Why no KG class AB ?

[Oneminde]

This is something I've been wondering about for a week or two. Why can't I find a class AB design by Kevin Gilmore ? Am I missing something. I know the difference between A and AB, but am still interested as to why. Is switching so bad that one of the best engineers stay away from it all together or is it simpler than that: "Why AB when you can A"

PS: I am a n00b so please be gentle :)

Oneminde

[luvdunhill]

This might help:

http://www.firstwatt.com/pdf/art_leave_classa.pdf

[Oneminde]

"The whole point of going to this trouble is to build an amplifier which sounds as good as possible." So then, one could say: You can chose an amplifier which operate in class AB and is many time more efficient than class A but you do so at the expense of musical quality. If you want the absolute best possible rendition of the source material you do so through class A.

So if I rephrase my original question, it would go something like this: Why did KG not design and released a poor performing amplifier ? ... LOL

Thanks for the PDF, it not only answered the question but also aided me in understanding the importance of biasing. But I am not done yet. Allow me to run something by you and see what you think.

 

 

[mwl168]

Another factor to consider is that building a true class A speaker amp is a far more demanding effort than building a class A amp to drive headphones. 

[Oneminde]

7 minutes ago, mwl168 said:

Another factor to consider is that building a true class A speaker amp is a far more demanding effort than building a class A amp to drive headphones. 

Is it my understanding then that because of general "simplicity" in designing an ideal class A headphone amp there is no real need to design class AB and D for headphone amps ? while for loudspeakers, the general consensus is that it is not as crucial to run class A and therefore class AB if designed properly is good enough ?

[JimL]

Well, I think things are a bit more complicated than that.

 

For speaker amps, you generally need a lot of power unless you have very efficient speakers, e.g. horns, which is another topic altogether. We're talking about 50-100 watts per channel or more. So the amp needs to produce 40 volts peak for a 100 watt amp, and on the order of 10 amps if we have a 4 ohm speaker.  Now, a class A push-pull amp has a maximum theoretical efficiency of 50% (single ended class A has a max efficiency of 25%), so that means it runs at 100-200 watts/channel all the time. This means lots of tubes (old days) or lots of heatsink for solid state. All this means lots of heat and lots of money (and that doesn't even include the cost of running the damn things). So basically, until Mark Levinson and Krell came along, pretty much all amplifiers were class AB because nobody thought that there was a big enough market to support a class A commercial amp. Class AB was considered the best compromise between cost, efficiency and sound quality. It was never considered the best, just that Class A was considered too expensive.  

 

The first Mark Levinson Class A amp was rated at 25 watts (but was reportedly class A down to 2 ohms, at which point it was a 100 watt amp), was mono, and was big, heavy and cost a few thousand dollars - I think you could have bought over 100 NAD3020 integrated amps for the same money, and the 3020 was stereo, had roughly the same power rating, and included a phono section. But, it turned out that there was a market for it, and Krell came out with stereo class A amps that had more power and generated more heat. And it's gone on from there. The limiting factors for speaker amps is basically how much power do you need and how much money (and size, and heat) do you want to spend getting it.

 

For non-electrostatic headphone amps it is significantly different. Here, you only need a few watts at most, into higher impedances than speakers have. So low to moderate voltages (say 15 volts at most)and relatively low currents (an amp or so at most).  Based on specs, the least efficient non-electrostatic headphone is the HFM Susvara, and 5 watts into that should produce  120 dB, which is enough to give you hearing damage in minutes if sustained. So in theory, a class A headphone rated at 5 watts into the Susvara's impedance should do the trick. Which means that an amp that draws 25-30 watts from the wall should be able get you close to the threshold of pain  In that case, there isn't a lot to be gained by dropping back to class AB. That said, some cheap headphone amps ARE run in class AB because the manufacturer doesn't want to lay out money for heatsinks. Every penny counts.

