Class A vs AB?

[shooter]

Got a question on A vs AB.  I get the DC inefficiency in A vs AB, on 100% vs on 60ish%, what I’m hung up on is where the AC  inefficiency is.  It seems like tube gain wouldn’t be affected.  I’m still at the “theory of operation” stage in my quest!
In my “brain model” I’m using 2 tubes for both classes.  Anyway, thanks for your help clearing the mud in my brain.

[jazbo8]

The key difference is how the tubes are biased, also for single-ended amplifiers, you are pretty much limited to Class A. I think your question is more for the push-pull amplifier, right? Anyway, here is a good summary of all the difference classes of operation.

[HotBluePlates]

... what I’m hung up on is where the AC  inefficiency is.  ...

I don't know what you mean by AC inefficiency. So you tell me what you mean by that statement, and I can probably answer how class A and class AB are different within that context.

[PRR]

Class A must idle with ALL the power to cover the PEAK of the wave, and waste it if not needed.

Class B can idle stone-cold, increasing its power demand as output power rises. At all power levels its input power is less than a Class A stage of the same peak power output.

AB is somewhere in between.

And yes, all *undistorted* single-ended amps must run class A. A single-ended Class B stage would cut-off half the wave. This is unacceptable in audio. It happens to be quite acceptable in some tuned radio transmitters.

[shooter]

 Most all the descriptions I've read talk, bias, operational efficiency, max theoretical for class A at about 40-50%.  I get bias, max values etc.
 what I'm wondering is if my tubes (2 class A,or 2 class AB, same type) having a gain of say 60 , how is class AB twice as loud? Class A I put the zero crossing point about center on the load line, AB it's shifted to cut the tube off somewhere close to the zero crossing. does that allow each tube (AB) to develop twice the amplitude AC signal.  I see the AC signal in either case as being limited by the same parameters.  I understand it takes twice the electricity in class A as AB but am I getting the same "volume" out the speaker?  Does  parallel Class A = class AB? as far as the AC signal gain is concerned?
Thanks again for the help.

[jjasilli]

Maybe the best thing is to look at the tube charts, say for a 6L6:  http://www.mif.pg.gda.pl/homepages/frank/sheets/093/6/6L6GC.pdf


Class A, pentode mode, one tube   (SE) : +/- 350 plate volts; up to 10.8W output
Class A, pentode mode, two tubes  (PP):  +/- 270 plate volts; up to 17.5W output (N.B.: less than 2X the power output of one tube, due to inherent circuit inefficiencies)
Class AB, pentode mode, two tubes (PP):  +/- 450 plate volts; up to 55W output


Note that Class A operation requires a signifigantly lower plate voltage.  There is much less plate & screen current; and less output power in Watts.  Class AB has much higher plate voltage with much higher plate & screen current.  The tubes can handle this, because they alternate doing the work; so each tube regularly cycles through a cooling-off period. 


In class Class A (biased for clean operation as PRR points out) the tube(s) draw nearly full current all the time.  I.e., idle current is almost the same as full signal current.  In Class AB, the tubes' idle current is less than in Class A.  More current is drawn only to match the desired signal strength.  Therefore, Class AB is more efficient than Class A.


For guitar amplifiers we don't care about efficiency, but rather about tone.  These modes of operation sound different, especially for overdrive tone.  Class A tends to have a vintage tone, Class AB more modern.


It is not correct that Class AB is twice as loud.  Class AB can have more than twice the output power in Watts.   But volume, or loudness, follows a logarithmic scale in decibels.  I.e., twice the volume requires 10X the output power, not 2X or 3X. See, www.guitarnuts.com

[tubeswell]

In Class AB there are two load resistances.


In your 2-tube 'brain model', the A-load-line occurs when both tubes are conducting, whereas the steeper B-load-line occurs when one tube is in cutoff and the other tube is conducting.


For the 'B load line' condition, the tube that is conducting sees 1/2 x as much load resistance through the OT (than both tubes would see in the 'A load line' condition). This situation is made possible because of the way the output tubes are biased and the amount of signal swing at the output tube grids. When the signal from the PI is sufficiently big enough in relation to the output tubes' bias voltage, it causes one of the tubes to go into cutoff when that tube's grid is deep into in the -ve part of the signal swing cycle. This 'shuts off' that tube, effectively rendering the load resistance that is 'seen' on that side of the OT primary, absent.

[HotBluePlates]

... what I'm wondering is if my tubes (2 class A,or 2 class AB, same type) having a gain of say 60 , how is class AB twice as loud? ...

