For an efficient output stage, split the input signal up into easier to amplify pieces, using more than one amplifier, then sum the outputs.
For example, use a class D amplifier to amplify the signal, then subtract a scaled down portion of the class D output from the input signal. What is left of the input signal (residual) gets amplified by a small class A amplifier and summed with the class D output. The class A amp could use the summed outputs for its N feedback correction versus the original input signal too. Ensuring accurate output.
Another way would be to use a power D to A of maybe 8 bits, with a small class A amp to amplify the residual for summing. Similar N feedback using summed outputs.
Since class D amplifiers natively provide current output, followed by a capacitor to convert to V, one could just run it using current output from the class D (skip the cap). The main channel analog V amplifier (small in size though) would then have a current sensed output, providing N feedback to the class D amplifier for input. So the class D tries to minimize currrent drawn from the analog amplifier. The analog amplifier gets conventional N feedback of the total Voltage output signal for accuracy. So the analog V output amp sets the output voltage, the class D amp unloads it. This would work best with a Cuk type class D amplifier, to eliminate PWM ripple current. Allowing the analog amplifier to run with a clean load.
.
For example, use a class D amplifier to amplify the signal, then subtract a scaled down portion of the class D output from the input signal. What is left of the input signal (residual) gets amplified by a small class A amplifier and summed with the class D output. The class A amp could use the summed outputs for its N feedback correction versus the original input signal too. Ensuring accurate output.
Another way would be to use a power D to A of maybe 8 bits, with a small class A amp to amplify the residual for summing. Similar N feedback using summed outputs.
Since class D amplifiers natively provide current output, followed by a capacitor to convert to V, one could just run it using current output from the class D (skip the cap). The main channel analog V amplifier (small in size though) would then have a current sensed output, providing N feedback to the class D amplifier for input. So the class D tries to minimize currrent drawn from the analog amplifier. The analog amplifier gets conventional N feedback of the total Voltage output signal for accuracy. So the analog V output amp sets the output voltage, the class D amp unloads it. This would work best with a Cuk type class D amplifier, to eliminate PWM ripple current. Allowing the analog amplifier to run with a clean load.
.
Last edited:
Perhaps this analogy will help? (Probably not though!)
A circus-lady buys a bicycle, but always rides it with the front wheel in the air. Her friend says ‘that is a very strange monocycle you have there!’ She replies ‘well no it is actually a bicycle but I like to operate it as a monocycle’.
That is a very nice analogy!
It seems to me that the statement that an amplifier is class A should, unambiguously, mean that when operated up to and including its stated maximum power output, it is operating in class A mode.
It is also of interest to know, in the case of a class AB amplifier, at what power level it transitions from class A operation into class AB, since this gives useful information about the characteristics of the amplifier.
What is not helpful, I think, is if, as sometimes happens, the claim is made that an amplifier that is actually class AB is instead class A, on the grounds that "neither output tube ever fully cuts off." If one took that argument to its logical conclusion all class AB amplifiers are actually class A, since there is probably always at least a pico-amp of current flowing through the tube, even when in ordinary parlance it is cut off.
I like the bicycle analogy. When we say that a machine is a bicycle we are saying something about the machine (e.g. that it has two wheels of similar size, in line with each other). This does not preclude someone using it in an unusual way. It does not cease to be a bicycle because someone rides it on one wheel. In the same way an AB amp does not become A just because someone puts a small signal into it (and what would it be with no signal?).
And yes I am serious...
Why do you care, unless simply the class of bias has some inherent special property that has to be accurately communicated? It is easy to demonstrate a "class A " follower that has more output impedance modulation (distortion) than a highly biased diamond ("AB") output.
If we are going to define cutoff, let us look at some tube data sheets and some solid state data sheets as a start.
I believe that some tube data sheets have a 'cutoff' spec (i.e. 10uA, or other value).
And some solid state data sheets have a 'cutoff' spec (i.e. 10nA, or other value).
But then (for the sake of discussion) lets say an output tube is (effectively) cutoff when its current is 250 times less than the other output tube, as a start.
How about a push pull amp with a pair of KT88s.