 

For electrostatic headphones, it is different again, stat headphones have very high impedances, so you need very high voltages but relatively minuscule currents - a few mA. because you need to have high voltage - the Stax amps generally have +/-350V supplies, the KG amps generally run +/-350V to +/-450 volts, the Stax and DIY T2 run +/-500V supplies. Now the Stax amps run around 5-7 mA/output device, the KG amps up to 20 mA/output device, and there are 2 output devices per channel. So given that, the Stax amps draw about 40-55 watts from the wall, the KG amps run about 140 watts to more than 200 watts (the latter from a DIY T2, which also has 8 tube heaters using energy).

Edited January 11, 2019 by JimL

[Oneminde]

@JimL Splendid and got it. I've read - just finished - the original Dynalo aka A Pure Class A Dynamic Headphone Amplifier - article that mister K.G produced back in 01 and while much of it was understandable, some other aspects need time to sink in and it will clear up more once I get to actually build one. As it happens, I so far plan on building almost all Dyn versions as well. But I get it, watts cost and I agree with you in that regard. Not only are we talking about signal purity, but are the source material and loudspeakers of sufficient quality to justify so much energy being thrown out ....

The DIY T2 is insanely awesome. Can't wait until I get the electrostatic project up and running.

Again, thanks for the reply, it made my life easier

[JimL]

One other detail is how the push-pull output stage is biased. This is basically what determines whether an amp is Class B, AB or A. For class B, the output stage is biased at zero, meaning with no signal there is no current running through the output devices. This is basically only used in radio frequency transmitters.

 

In class AB, there is some current running through the output devices with no signal, but when a signal is sent through the amp, each of the two output devices only reproduce half plus a bit more of the output, and the combination of both devices is needed to produce the whole signal. In reality the output devices are turning on and off during the output, which can produce crossover distortion, which is a transient that occurs when the device turns on or off. The problem is that this distortion actually gets worse as a proportion of the output was you go to lower power, until you get down to the point where the signal is small enough that the turn on and turn off does not occur.

 

Now with most music other than the compressed crap that some modern producers are putting out (i.e. loud all the time), the fact is that much of music is at a volume that is 10-20 dB below maximum, so that an amp is usually running a 1-10% of max power. For example, take a typical speaker with a sensitivity of 86 dB/watt/1 meter. Now, most people listen a few meters from the speakers, so the sound level there is several dB lower, but this gives you a rough idea. Typical conversation is about 60 dB, so 86 dB level is like standing a few feet apart and yelling at each other. What you can see is that with most music, the amount of power an amp is putting out is usually on the order of a watt or less.

 

In early transistor amps crossover distortion was a problem, because they were run at such a low bias so that there was lots of crossover distortion being produced a lot of the time. The solution is to increase the bias so that most of the time the amp is running in class A (ie. both output devices reproducing the whole music cycle), but that means more current and bigger heatsinks. Tube amps, OTOH, generally ran with enough bias current that they were basically class A up to a couple watts, which for most music meant they were actually class A most of the time. .

 

Now for true class A, there is enough current running that both devices output some signal current at all times. In theory this should only matter for music peaks. In reality, since active devices (tubes, transistors) are more linear in the middle of their operating range rather compared to their limits, this decrease distortion at all levels.

[Oneminde]

@JimL Thank you. So much has cleared up from that explanation, because when I read that the Dynalo is a PP class A my mind was confused. I though of class A as SE and class AB as PP (tubes territory), but that is obviously wrong and I wholeheartedly wana thank you for sorting it all out.  I have a new found respect for the differences between A and AB now.

So, am I getting it right that a push-pull class A is in reality class B with the bias set at the same Q-point as single ended class A ? (yes, PP amplify only half of the signal, either + or - portion)
 
Theoretically, one could then switch between class A and class AB operation - hard switch (resistor bank) - which changes the bias or Q-point. That is kinda interesting. But ... so far I have not seen an amp which feature an A/AB switch. If we look beyond the glaring reasons for choosing class A. There is a place for class AB and the Dynalo could potentially be featured with this option if lets say under battery operation and one wish to extend the battery life ... as an example for the Dynalo Mk2 Mini version. Still very early in the process for me and I am collecting info the different version of the linear class A amp (Dynalo/Dynahi), but not going to spend time and energy for does in this thread.

