Why no Class B amps?

[crouse]

First, know that I know NOTHING about about amplifier design, so I am asking this purely to educate myself. Since there are Class A amps and Class A/B amps, why do there seem to be no Class B amps? I have read that the crossover from A to B can be a source of distortion. If there actually are no pure Class B amps, I imagine there is a very good reason. What's that reason?
thanks.

 

[grateful]

Class B amps have crossover distortion. Generally they are not suitable for hifi use. There are exceptions, Early McIntosh ss units have class B output stages and have managed to get around those problems.

 

[crouse]

Class B amps have crossover distortion. Generally they are not suitable for hifi use. There are exceptions, Early McIntosh ss units have class B output stages and have managed to get around those problems.

Thanks. Do Class A/B amps have the same problem when operating in B? Sorry if it's a stupid question. Like I said, just trying to learn something here.

 

[mhardy6647]

Indeed, as implied by Grateful above - the crossover distortion isn't the switch from class A to class B operation, it's the crossover from one amplifying device (the positive half of the waveform) to the other device (the negative half of the waveform). Pure Class B is very efficient (well, by analog amplifier standards) but not low-distortion (in terms of hifi expectations).

 

[mhardy6647]

To answer your question -- there's a reason why some 'audiophiles' rediscovered -- and now espouse -- the profoundly inefficient and long-ago abandoned (by audio circuit engineers) single ended (Class A) operation for power amplification

 

[Superampman]

Naim is a highly respected world class amp and is considered pure class B by the manufacturer. How they define this is probably up to debate, however.

 

[SuperjazzyJa]

To answer your question -- there's a reason why some 'audiophiles' rediscovered -- and now espouse -- the profoundly inefficient and long-ago abandoned (by audio circuit engineers) single ended (Class A) operation for power amplification

This is probably a stupid question too. A single ended amp isn't really a power amp right? I know it has an output stage, but doesn't the output stage need to consist of 2 or more gain stages in parallel in order to truly be a power amp? :scratch2:

 

[crouse]

Indeed, as implied by Grateful above - the crossover distortion isn't the switch from class A to class B operation, it's the crossover from one amplifying device (the positive half of the waveform) to the other device (the negative half of the waveform). Pure Class B is very efficient (well, by analog amplifier standards) but not low-distortion (in terms of hifi expectations).

So, the crossover distortion is present when A/B amps are operating in B only?
thanks

 

[grateful]

Crossover distortion is caused by what percentage of the signal is passed through the output device and where they transition, consider a sine wave going from peak to peak. Going from 0 to the positive peak is 90 degrees back to 0 is 180 to the negative peak 270 and back to 0 360. A class A output stage one transistor conducts the whole 360. In a class B one transistor does the top 180 the other the bottom 180. AB operation is similar to class B but both transistors conduct for more than 180 but not 360. The overlap where they are both conducting smooths out the notch.

 

[Guest105]

In my understanding, Class A/B amps never operate exclusively in Class B. Crossover distortion is reduced by the A/B design, but not eliminated.

Edit: I should have waited for Grateful's post....excellent.

 

[Dr Tinear]

So, the crossover distortion is present when A/B amps are operating in B only?
thanks

If it's a class AB amp, by definition it doesn't operate in class B.

If output transistors were perfectly linear, a class B push-pull amp would work without crossover distortion because the output devices would "hand off" the signal from one to the other without a gap. Unfortunately there's no such thing as a perfectly linear output device, so crossover distortion occurs in a class B amp when one output device stops conducting before the other one starts conducting. Class AB was introduced back in the tube era to bias both devices on around the point where the signal handoff occurs. The technique was well known by the late 1930s and is discussed quite thoroughly in Frederick Terman's Radio Engineering and Fundamentals of Radio, two classic engineering texts from that era.

There have been a few Class B solid-state audio amps. The Dynaco Stereo 120 and Stereo 80 were both Class B designs that relied on feedback to smooth out the crossover notch, and some other designs from the 1960s followed a similar design philosophy.

 

[crouse]

If it's a class AB amp, by definition it doesn't operate in class B.