250ma on push, and 1mA on pull, and then the opposite values during the other alternation of the sine wave?
I believe that some tube data sheets have a 'cutoff' spec (i.e. 10uA, or other value).
And some solid state data sheets have a 'cutoff' spec (i.e. 10nA, or other value).
But then (for the sake of discussion) lets say an output tube is (effectively) cutoff when its current is 250 times less than the other output tube, as a start.
How about a push pull amp with a pair of KT88s.
250ma on push, and 1mA on pull, and then the opposite values during the other alternation of the sine wave?
I don't think we have to get so specificatorily particulate... just glance at the bottom and see how close to 0 the plate current is at the bias voltage and plate voltage. For an EL34/6CA7 you get about 3-4 (-)V below -32v to get to 0 current. A Dynaco ST-70 is biased about -32v. So you might get 3W of music power in Class A before you reach cuttoff. Well, for me that means my amp runs in Class A 99% of the time because the volume control never goes high enough to need Class AB operation.
Last edited:
I think it is necessary to know what the quiescent current is too. In you example, if the quiescent current through each tube were about 125mA then your case would correspond to just about the maximum power output when it is still operating in class A. But if the quiescent current were, say, 40mA, then it would be operating well into class AB with your two currents in your example.
A case I saw recently was an OTL amplifier described as class A, and giving 30W into 8 ohms. The output stage consisted of five 6AS7 tubes, so 5 triode sections for each bank (it was a circlotron). The quiescent current couldn't have been more than, say, 70mA per triode section, so that means 350mA for the whole bank. The maximum power it could deliver in true class A would therefore be that corresponding to 700mA peak, which would mean about 2W average power, into 8 ohms. No way it could be called class A at 30W!
Last edited:
20to20,
I agree.
I simply purposely overshot the mark with such small specified cutoffs; and then with such a large ratio as 250:1.
Your Dyna is a good example of reality.
I have "ruined" my Dyna ST-70.
First, I re-stacked the OPTs and air gapped them, it became a single ended stereo amp with one EL34 per channel.
Later, with another manufacturer's Push Pull OPT, and a better choke, it has also been made into a push pull mono block with 2A3s (with dropping resistors, balance pots, and AC filaments), 300Bs (with DC filaments);
and Triode wired (100 Ohm plate to screen): KT66s, KT88s, and KT77s.
There were split load invertors, log tailed cathode coupled invertors, and "serial" invertors with the classical resistive divider to drive the other grid phase.
There has also been a single ended driver, with a choke as a current source for the output tubes coupled cathodes.
I took the value out of the once-stock ST-70, but then I put the value back in by using it as a test bed for so many topologies, and configurations.
The worst feature of using this amp as a test bed, is the steel chassis. It is hard to drill, and the fields of the power transformer, choke, and OPT are spread all across the chassis.
Along the way, I put 1 Ohm (and sometimes 10 Ohm) resistors between ground and the bottom of the Cathode self bias resistor/bypass cap pair. Each tube had its own bias pair and sense resistor. That way, I could monitor the dynamic cathode current of the push and the pull tubes.
Depending on how you bias the tubes, it is amazing at what low level or medium signal level causes the tube currents to 'flatten out' during each tube's cutoff region.
AB anyone?
For many push pull amplifier topologies, it is easy to put sense resistors in without materially disturbing any other circuitry or ruining the amp performance, and see for yourself the shape of the dynamic currents on your amp.
You can do this for sine waves, and more importantly on music and listening at the same time as watching the 2 channels on an oscilloscope (if you have one).
At the price of entry level scopes, versus your $$$ invested in CDs, Records, Amps, Loudspeakers, etc., how can you not afford a scope.
It is nice if you have a digital scope with FFT, XY, and measurement features, such as RMS, between the cursor RMS, max, min, frequency, etc.
But only purchase a scope if you will use it (you will, when you discover all a modern digital entry level scope can do for you).
I agree.
I simply purposely overshot the mark with such small specified cutoffs; and then with such a large ratio as 250:1.
Your Dyna is a good example of reality.
I have "ruined" my Dyna ST-70.