Voltage gain does not equate to power output. Power equals Voltage * Current. The current of a 12AX7 stage is very small, maybe swinging 0.5mA. The voltage gain of a 6L6 from grid to plate might be ~7-8 (45v peak a.c. input for 350v peak a.c. output). But the a.c. current swing might have a peak of well over 100mA.

So your process for preamp voltage amplifier stage loadlines might have little bearing on a power amplifier stage at the output of an amp.

Tubes operating in Class A never stop conducting current; they're working all the time. Tubes operating in Class AB turn off for part of the time, so they get short breaks. So the Class AB tubes can swing larger voltage and current peaks, because the rest period offsets the higher plate heating to keep the tube within its dissipation limit. Greater output power is the result.

When running in the Army, our quarterly requirement of 4 miles in 36 minutes (9 minute/mile pace) felt like I almost couldn't make it to the end. But at the same time, I could run 2 miles for a PT Test in 16.5-17 minutes (8.25-8.5 minute/mile). When we did 1/8th mile wind sprints on the track, I could cover that ground in 35-40 seconds (4.67-5.3 minutes/mile).

I could do more work during the wind sprints because I got a breather between each sprint; no way I could ever run a 5-minute-mile over a sustained distance. The 4 mile pace was "Class A" and the sprint was "Class AB".

[shooter]

Thanks guys, that does help a lot.  I've been able to cut n paste pieces of schematics n they all work, there is like 90yrs of design already out there, I'm just trying to understand the why's n this is the best place I've seen for truly great knowledge.  I joined the Navy, they didn't make me run or catch bullets!!  Thanks again.

[PRR]

> having a gain of say 60

Why do you keep saying "gain"??

Power output and efficiency has nothing to do with gain.

Two identical cars, but one has a booster on the gas pedal so it floors at a touch. Which one goes faster? Which one uses less fuel? While a super-sensitive pedal may require driver re-adjustment, the car won't go faster and (for the same speed) won't get different efficiency.

> how is class AB twice as loud?

Not "twice as loud", but whatever.

The tubes can idle much cooler. Therefore we can often raise the B+ or lower the load impedance (raise the current) without cooking the tube.

[shooter]

I used "gain" in the context; if I have 10vac in, 100vac out class a, (gain of 10), what would I expect out class ab for the same 10vac n why? (I'm assuming same 2 tubes in both A and AB running at 90% max plate dissipation).  Where all this is leading for me, I have a good working/sounding quadSE amp, I am thinking of making an amp section like a Vox ac30? (4tubes, 2 sets paralleled in AB) and I'm just trying to understand the differences on the AC side of things.

[jjasilli]

See:  http://www.enjoythemusic.com/magazine/sound_practices/2/triode_pentode.htm


I used "gain" in the context; if I have 10vac in, 100vac out class a, (gain of 10).  The term Gain is not used "in context".  It has a definite meaning, exactly as your example shows:  it is a multiplication factor.


"The term gain alone is ambiguous, and can refer to the ratio of output to input voltage, (voltage gain), current (current gain) or electric power (power gain). In the field of audio and general purpose amplifiers, especially operational amplifiers, the term usually refers to voltage gain. . ."  http://en.wikipedia.org/wiki/Gain  This I think is what PRR is refering to.


The significant thing about power tubes is not so much gain, but their ability to handle high wattage.  E.g., a 6L6 can dissipate about 30 Watts. Watts = Voltage X Current (a version of Ohm's Law).  The power tubes achieve their high Wattage with High Voltage X Low Current, on their side of the OT.  The OT performs a mathematical operation on this circuit:  it maintains the Watts but converts to Low Voltage X High Current output, to drive loudspeakers.


Power triodes can also handle relatively high watts, but need lots of signal voltage at their input.  The extra grids in tetrodes and pentodes make them more efficient.  A significant benefit is that they can be driven with relatively low doses of signal voltage, easily provided by a preamp tube.  No need for expensive and bulky interstage transformers, etc.

[PRR]

You have a string of little tubes to give voltage gain. You have a VOLume pot so you can get the gain you want.

Put in your other output stage and adjust the Volume knob as needed.

There's a difference in gain but it is too small to fret about.

[shooter]

Thanks guys for workin me through the language of tubes, hopefully I will have intelligent questions next time around.

[JPK]

Shooter thanks for asking this. It made me go back and study it again. Every time I go back and study it I understand it a little better. I Google'd it and found an article that I have seen before. I'll post a link.

http://www.mesaboogie.com/US/Smith/ClassA-WebVersion.htm

[shooter]

I read the link - 2wice n all the help here.  I know I probably sound like a rock but I was missing 1 piece that kept buggin me n that was how the AC signal got amplified more in AB.  with everyone's help I do get it now!
Thanks again, this is gotta be the best knowledge base on the net!

