Need an Amps Classification 101 Tutorial

[thebes]

Ok. In another of my continuing series of roll your eyes and answer the poor fools question attempts to educate myself in all things stereo I raise today the subject of amplifier classifications.

From the Audio Answers section of those fine folks at AudioAsylum, I have extracted the following:

What is Amplifier Class A? What is Class B? What is Class AB?

What is Class C? What is Class D?

All of these terms refer to the operating characteristics

Of the output stages of amplifiers.

Briefly, Class A amps sound the best, cost the most, and are the

Least practical. They waste power and return very clean signals.

Class AB amps dominate the market and rival the best Class A

Amps in sound quality. They use less power than Class A, and

can be cheaper, smaller, cooler, and lighter.

Note: Tried to but couldn't attach the entire section on this issue in respect of the authors copyright.

Lets pass on C and D amps right now. What Im trying to figure out in common English is why class A would generally sound better than an A/B, why they cost so much more and what da hell is this talk Ive been hearing about class T.

Also Ive seen passing reference to this is a Stereophile Class A recommendation. Huh? Is this the same thing? Picked up a copy of Stereophile once and the charts and graphs gave me a monster headache. What gives?

Need our resident sages to weigh in on this one. After all, (optimistically) I cant be the only dolt here who doesnt fully understand this stuff.

[scriven]

Your not!

[lynnm]

I did a Google search for amplifier classes and got a large number of hits. Here is a good short summary:

http://hometheaterhifi.com/forum/showthread.php?t=432

[artto]

It basically boils down to a thing called crossover distortion. The point at which the waveform crosses 'zero' from positive to negative and back again. Class A amps are biased to have quiescent current bias for the full cycle of the the waveform. Class AB amps have enough 'bias' so that there is quiescent current flowing at the waveform's zero crossing point all the time, but not for the maximum output of the waveform cycle if no current is flowing (hence more efficient). There are some amps such as the Luxman MB3045 that have a design topology that basically eliminates crossover distortion in a Class AB design, yielding Class A performance however these amps have transformers that are very difficult & expensive to make, but they do provide the best of both worlds. The class AB designs are also classified into two catagories, class AB1 and lass AB2. Again t his has to do with how much quiescent current is applied during no or low level signals.

"quiescent"=marked by inactivity, at rest

[Deang]

In adjustable bias amplifiers, if the tubes are a close match, whatever crossover distortion that exists is inaudible. In fixed bias products, it's a complete non-issue if the tubes are biased correctly.

All SET amps are Class A, and some push-pulls are Class A as well. The marked characteristics of Class A amplifiers are their very smooth and open sound. For reasons I can't explain, my experience is that good Class AB amps seem to be more dynamic and "alive" sounding. Things come to your ears more like waves rolling onto the beach, as opposed to a slow rising tide.

[79 Scottsdale]

----------------

On 7/22/2004 9:06:29 PM DeanG wrote:

In adjustable bias amplifiers, if the tubes are a close match, whatever crossover distortion that exists is inaudible. In fixed bias products, it's a complete non-issue if the tubes are biased correctly.

All SET amps are Class A, and some push-pulls are Class A as well. The marked characteristics of Class A amplifiers is their very smooth and open sound. For reasons I can't explain, my experience is that good Class AB amps seem to be more dynamic and "alive" sounding. Things come to your ears more like waves rolling onto the beach, as opposed to a slow rising tide.

----------------

Regarding that first paragraph: So you're saying that whatever crossover distortion may exsist would be inaudible IF the amplifier had fixed adjustable bias, and matched tubes?

A few questions for you to think about:

(1) What is the first stage just directly in front of *almost* every push-pull tube amplifier?

(2) What type of circuit is being used as the stage mentioned in question 1

(3) Is the primary winding of the output transformer matched turn for turn, and are the primary halves equal in impedance and DCR?

(4) How much feedback is employed, and what is the purpose of inverse feedback? What would happen if no feedback was employed?

(5) Are the tubes matched for transconductance and Ik?

Your second paragraph is subjective crap, but we do have something in common, we both prefer push-pull amplifiers.

[WMcD]

Well, I'll give a try.

All of this starts with the basic fact that tubes or transistors only conduct in one direction. They have some sort of input control lead to which the input signal is connected. This is usually the grid (tube), the base (bijunction transistors), or the gate (field effect transistors or FETS). This controls the amount of current the device conducts.

The British call tubes, "valves". This is fairly well decriptive in that the input lead is the "handle" on the valve. The power supply provides the energy, or water under pressure, and the rest of the "valve" simply lets more or less water flow through the device to the output.

Of course in the analogy we don't have your hand on the handle of the valve. Rather, it is the input voltage or current which can open or shut the valve.

Solid state devices basically do the same thing and should be called valves too, though they are not.

An important concept is that the valve does not really "amplify" the input to make the input signal bigger by some magical enlarging machine. Rather, the input allows more or less current from the power supply to go to the output. So we are listening to the power supply as controlled though the valve.

And continuing, one water valve and water source can only work to vary output between zero and some plus value. We can't suck water back into the system by turning the valve handle the the other way. At least with one valve and power supply alone.