If output transistors were perfectly linear, a class B push-pull amp would work without crossover distortion because the output devices would "hand off" the signal from one to the other without a gap. Unfortunately there's no such thing as a perfectly linear output device, so crossover distortion occurs in a class B amp when one output device stops conducting before the other one starts conducting. Class AB was introduced back in the tube era to bias both devices on around the point where the signal handoff occurs. The technique was well known by the late 1930s and is discussed quite thoroughly in Frederick Terman's Radio Engineering and Fundamentals of Radio, two classic engineering texts from that era.

There have been a few Class B solid-state audio amps. The Dynaco Stereo 120 and Stereo 80 were both Class B designs that relied on feedback to smooth out the crossover notch, and some other designs from the 1960s followed a similar design philosophy.

Ah, I think you've hit upon part of my confusion. So, class AB is a separate type of amp that has nothing to do with class B?

 

[grateful]

They are modes of operation. Some amps have sliding biases that change the crossover point depending on power output. In some cases they can slide from class A to AB to B operation dynamically. Most are designed for one mode of operation.

 

[fred soop]

Class AB is sort of a "fuzzy" definition. According to amplifier expert Douglas Self, in his power amplifier books, these should all be called class B even though there must be some idle current to avoid the crossover distortion. That will result in class A operation up to a few milliwatts. This is true even for the Dynaco and McIntosh designs mentioned above.

Self also points out that proper class B design these days can now be made with low enough distortion, even at the crossover, (and this is backed up by Bob Cordell in his power amplifier book) that there is really no point in tolerating the expense, heat, power consumption, and overall inefficiency of class A for anything above a few watts.

 

[for_p1]

So, the crossover distortion is present when A/B amps are operating in B only?
thanks

Yes crossover distortion does present in AB amps. Feedback is generally used to reduce it further. In pure class B feedback won't help - at some point it needs infinite gain, which is not possible to achieve. There are some other clever ticks like sliding bias, pre-distortion, separate low and high signal path and others, used to increase efficiency and stay mostly in class B mode.

 

[Adda]

No one has mentioned extended class AB also known as class AAB.

These are basically class AB amps with very high idle current, as a result, they stay in class A at higher output power then conventional AB amps, typically up to around 10w.

A very inefficient way of getting rid of crossover distortion.

 

[crouse]

Thanks everyone! I can always count on AK for answers.

 

[MacPhantom]

A bit late, but let me note this down to avoid confusions (help me by correcting any mistakes you can find).

Most basic reasons for crossover distortion:

• A transistor is only able to amplify DC voltage (either positive or negative). Audio signals are AC though.
• The used semiconductor material has a certain minimum voltage that has to be present at the transistor's Base (or Gate). Silicon has around 0.7V, Germanium around 0.4V iirc.

Solutions for this in a nutshell (i.e., most basic amplifier topologies):