[jeff]

I'd really like to see what everyone says on this take but here goes:
I like to think of it in terms of eveolution. 
 In the begining there was the single ended amp. one power tube.That tube reproduces the entire signal positive and negative. But the "problem" with using one tube is that it's not perfectly linear. Inotherwords the output is asymetrical. That's just the way tubes work.
 OK so how can we fix that? Enter the push pull. Let's take the input signal and make an inverted copy of that. The original signal is amplified by one tube and it's output is asymetrical. The inverted signal is amplified by the other tube and it's output too is asymetrical. The output of one tube has one half of the wave a little bigger than the other. The output of the second tube also has one half of the wave bigger than the other. But since the second tubes input was inverted(from the phase inverter) and it's output is inverted(from the output transformer) both those signals are asymetrical in opposite ways. Those two outputs are combined resulting in an output wave that is symetrical. (one tubes asymetry is ballanced by the other tubes asymetry. Each tube is, just as in our single ended amp, amplifying the entire singal. So if we can swing from 0 to X we must bias the tube at 1/2X. Starting at 1/2X we can swing up to X for the positive half and down to 0 for the negative.
Well if we are gonna use two power tubes anyway let's see if we can get more power.
 Class B. With class B we can bias the tubes differently and get more power. Instead of biasing both tubes at 1/2X and swinging from 1/2X to X to 1/2X to 0 bias both tubes so that they're off. The top tube starts at 0 and during haf the input wave swings from 0 to X and back, but does nothing during the nevative half of the input wave. The bottom tube is also starting at 0, but because it's input is in verted it is only on when the other is off and is ampilfing the other half of the input wave. It's output is also swinging from 0 to X. That output is reinverted through the output transformer and combined with the top tubes output. So one tube is amplifing one half of the input singal from 0 to X, and the other is amplifing from 0 to X the total output is larger: 2X. Think of it as one tube amplifing 0 to X back to 0 then the other tube taking over and amplifing from 0 to (seemingly because the inversion and reinversion) to -X. the whole swing is twice as big.
But tubes aren't too good going dead stop to full on. So they are biased so they never completly shut off: ClassAB. One tube does most the work amplifing the positive half the other on the negative half.

So Single ended A: One tube's amplifing the entire input signal(starting at 1/2X up to X down to 0)
Push Pull A: Both tubes are amplifing the entire singal(Each swinging from 1/2X up to X down to 0)
Push Pull AB: Each tube mainly amplifies only it's half of the signal so each half of the output singal can swing more than 1/2X.
That's my take on it. I'd love to hear comments both positive and negative if you agree of if I've got this wrong.
(I crossed this out because I think I got it wrong and I don't want to screw anyone up)

[tubeswell]

So Single ended A: One tube's amplifing the entire input signal(starting at 1/2X up to X down to 0)

In either case; an 'A' load line. The tubes are conducting all the time, whether its a single-ended amp or a push-pull amp

Push Pull AB: Each tube mainly amplifies only it's half of the signal so each half of the output singal can swing more than 1/2X.

The 'A' load line exists when both tubes are conducting.

The (steeper) 'B' load line exists in only that part of the signal cycle when the tube(s)* on only one side of the OT primary is conducting. (This is when the on-tube's plate voltage is in the 'downward' part of the cycle). This is because the reflected load through the OT is halved, resulting in the steeper load line. The steeper load line results in more plate current for a given grid voltage swing when the plate current is highest (on the 'downward' part of the plate voltage swing), resulting in higher output power from the 'on-tube(s)*' when it it the only side that is conducting. The on-tube(s) is able to operate comfortably in this condition because it/they are switched 'off' for another part of the signal cycle - quid-pro-quoing the overall plate dissipation.

* singular or plural depending on whether or not you have several tubes in parallel on each side of theOT primary.

[shooter]

I've settled on my 1st PP build, I have the A vs AB straight in my head - I think, I still have a lingering question.  Let’s say the fixed bias is set for -30 n I’m using push-pull.  Does that mean as long as the input AC signal is less than 30Vac Peak the tubes will be operating in class A?

thanks all for the lessons.