First, I re-stacked the OPTs and air gapped them, it became a single ended stereo amp with one EL34 per channel.
Later, with another manufacturer's Push Pull OPT, and a better choke, it has also been made into a push pull mono block with 2A3s (with dropping resistors, balance pots, and AC filaments), 300Bs (with DC filaments);
and Triode wired (100 Ohm plate to screen): KT66s, KT88s, and KT77s.
There were split load invertors, log tailed cathode coupled invertors, and "serial" invertors with the classical resistive divider to drive the other grid phase.
There has also been a single ended driver, with a choke as a current source for the output tubes coupled cathodes.
I took the value out of the once-stock ST-70, but then I put the value back in by using it as a test bed for so many topologies, and configurations.
The worst feature of using this amp as a test bed, is the steel chassis. It is hard to drill, and the fields of the power transformer, choke, and OPT are spread all across the chassis.
Along the way, I put 1 Ohm (and sometimes 10 Ohm) resistors between ground and the bottom of the Cathode self bias resistor/bypass cap pair. Each tube had its own bias pair and sense resistor. That way, I could monitor the dynamic cathode current of the push and the pull tubes.
Depending on how you bias the tubes, it is amazing at what low level or medium signal level causes the tube currents to 'flatten out' during each tube's cutoff region.
AB anyone?
For many push pull amplifier topologies, it is easy to put sense resistors in without materially disturbing any other circuitry or ruining the amp performance, and see for yourself the shape of the dynamic currents on your amp.
You can do this for sine waves, and more importantly on music and listening at the same time as watching the 2 channels on an oscilloscope (if you have one).
At the price of entry level scopes, versus your $$$ invested in CDs, Records, Amps, Loudspeakers, etc., how can you not afford a scope.
It is nice if you have a digital scope with FFT, XY, and measurement features, such as RMS, between the cursor RMS, max, min, frequency, etc.
But only purchase a scope if you will use it (you will, when you discover all a modern digital entry level scope can do for you).
cnpope,
I agree. You can do that with my suggested sense resistors, post #68, which I was writing, but your post beat me to the punch.
A DMM gives the quiescent current, a scope gives both the quiescent and dynamic currents.
Marketing always wins if you believe them. The OTL you referred to was marketing rated class A, then 30 Watts (hmm, the class AB requirement for 30 Watts was unstated by marketing).
I was in marketing for 5 years.
I had to tell the truth when I checked the rated specifications of our electronic test equipment that we were bidding to the US government (I did not want to go to jail).
And for the Navy Accuracy and Environmental Torture tests, we would not win any bid
if it did not do as advertised and stated.
But I also gave a lot of advice to commercial customers what our test equipment could do for them, the limits, accuracy, what it could not, unique features the were useful and nobody else could do.
And then I sent them to another vendor if we could not measure it, etc. But of course, then they always came back to me, because they trusted me to give them the whole scoop.
Integrity still counts in some people's minds.
Don't get me wrong, I have mis-calculated, and made other mistakes.
I agree. You can do that with my suggested sense resistors, post #68, which I was writing, but your post beat me to the punch.
A DMM gives the quiescent current, a scope gives both the quiescent and dynamic currents.
Marketing always wins if you believe them. The OTL you referred to was marketing rated class A, then 30 Watts (hmm, the class AB requirement for 30 Watts was unstated by marketing).
I was in marketing for 5 years.
I had to tell the truth when I checked the rated specifications of our electronic test equipment that we were bidding to the US government (I did not want to go to jail).
And for the Navy Accuracy and Environmental Torture tests, we would not win any bid
if it did not do as advertised and stated.
But I also gave a lot of advice to commercial customers what our test equipment could do for them, the limits, accuracy, what it could not, unique features the were useful and nobody else could do.
And then I sent them to another vendor if we could not measure it, etc. But of course, then they always came back to me, because they trusted me to give them the whole scoop.
Integrity still counts in some people's minds.
Don't get me wrong, I have mis-calculated, and made other mistakes.