As we'll see below, we can wire an additional amplifer and power supply backwards to make it negative, and switch it on at the right time.

This "only works one way" is a bit of a shame in that our music input signal is an a.c. signal which does vary between, say plus one and minus one volts. (The guitar string moves back and forth, the air moves back and forth, the microphone diaphragm moves back and forth, it makes an electical signal going plus and minus.)

We want the amp output to mimic that so that it can move the speaker diaphragm back and forth, not just zero to plus, or zero to minus.

There are a many work arounds. Such is the history of amplifier design. These all originate with the one way valve.

Consider if you have only one "valve" device.

On the input end, we can bias the input signal voltage with an additional, constant voltage or current. We'll keep in the world of solid state (Mosfets) with low voltage and say the bias is 13 volts. You see a lot of talk about "bias". This is the concept.

Now the signal at the gate (input lead) goes between, say 12, volts and 14 volts. 13 volts is the equivalent of zero, and we're driving it plus and minus 1 from there.

This means the valve device is ALWAYS conducting some current from the power supply. This is so even when there is no input signal before the bias. This is called "class A" operation of the valve. It is inefficent in that at no input, some conduction is going on.

We have to continue though. Suppose the overall voltage gain of the valve and power supply combination is 10.

That means the output voltage is varying between 120 and 140 volts. (This is might not be typical in the real world, but let's work with the simple numbers.)

We can't send that type of signal to the speaker. We need an additional device at the output to couple to the speaker so that there is a zero voltage or current output when the overall input is zero.

One simple one is a capacitor. The current through the capacitor is the CHANGE of voltage across it. So it responds to the change between 120 and 140 volts (or 20 volts). The simple class A, "Zen" amps by Nelson Pass use this.

In the vacuum tube world, we use an output transformer. The output of transformer (secondary windings) is related to the CHANGE of current through the input (primary windings).

Now you may see a tiny little problem. These output coupling device respond to not the value of the output of the valve, but instead the rate of change, or slope.

Now that seems like a bad thing, but it is not . . . if you consider that complex music is made up of simple sine waves. If you look at the slope (rate of change) of a sine wave, you get the same wave shape. Except, it looks like it has been shifted by 90 degrees.

Let's do some pre-calculus here. Don't panic, look at the numbers.

The VALUE of the sine at 0 degrees is 0. At 90 degrees it is 1. At 180 degrees it is O. At 270 degrees it is -1. At 360 degrees it is 0.

The SLOPE of the sine at 0 degrees is 1. At 90 degress it is 0. At 180 degrees it is -1. At 270 degrees it is 0. At 360 degrees it is 1.

This last matter gives some people the heeby geebies. Here it is not really a time delay (though in complex filters, there can be time delay). Rather, we've avoided looking at rate of change of the signal (it is calculus), and substituted the simple phase shift principle. See how the numbers of value and slope are the same, except for the 90 offset?

But back to the orignal issue. The valve being always on, even with zero input, is creating a demand on power from the power supply. And we're forced to bias the valve ON all the time. There is power disapation without input. We'd rather have that for peaks in the music.

There are work arounds to this too. A simple one is the push-pull set up of vacuum tubes. This gets really complicated. However, now we have two tubes and, essentially, two power supply voltages (a "relative plus and a minus), and an additional input device (phase inverter).

Essentially we have two amplifiers. When the input signal goes positive, the valve connected to the positive power supply turns on. When the input signal goes negative, the valve connected to the negative power supply turns on. These push and pull current through the primary winding of the output transformer. Hence, "push-pull" operation.

So you can see that at zero input voltage to the complicated push-pull topology, there is no conduction in either valve. This is good for power consumption. This is essentially, class B. And looking at one valve in the push pull, it is ON for only half the cycle of the input.

There are a lot of variations on this idea and thus a lot of "topologies." In some, we can bias the valves so that they conduct though more than half of the cycle of the sine input. Too complicated. Meaning: I don't understand all of it well enough to explain. This, though, is class AB operation of any one valve device in the very complicated amplifer box.

Lets go back to what is potentially bad about class B or AB. It arises, in one instance, in low levels of input signal.

Consider the push pull. When the driving signal switch over in plus and minus to turn on the the pushing amp to the pulling amp, and vice versa, things are not exact. We can't really be sure they are turning on and off exactly correctly.

But, on the other hand, in the class A operation with a single device, it is always ON, or at least always partially on. So there are no issues of trying to get two strings of device to work in harmony.

This is an important concept. Class A may have its power consumption issues, but it does not require another amplifier beside it to handle the rest of the sine wave. We have our issues in one basket. Many say it works better this way.

There are many other issues. One is that the electronic valves are not actually linear in their response to the input. E.g. the output is not an exact multiple of the input. It works on a non linear curve (redundant-a line -- linear is not curved), which causes distortion.

One thing which can help is a negative feedback loop (which IS linear).

This is sort of like driving a car with one foot on the gas (the raw gain, or open loop gain, or non feedback gain) and the other on the brake (the negative feedback loop). The output signal is fed back to mix with the input, but negative in magnitude to cut down on the combined input signal.