• Class A: Bias the audio signal to raise it entirely into variable DC (example: if the signal has a negative peak of -1.4V, add at least +2.1V bias if you use silicon transistors to raise the lowest signal point to just about enough for the transistor to turn on, i.e. 0.7V).
  With this configuration, there's only one transistor needed in theory that is always turned on and amplifies both "positive" and "negative" waveforms (biased to more and less positive).
  Addendum (this may go too far): due to the DC bias, speakers would blow after a few seconds with such an output. There's three strategies to avoid this problem: (i) Use an output transformer that only lets through AC; (ii) Add a second (PNP) transistor to the emitter of the first one (NPN) as well as an output capacitor that blocks DC; (iii) do the same as in (ii), but omit the output cap and give the two transistors positive and negative supply voltages.
• Class B: at least two transistors do the amplification job for the positive resp. negative waveform (PNP and NPN transistors). Their bases (resp. gates) are not biased at all. Therefore, class B amps only amplify what's above the absolute value of the minimum required voltage, and everyting in the range (-0.7V, +0.7V) for silicon transistors is not amplified. This gives an enormous amount of crossover distortion and is the main reason why standalone Class B amps are not encountered anymore in audio technology.
• Class AB: this is NOT an amp that switches from B to A mode or vice versa as often (mis-) explained. Some amps (e.g. by Accuphase) can be switched from A to B mode, but these are informally called "A/B", but certainly not "AB". Instead, a class AB amp is, to a certain extent, a class B amp which, just like the class A, has a DC-biased base/gate. However, the bias is not as high as in class a, but just about enough to bridge the ±0.7V gap. Example: if the amp is currently amplifying a sine wave at a momentary amplitude of -0.9V, the negative transistor is working as expected. At the same time, due to the bias, also the positive transistor is already turned on (base voltage >0.7V) and is just about ready to take over the waveform. The same happens the other way round.
• Class D: the problem about class A amps is that they are ridiculously inefficient (efficiency of about 10%), and hence they run extremely hot if you listen at low volume. A transistor works best either at the minimum base current (bi-polar transistors) / minimum gate voltage (FETs) respectively their maximum current / voltage. At the minimum, the transistor does not conduct at all; at the maximum, it has barely any resistance (thus does not abduct energy to generate heat). Class D amps make use of this very property insofar as they turn a transistor entirely on and off at very fast frequencies (above the audible limit of ~20kHz). Then, this carrier frequency is modulated, usually in its width ("pulse-width modulation"; PWM): if for instance a 50Hz sine wave is fed, the high-frequency carrier signal pulses become wider. At the end, a low-pass filter attenuates the then unwanted carrier and out comes "only" the audible part. Fyi: frequency modulation is also used in many other domains (ICT, SAR, etc.). Class D amps thus have an efficiency of >80%. If combined with a switching power supply (works more or less similarly), you get VERY powerful amps. Ever wondered why PA rackmount gear is so powerful? This is why. Of course there's more tricks to do (e.g. class H), but I just cannot go into detail for this.

Now for some special cases:

• Class A+: This is Technics (Matsushita) only and was only ever used in *one* device: the famous SE-A1. In principle, the SE-A1 is a class A amp. However, Matsushita applied an ingenious trick: the power supply for this class A stage comes from a huge class B amp which also gets the audio signal as a trigger. If the signal has louder passages and requires more output power, the class B amp provides more supply current. In quiet passages, the current drops down and the class A stage does not generate that much heat. As already said: very intelligent, but they could do equally good:
• New Class A: also a Technics label. This technique included a thermistor and few transistors ("synchro-bias") which did not affect the supply voltage, but rather increased the idling current if required (i.e. at quiet passages) and did the inverse during loud sequences. Competitors (Denon, Pioneer, Dual) and daughter companies (JVC) achieved similar effects with slightly different techniques (i.e. Denon's "Optical Class A").
• Class AA: again Technics-affiliated, but allegedly invented by a certain Mr. Sandman. This design includes both a class A and class B stage. During quiet passages, speakers are mainly driven by the class A stage ("Voltage Control Amplifier"). As soon as a certain voltage threshold is reached (determined by the "class AA bridge" resistor network), the (beefier) class b stage takes over and does its job ("Current Drive Amplifier") until the input signal falls below the threshold. This threshold is logically chosen to be slightly higher than the required minimal voltage of 0.7V in silicon transistors.

Now, last but not least, the most important take home messages:

• Don't judge an amp by its topology. A well-made class AB amp can sound better than a class A amp.
• Class A power amps are unecological, unneccessary (you can get equal or better THD ratings with other design principles, if constructed well), and therefore obsolete.
• Special and unofficial names like "New Class A" are marketing, yes. However, this is only partly true for they are very great design principles.

 

[E-Stat]

The Dynaco Stereo 120 and Stereo 80 were both Class B designs that relied on feedback to smooth out the crossover notch, and some other designs from the 1960s followed a similar design philosophy.

Yep. Another from that vintage was the AR Integrated amp. I purchased one in '72 based upon a glowing Julian Hirsch review and its looks. Sounded fine at full power, but resolution disappeared at low levels.

It wasn't long before I got rid of it.

 

[chicks]

Yep. Another from that vintage was the AR Integrated amp.

Sorry, but you're incorrect. A major issue with the original AR amps was, in fact, the bias pots for the output transistors, which were wirewound, and would open up. Class-B amplifiers don't use bias pots, as they aren't biased into Class AB.

http://www.classicspeakerpages.net/IP.Board/index.php?showtopic=971