So now, since you so kindly revealed some crucial information, I will read some info on the classes and biasing. Found a site which seem to explain things properly:
https://www.electronics-tutorials.ws/amplifier/amp_5.html

[kevin gilmore]

long time ago there was a Yamaha integrated amp that had an A/AB switch. like about 1978. really nice thing, would make a perfect headphone amp if you can find one that still works.

in fact the CFA amp is a A/B amp with the bias between 150 and 200ma. These days that kind of bias into just about any headphone results in pure class A operation. The uberamp2 is also a A/B amp with the bias between 2 and 3 amperes. about 100 watts into 8 ohms balanced.

[Oneminde]

Yamaha CA-2000/2010 - "Normal / Class A" Nice of you to drop in Kevin. Yeah, does Yamaha amps are sought after still to this date, maybe one day I can get one for my collection. Too bad no one makes the V-Fet these days.

CFA, not sure which amp that is ?

Uberamp2 .. I've not gotten around to that one, I am currently focusing on your Dynalo and Dynahi amps, yep, late to the game, but not to late since I know I can get the enough 2SA1015 / 2SC1815 transistors and also match them... which also mean I need to get my hands on / build a curve traces. I am sure there are paralleled quality or better, but, it has to be in the spirit of what you did back in the day, its a rule - the Dynalo Mk2 is a modern take on that and looking forward to that one as well.

I have plenty of amps to catch up to so to say, especially since I will venture into Electrostatics as well. This hobby will be expensive.

"KG Amps, the meaning of life" ...

[mikeymad]

and all the way back to the CA1000 - 1973

"operation"

 

[Oneminde]

Yeah ... why did that sort of go away, what the heck happened.

[spritzer]

I have a Marantz PM-64 here which is Class A up to 25W I think but no switch.  I need to check...

[Oneminde]

2 minutes ago, spritzer said:

I have a Marantz PM-64 here which is Class A up to 25W I think but no switch.  I need to check...

How does that work btw. How is amp on A mode for a while then go into AB mode at higher wattage ?

[jose]

 

19 minutes ago, spritzer said:

I have a Marantz PM-64 here which is Class A up to 25W I think but no switch.  I need to check... 

 Pass amps (not First Watt) work as well. No?

[spritzer]

Yup, you set the desired current and then it goes out of that range.  Odds are though that most of the time...people never will. 

[Oneminde]

CFA - current feedback electrostatic amp ... found it.

[Pars]

There is a dynamic CFA2/3 as well. Now for something completely different...


Sent from my iPhone using Tapatalk

[Oneminde]

I am back and while the previous debates cleared up differences between class A and AB, I have some further remarks. In order to make these remarks I will use available information through [electronics-tutorials . ws] which I will link to down bellow. If not for anything new, then this entry is in large for myself and possibly other who read and have the same questions / concern as I do, and remember, this is for educational purposes.

***
I started this thread with a question: Why no KG class AB ? because my perspective was that if done right, class AB could sound as good as class A but retain a better efficiency. Not that I made that clear in the first entry, at least it was in the back of my mind and I wanted to check for explanations from other more enlightened head-case members who could explain things in a way I could not. I am still learning and the information which I have come over in the past few day is an eye opener. Yes yes I know, for most of you the information I will present is old news, no brainer or perhaps jaw dropping. No matter, I've had a blast, learned things I did know would transpire and also made it very clear how things should be done from now on. So lets get going.

Edited reply's for the convenience of this entry, but you can always go up and read the full statement(s).