This analogy requires that your "lead foot" on the accelerator gets the car going faster than your brain (the valve) would do if perfectly "linear". But, through the feedback loop (which IS linear), it also steps on the brake in exact proportion to the to the lead foot, with its inaccuracy. So there is a type of correction.

As you might imagine, this requires big horsepower (large open loop gain) and big brakes (big feedback levels in the loop). And they are fighting each other all the time to find a linear relation between input and output. And maybe they don't. Contrary to the "self correction" theory, the feedback loop creates some subtile problems and actually may create some forms of annoying distortion.

Nelson Pass has a theory that this feedback solution is a bad one in that it requires more and more gain and more and more anti-gain (my concept) feedback corrections when we should have done it right in the first place with a somewhat linear set up. He uses very high biased class A hexfets and gets good reviews.

Sorry if that went too long. But I've not seen anyone else setting it out. You should look at the Pass Labs page.

Best,

Gil

[Deang]

Yikes Gil.

Scottsdale -- I can't answer your questions, they are out of my realm. As far as "subjective crap" goes, well, what can I say -- it's pretty much all subjective crap anyways. I still don't think Class A sounds as lively as Class AB (at least as far as tube amps are concerned).

[scriven]

Thanks Gil! (I think I even understood it!)

[Soundthought]

I'm always learning something from you, Gil.

Thanks for breaking that down for us.

Regards,

John.

[NOSValves]

79 Scottsdale = Ryan Inman 3469th

[JohnWhite]

Wow, Gil, Thanks for posting that! That is by far one of the best explanations I've read so far. Although I started to get a headache reading it I've printed it out to re-read.

[richinlr]

Well, not being a very learned about amplifier design (SS or tube) I can at least bring up something that hasn't been brought up yet.

There are also amps out there that run Class A at low output and then change over to AB at some point when out put increases. I have such an amp. Imagine a 1khz sine wave on a graph. There is a zero line but the sine wave is not crossing over the zero at any point because the entire wave is positioned above it. As I understand it this is class A operation. In the case of my power amp, it runs here at low output. As the output increased, the amplitude (space between peaks and valleys) of the sine wave gets large enough for the valleys to begin dropping below the zero line. Again, as I understand it, this is when my amp crosses over to class AB.

There was a mention that low feedback limits frequency response. This is not always the case, depending on the design. My amp has NO overall feedback and only enough feedback (very low) on individual stages to make it stable. The frequncy response is approx. 5Hz to over 100khz.

If someone here sees any garbage in what I've said above, I would appreciate the correction.

[Strabo]

I work better with pictures and found this description easier to understand.

http://www.vac-amps.com/page0027.html

[Strabo]

As Mark already stated, and to quote Kevin Hayes of VAC.

"It is important to note that in all cases the class of operation is determined by the character of the output at maximum power output. It is therefore inaccurate to refer to an amplifier as operating in Class A up to X watts and class AB above X watts; this is really a Class AB amplifier where cut-off first occurs at a level of X watts."

[Strabo]

Last one, Class B

[thebes]

When I said "resident sages" I wasn't kidding. Wonderful responses exspecially Gil's. Your use of analogies greatly enhances the understanding of a difficult subject for the non-technically oriented.

If I got this correct it would seem that when all is said and done, the Class A amps produce a cleaner sound at the cost of power consumption, yet the A/B's probably bring a little more slam or livileness to the music at the cost of distortion. While an elegantly simple solution in one respect the Class A energy use drives engineers nuts because they hate the trade off.

Still have a couple of questions, though.

Why are the preponderence of maps procuded a/b's. Are they simpler to manufacture or because of their higher wattage they are more responsive to need to drive inefficient speakers?

What happens if you build a realy high wattage Class A. Do you need your own nuclear power plant or is it because at certain point the signal degrades?

Why are there no mid-fi price range class A's they all seem to cost a lot of dough.

Finally, again what's this Class T stuff all about.

Not trying to drag out the thread but it appears there is interest in learning about this subject from others besides myself.

[3dzapper]

I think that the main reason that there are not more mid-power class A amplifiers is the invention of the beam power tube. This tube supplanted the triode for audio applications and further developement of the triode was stopped except for high power transmitter tubes. Thus SET class A amps go from 8 Watts(300B) to 150 Watts (833) with nothing much in between.

Rick

[mike stehr]

"Thus SET class A amps go from 8 Watts(300B) to 150 Watts (833) with nothing much in between."

There is a few that I think are in between.

845? 211? 811?(Not sure on the 811 at all...)

One thing is these bigger triodes require larger voltages.

And when you get up around 1000 or more volts, the practicality factor kinda comes in, at least from a DIY aspect.

I guess a guy could buy a SET amp/amps with a big hog triode, but it's obviously a spendy venture.

That's a whole different plane for DIY tube audio dealing with those voltages, something akin to big ham radios. To crazy for me.

[richinlr]

Thebes,

I used to own a 50 W per channel pure class A SS amp. Regarding why more folks don't make them, in a word HEAT! and more HEAT! Over time everything in the amp gets cooked and things start failing.

I used to joke about using it to cook hamburgers instead of firing up the grill.