Actually, Class C is required for high level AM modulation. Modulation is a form of mixing and that means:
1) Nonlinear
2) Time varying linear
3) Piece-wise linear
Time invariant linear systems are linear adders that produce no new frequencies except for whatever incidental modulation exists due to irreducible non-linearities. Since Class C is piece-wise linear (mostly) it will mix frequencies.
Since AM is an audio function, you need the same types of VTs as you'd use for an audio power final. From the spec sheet of the AW220:
Pd= 150KW
rp= 640R
u= 10.5
This being for the transmitter of a commercial AM station. It has the same low rp and u-factor as an audio power final.
The same reason why there is an 812 that's very similar to the 811, except that the former has lower u-factor and rp: improved sonic performance as a plate modulator.
If you're amplifying pre-existing AM, you need a linear amp as you don't want to be making more frequencies than what you have already.
And yes I am serious...
The Last Word on Class A
AM
Gee - this guy ....
But seriously: In totality the thread has become a very good read for those who would like to inform/update themselves.
Yes - development can discombobulate/warp/
well-intended original definitions. But (as said) I find the definition of class-A inclusive of the premise that anode current in a specific tube flows at all times. And I do not think the definer meant that minutia should cloud the scene, as on spending time with a protractor on a times 40 enlarged Va-Ia graph (or the semiconductor equivalent) pondering about where exactly the current starts.My question: Why/when is this important enough to make a spiel about it? ..... In my modest time as researcher I have not experienced that this was ever the main point of discussion at the tea table. So for whom is this important? In my experience, mainly for promotion! (Scott hinted as much in post #64.)
Yes, script writers have a task to inform (pious description
) the public blah blah blutsh .... But I am an EE and I dislike misuse of my profession's technical terms to mislead.Again: Why? Because of the ingrained perception that class-A is best, what it should be, supremo; way-to-go. Slip it in where you can to 'inform' (read bluff) any prospective clients. (That 'class-A' is not necessarily distortion-free itself and class-AB not necessarily inferior should be kept quiet - but not to open another OT discussion now!)
Most knowledgable folks would know that in class AB there is a lower portion of signal where anode/collector current always flow, not sure that others care. Why such accent on that particular characteristic? I fear as said.
May be because all people around your tea table spoke Afrikaans, while we here on the forum speak multiple languages, and have to define terminology in order to communicate properly?
Imagine a car that has a turbo that functions only when the rpm is below 900. (Imagine.) Then they market the car as Turbocharged. Other car makers would scream and I bet that “false” advertising would get shut down.
Imagine you make amplifiers for sale and you market their Class honestly while the other guys lie and call everything they sell “Class A.” (Many guitar amp makers tell this lie.)
It is a shame that the Class A phonies get away with it and they should be called out. It’s not only a lie but it’s especially egregious considering that it implies it is “the best” and 99%+ of the potential customers don’t know what it truly means.
If I made and sold amps I would probably try to call them out somehow. Maybe a little page showing their models and bogus claims. I’d rather see an amp advertised as Turbocharged than Class A, when it is not. A “Turbocharged” amp is legal hyperbole while I consider (bogus) Class A to be consumer fraud.
Imagine you make amplifiers for sale and you market their Class honestly while the other guys lie and call everything they sell “Class A.” (Many guitar amp makers tell this lie.)
It is a shame that the Class A phonies get away with it and they should be called out. It’s not only a lie but it’s especially egregious considering that it implies it is “the best” and 99%+ of the potential customers don’t know what it truly means.
If I made and sold amps I would probably try to call them out somehow. Maybe a little page showing their models and bogus claims. I’d rather see an amp advertised as Turbocharged than Class A, when it is not. A “Turbocharged” amp is legal hyperbole while I consider (bogus) Class A to be consumer fraud.
Jjman, nothing is "turbocharged" in class A, except electricity consumption. It is the lest efficient amplifier that was designed to work in class A.
When I see "Pure class A SINGLE ENDED" amplifier ad I smile, because how can single ended amp be other than class A? It is like, "In our candy we use only sweet sugar", as if sugar can not be sweet.
When I see "Pure class A SINGLE ENDED" amplifier ad I smile, because how can single ended amp be other than class A? It is like, "In our candy we use only sweet sugar", as if sugar can not be sweet.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.