JimL;
1) - So basically, until Mark Levinson and Krell came along, pretty much all amplifiers were class AB because nobody thought that there was a big enough market to support a class A commercial amp. Class AB was considered the best compromise between cost, efficiency and sound quality. It was never considered the best, just that Class A was considered too expensive.
 2) - The first Mark Levinson Class A amp was rated at 25 watts (but was reportedly class A down to 2 ohms, at which point it was a 100 watt amp), was mono, and was big, heavy and cost a few thousand dollars - I think you could have bought over 100 NAD3020 integrated amps for the same money, and the 3020 was stereo, had roughly the same power rating, and included a phono section. But, it turned out that there was a market for it, and Krell came out with stereo class A amps that had more power and generated more heat. And it's gone on from there. The limiting factors for speaker amps is basically how much power do you need and how much money (and size, and heat) do you want to spend getting it.
3) - For non-electrostatic headphone amps it is significantly different. Here, you only need a few watts at most, into higher impedances than speakers have. So low to moderate voltages (say 15 volts at most)and relatively low currents (an amp or so at most).  Based on specs, the least efficient non-electrostatic headphone is the HFM Susvara, and 5 watts into that should produce  120 dB, which is enough to give you hearing damage in minutes if sustained. So in theory, a class A headphone rated at 5 watts into the Susvara's impedance should do the trick. Which means that an amp that draws 25-30 watts from the wall should be able get you close to the threshold of pain  In that case, there isn't a lot to be gained by dropping back to class AB. That said, some cheap headphone amps ARE run in class AB because the manufacturer doesn't want to lay out money for heatsinks. Every penny counts.
4) - One other detail is how the push-pull output stage is biased. This is basically what determines whether an amp is Class B, AB or A. For class B, the output stage is biased at zero, meaning with no signal there is no current running through the output devices.
5) - In class AB, there is some current running through the output devices with no signal, but when a signal is sent through the amp, each of the two output devices only reproduce half plus a bit more of the output, and the combination of both devices is needed to produce the whole signal. In reality the output devices are turning on and off during the output, which can produce crossover distortion, which is a transient that occurs when the device turns on or off. The problem is that this distortion actually gets worse as a proportion of the output was you go to lower power, until you get down to the point where the signal is small enough that the turn on and turn off does not occur.
6) - Now with most music other than the compressed crap that some modern producers are putting out (i.e. loud all the time), the fact is that much of music is at a volume that is 10-20 dB below maximum, so that an amp is usually running a 1-10% of max power.
7) - Now for true class A, there is enough current running that both devices output some signal current at all times. In theory this should only matter for music peaks. In reality, since active devices (tubes, transistors) are more linear in the middle of their operating range rather compared to their limits, this decrease distortion at all levels.

So far so good right. I sat down an made a cup of coffee with a thought in the back of my mind: Remember, KG Dynalo and Dynahi is push-pull (PP) class A amplifier. The PP class A was new to me because I viewed class A as single ended and push-pull as class AB.

• Class A: – The amplifiers single output transistor conducts for the full 360^o of the cycle of the input waveform.
• Class B: – The amplifiers two output transistors only conduct for one-half, that is, 180^o of the input waveform.
• Class AB: – The amplifiers two output transistors conduct somewhere between 180^o and 360^o of the input waveform. Clarification: AB which use a minimum of two transistors conduct 0^o to 180^o for the positive cycle and 180^o to 360^o for the negative cycle.

Class A: The most commonly used type of power amplifier configuration is the Class A Amplifier. The Class A amplifier is the simplest form of power amplifier that uses a single switching transistor in the standard common emitter circuit configuration as seen previously to produce an inverted output. The transistor is always biased “ON” so that it conducts during one complete cycle of the input signal waveform producing minimum distortion and maximum amplitude of the output signal. This means then that the Class A Amplifier configuration is the ideal operating mode, because there can be no crossover or switch-off distortion to the output waveform even during the negative half of the cycle. Class A power amplifier output stages may use a single power transistor or pairs of transistors connected together to share the high load current.

