In Your Amp", in which I described this kind of tuning mass produced audio gear, with some nice shots of the carnage I wreaked in my own amps. This is an easy tweak, and one that becomes obvious inside of 30 minutes after doing the evil deed. But it never ever fails. Point made.

Cheers,
DVV