[jjasilli]

Jeff: your analysis re linearity is more apt for preamp tubes, not power tubes.  Running a headphone, which has a tiny drive,r can be done with preamp tubes.  Also, small driver can be placed in a horn, but horns pose a new set of technical issues and limitations.  Hence the need for a loudspeaker; but this requires significant power.  Here, the beginning is power triodes.  They have great linearity, but lack power. They also have a lot of internal capacitance.  Due to the lack of power, they have drawbacks: they need a huge input signal; SE requires huge OT's.  And their output is low so they were still used to drive horns; especially because loudspeakers were very inefficient back then.  Horn drivers are very efficient and responsive because of their low mass.  With today's highly efficient loudspeakers, triode power amps are making a comeback in a niche of the hi-fi world, because of their linearity.  Tetrodes and pentodes are less linear; but they are more powerful and have far less internal capacitance.  IN PP, they have tremendous power, require small OT's, and have huge noise & hum cancellation.

[HotBluePlates]

Let’s say the fixed bias is set for -30 n I’m using push-pull.  Does that mean as long as the input AC signal is less than 30Vac Peak the tubes will be operating in class A?

No.
 
If the bias voltage is -30vdc, then an input signal which reaches 30v peak causes the control grid to just touch 0v. At/Above 0v, the grid stops looking like an infinite impedance, conducts current and causes the phase inverter's output signal to collapse (the inverter is typically designed assuming the output tube grids are infinite impedance).
 
Looking at one side of the push-pull output, you would need the inverter to deliver 30v peak/1.414 = 21.2v RMS.
 
If you look at the output stage the way your inverter does, with the input being the total voltage from push side to pull side, you need 60v peak-to-peak or 42.4v RMS grid-to-grid. This way of looking at the problem could be important depending on which type of phase inverter you're using.
 
When the output tubes transition from Class A to Class AB is more a function of the supply voltage and load impedance, not bias voltage.

[shooter]

Ok, another question, do I want to limit the input signal to the PA stage so that (G1=0 or +) is avoided, or just reached?  If so how does PA distortion, clipping/saturation occur?   Thanks all for the insight.

[HotBluePlates]

do I want to limit the input signal to the PA stage so that (G1=0 or +) is avoided, or just reached?

You would design the output stage to make the desired output power, using available tubes, output transformers and power supply voltage/current. The choice of output tube, supply voltage and class of operation (which impacts power output) will be contributing factors that determine the output tube bias voltage.

The phase inverter for most guitar amps is designed to deliver a peak output voltage equal to the bias of the output tubes, assuming a fairly high load impedance (often 100k-220kΩ from the output tube's grid to ground or fixed-bias supply). The phase inverter is often designed to have to reserve capability beyond this output so that the amp's clean output power isn't limited by distortion in the phase inverter (meaning the output tubes should run out of steam and distort first).

Often, the phase inverter is designed to deliver its required output with an input of about 1v. So the balance of the preamp is generally designed so that with an average guitar input signal, and after considering losses in tone circuits, etc., it will deliver 1v to the phase inverter with the volume control set at about half- to 3/4-full volume. That gives wiggle room for weak pickups or a light touch.

do I want to limit the input signal to the PA stage ...

So in other words, you're not limiting drive to the output tubes but insuring you have ample drive even with weak pickups.

When the output tube grids get driven positive, they will limit and clamp the drive signal themselves. Once positive, the output tube grids draw grid current. The phase inverter generally is not capable of maintaining its voltage output while also delivering this current, so its voltage output will collapse. Class AB2 and Class B2 amps are designed to drive the grid positive and source grid current, and are pretty uncommon.

... how does PA distortion, clipping/saturation occur?

If a tube had an output signal exactly like the input, only bigger, it wouldn't distort. If the tube's characteristic curves were perfectly straight, equidistant lines, the tube wouldn't distort. Outside of very small input signals, neither of these cases happen. So even before a tube runs into some hard limit, it will distort to some degree. It's just that distortion might increase dramatically as a hard limit is approached.

[PRR]

> do I want to limit the input signal to the PA

No. That's what volume/gain knobs are for.

[shooter]

thanks hotblueplates, that's great info, I have my tubes, tranny, n bias-point picked, still dialing in the pre's one last time.  I fixed a ga5t with an interstage tranny used as the PI n loved how simple it was n how the tremolo modulated the center tap.  I found one on the web for like 35$ but I gotta get some answers from the maker.

"PRR    No. That's what volume/gain knobs are for." 

the reason I asked is cuz, well, all the guys that play through my builds the first time start at about 5, go to 10, then find some happy place between them.  I just didn't want to draw enough +grid current to let smoke outta stuff.

Thanks again for the help

[PRR]

> I just didn't want to draw enough +grid current to let smoke outta stuff.

With capacitor-coupled drive, NOT a problem.

> a ga5t with an interstage tranny

With the usual driver tube before the interstage transformer, probably not a problem.