This is the simplest type of Class A power amplifier circuit. It uses a single-ended transistor for its output stage with the resistive load connected directly to the Collector terminal. When the transistor switches “ON” it sinks the output current through the Collector resulting in an inevitable voltage drop across the Emitter resistance thereby limiting the negative output capability. The efficiency of this type of circuit is very low (less than 30%) and delivers small power outputs for a large drain on the DC power supply. A Class A amplifier stage passes the same load current even when no input signal is applied so large heatsinks are needed for the output transistors.

However, another simple way to increase the current handling capacity of the circuit while at the same time obtain a greater power gain is to replace the single output transistor with a Darlington Transistor. These types of devices are basically two transistors within a single package, one small “pilot” transistor and another larger “switching” transistor. The big advantage of these devices are that the input impedance is suitably large while the output impedance is relatively low, thereby reducing the power loss and therefore the heat within the switching device.

It is possible to obtain greater power output and efficiency than that of the Class A amplifier by using two complementary transistors in the output stage with one transistor being an NPN or N-channel type while the other transistor is a PNP or P-channel (the complement) type connected in what is called a “push-pull” configuration. This type of power amplifier configuration is generally called a Class B Amplifier and is another type of audio amplifier circuit that we will look at in the next tutorial.

Class B: To improve the full power efficiency of the previous Class A amplifier by reducing the wasted power in the form of heat, it is possible to design the power amplifier circuit with two transistors in its output stage producing what is commonly termed as a Class B Amplifier also known as a push-pull amplifier configuration. Push-pull amplifiers use two “complementary” or matching transistors, one being an NPN-type and the other being a PNP-type with both power transistors receiving the same input signal together that is equal in magnitude, but in opposite phase to each other. This results in one transistor only amplifying one half or 180^o of the input waveform cycle while the other transistor amplifies the other half or remaining 180^o of the input waveform cycle with the resulting “two-halves” being put back together again at the output terminal.

Then the conduction angle for this type of amplifier circuit is only 180^o or 50% of the input signal. This pushing and pulling effect of the alternating half cycles by the transistors gives this type of circuit its amusing “push-pull” name, but are more generally known as the Class B Amplifier.

The Class B Amplifier has the big advantage over their Class A amplifier cousins in that no current flows through the transistors when they are in their quiescent state (ie, with no input signal), therefore no power is dissipated in the output transistors or transformer when there is no signal present unlike Class A amplifier stages that require significant base bias thereby dissipating lots of heat – even with no input signal present. So the overall conversion efficiency ( η ) of the amplifier is greater than that of the equivalent Class A with efficiencies reaching as high as 70% possible resulting in nearly all modern types of push-pull amplifiers operated in this Class B mode.

The Class B amplifier circuit above uses complimentary transistors for each half of the waveform and while Class B amplifiers have a much high gain than the Class A types, one of the main disadvantages of class B type push-pull amplifiers is that they suffer from an effect known commonly as crossover distortion.

Hopefully we remember from our tutorials about Transistors that it takes approximately 0.7 volts (measured from base to emitter) to get a bipolar transistor to start conducting. In a pure class B amplifier, the output transistors are not “pre-biased” to an “ON” state of operation. This means that the part of the output waveform which falls below this 0.7 volt window will not be reproduced accurately as the transition between the two transistors (when they are switching over from one transistor to the other), the transistors do not stop or start conducting exactly at the zero crossover point even if they are specially matched pairs.

The output transistors for each half of the waveform (positive and negative) will each have a 0.7 volt area in which they are not conducting. The result is that both transistors are turned “OFF” at exactly the same time. A simple way to eliminate crossover distortion in a Class B amplifier is to add two small voltage sources to the circuit to bias both the transistors at a point slightly above their cut-off point. This then would give us what is commonly called an Class AB Amplifier circuit. However, it is impractical to add additional voltage sources to the amplifier circuit so PN-junctions are used to provide the additional bias in the form of silicon diodes.

The Class AB Amplifier: We know that we need the base-emitter voltage to be greater than 0.7v for a silicon bipolar transistor to start conducting, so if we were to replace the two voltage divider biasing resistors connected to the base terminals of the transistors with two silicon Diodes. The biasing voltage applied to the transistors would now be equal to the forward voltage drop of these diodes. These two diodes are generally called Biasing Diodes or Compensating Diodes and are chosen to match the characteristics of the matching transistors.

The Class AB Amplifier circuit is a compromise between the Class A and the Class B configurations. This very small diode biasing voltage causes both transistors to slightly conduct even when no input signal is present. An input signal waveform will cause the transistors to operate as normal in their active region thereby eliminating any crossover distortion present in pure Class B amplifier designs. A small collector current will flow when there is no input signal but it is much less than that for the Class A amplifier configuration. This means then that the transistor will be “ON” for more than half a cycle of the waveform but much less than a full cycle giving a conduction angle of between 180^o to 360^o or 50% to 100% of the input signal depending upon the amount of additional biasing used. The amount of diode biasing voltage present at the base terminal of the transistor can be increased in multiples by adding additional diodes in series.

Class B amplifiers are greatly preferred over Class A designs for high-power applications such as audio power amplifiers and PA systems. Like the class-A amplifier circuit, one way to greatly boost the current gain ( A_i ) of a Class B push-pull amplifier is to use Darlington transistors pairs instead of single transistors in its output circuitry.

This section is the most interesting and important one.

For Class A amplifier operation the switching transistors Q-point is located near to the center of the output characteristic load line of the transistor and within the linear region. This allows the transistor to conduct for the complete 360^o so the output signal varies over the full cycle of the input signal. The main advantage of Class A is that the output signal will always be an exact reproduction of the input signal reducing distortion. However it suffers from poor efficiency, because to bias the transistor in the center of the load line there must always be a suitable DC quiescent current flowing through the switching transistor even if there is no input signal to amplify.

For Class B amplifier operation, two complimentary switching transistors are used with the Q-point (that is its biasing point) of each transistor located at its cut-off point.

This allows for one transistor to amplify the signal over one half of the input waveform, while the other transistor amplifies the other half. These two amplified halves are then combined together at the load to produce one full waveform cycle. This NPN-PNP complimentary pair is also known as a push-pull configuration.

Because of the cut-off biasing, the quiescent current is zero when there is no input signal, therefore no power is dissipated or wasted when the transistors are in the quiescent condition, increasing the overall efficiency of a Class B amplifier with respect to Class A. However, as the Class B amplifier is biased so that the output current flows through each transistor for only half of the input cycle, the output waveform is therefore not an exact replica of the input waveform since the output signal is distorted. This distortion occurs at every zero-crossing of the input signal producing what is generally called cross-over distortion as the two transistors switch “ON” between themselves.

This distortion problem can be easily overcome by locating the biasing point of the transistor slightly above cut-off. By biasing the transistor slightly above its cut-off point but much below the centre Q-point of the class A amplifier, we can create a Class AB amplifier circuit. Then the basic purpose of a Class AB amplifier is to preserve the basic Class B configuration while at the same time improving its linearity by biasing each switching transistor slightly above threshold.

End of article entry.

Conclusion - Bias, bias and some more bias.
So now that we have a grasp of the difference between class A, B and AB, how do we tie together class AB and push-pull class A ? We do so by adjusting the bias to the transistors quiescent point or Q-point to the linear portion. The transistor is a linear device in the active region and nonlinear in its cutoff and saturation region. - That's it. This is the answer and "secret" wo why Kevin Gilmore use push-pull class A in his dynamic headphone amplifiers.

"For now I have seen the light". We can ofc achieve the same performance via single ended but we are power limited, by using several transistors in a chain we can obtain a much higher output power but maintain in class A the whole time.

That's it. I don't think I have anything els to add here. I totally get why KG is only interested in push-pull class A.

- Oneminde

Sources:
1. https://www.electronics-tutorials.ws/amplifier/amp_5.html
2. https://www.electronics-tutorials.ws/amplifier/amp_6.html
3. https://www.electronics-tutorials.ws/amplifier/class-ab-amplifier.html