Yes, there are the 300 Watt Fenders with a 6V6 driving the innerstage. Well-designed and not known to smoke too much. If you hay-wire a supersize transformer drive, open the windows.

BTW: "immolation".

[shooter]

!  My summer builds are in the barn!!!  I was gonna cap couple the IT at both ends, found a link that shows how.  I started the PA and PS schematic last night, hopefully i'll get it posted for comment in a day or so, most is "stock" but I have a couple questions on my PS choices.  Thanks again for the help.

[HotBluePlates]

I was gonna cap couple the IT at both ends, found a link that shows how.

If you have to use coupling caps into/out of the interstage transformer, the circuit probably doesn't need the interstage transformer.
 
Said another way, you would only use an interstage transformer with a compelling reason to do so; if cap-coupling is useable in the circuit, you probably don't have a compelling reason to deal with the problem the transformer might cause...

[HotBluePlates]

I did a quick look at GEC approach, a book which includes several circuits with interstage transformers.  ... unless you are planning to run Class B ...

The GEC Approach mostly shows very powerful Class AB₂ or Class B₂ stages for a lot of output power. You might have an KT66 feeding the interstage transformer on those to drive the grid current of the bigger output tubes.

I think Shooter is talking more along the lines of the Gibson BR-3, which uses a transformer instead of a phase inverter tube stage. You can get perfect output balance and save a tube/components, but you're more susceptible to hum pickup by the interstage transformer (and now they're expensive).

[HotBluePlates]

...   Regarding interstage transformers (IST) as PI's there was solution to IST by using a pair of PP O/T's.  It was in one of the GE publications.  Its worth exploring if one is building a frankenstein amp, or breadboarding.  ...

I don't think you're following... a normal interstage transformer is the size of a Fender reverb transformer. So anything to do with re-used output transformer is tremendously wasteful. And the resulting impedances are not the right values.


But that trick would be useful if you're making a huge amp which drives the output tube grids positive. The interstage transformers in the GEC Approach circuits are single-ended output transformers, with a center-tapped secondary.

[shooter]

I was gonna use the IS tranny just because, the one I found is on triodes website, $35, can be set up as 1:1 or 1:2, I'm doing an AB1 kt88 PA section, sorta Marshall master PA/PS sections.  I think i'll start 1:1 n re-wire 1:2 if it needs more grid drive.  I'm re-working the pre sections, 2channels now, which unloaded by a PA section give me about 100Vacp-p.  I'm thinking grid bias will be in the 45-55v range, PA B+ around 450, G2 @300. I'd like to have PA "distortion", clipping/saturation, from 8-10 on the volume, and Pre distortion around 7-10 on gain  have all the pieces sept the IS tranny.  At $35 it's worth "seeing what happens".
Thanks again for the inputs

[shooter]

I searched the threads on VVR, understand the electronic aspects of a voltage regular but not sure the "whys" as it relates to sound.  Does it give you a "brown" tone?  Is it left to the end user to play with, ie the pot is accessible?  thanks for the help

And I am using the IS tranny as my phase invertor similar to the Gibson

[HotBluePlates]

I searched the threads on VVR, understand the electronic aspects of a voltage regular but not sure the "whys" as it relates to sound.  Does it give you a "brown" tone?

Forget the "regulator" part of the name for now; focus on "variable voltage".

One permutation of the equation for power is Power = Volts²/Resistance. "Resistance" in this case is the load impedance presented by the output transformer primary, which doesn't change. So if you're able to to cut the supply voltage by half using the VVR, that cuts power output to 1/4.

The point is if your preferred flavor of distortion comes from the output tubes, the reduced supply voltage available from powerscaling/VVR let's you get that at reduced output power and volume.

Is it left to the end user to play with, ie the pot is accessible?

Yes, otherwise it wouldn't be a variable voltage regulator. Think of it akin to a master volume that controls the amount of power the output section can make.

[PRR]

> not sure the "whys" as it relates to sound.

If you feed a Bassman 5F6A the usual 400 Volts it distorts at 45 Watts. Say with a certain speaker and room this gives 113dB SPL.

If you feed it 100 Volts it distorts at something less than 2.8 Watts and 101dB SPL. A similar overload but at much lower volume.

Typically the knob goes down to less than 1 Watt, though designs scaled for 400V don't act really the same at 50V.

And there is a built-in human-body reaction (fear?) to HIGH acoutic levels which happens less at low acoustic levels.

It is a valuable tool.

[shooter]

Thanks guys, that does clear the mud. I think i'll work that into my next build, after some more reading on component specs.