PLate voltage and power tube wattage.

[Gregg]

When measuring output power it's usually a sine wave before clipping or full "clean" power.

[Merlinb]

Bias has nothing to do with output power. You could bias the amp to class B (zero idle dissipation) or class A (way high idle dissipation), and it would still give the same audio output power in either case.

To a first approximation for pentodes, output power is:

P = 2 * HT^2 / Rload

Where Rload is the anode-to-anode transformer impedance.

(It will be less in practice)

[R.G.]

Bias has nothing to do with output power. You could bias the amp to class B (zero idle dissipation) or class A (way high idle dissipation), and it would still give the same audio output power in either case.

Let me expand on that a bit.

Bias does have something to do with output power, but it is a backwards and subtle relationship, not the idea that output power is directly related to output power.

To start with an extreme simplicity, output tubes gots limits. They can only conduct so much current, withstand so much voltage, and dissipate so much power before they die. These limits, and how the tubes are set up within them, determine output power. Bias is a part of that mixture.

Tube datasheets usually show a maximum B+ voltage. This is generally at most half of some vague limit not specified where the tube can't reliably hold off the voltage applied from anode to cathode. It is (well, OK, was) done this way because output tubes are nearly universally used with output transformers, and the use of an output transformer nearly always subjects the tube to 2x the B+ voltage one way or another. In general tubes can withstand bigger B+ much better than they can cope with more current or more power being dissipated.

In single-tube Class A, the tube must be biased to at least the peak current that will flow in the primary on peaks of signal, so it can have that current stored in the OT for signals of a polarity where the tube is turning off. That means that the tube has to conduct twice that amount of current when it's fully on on the opposite polarity, so the bias current for single tube class A is half the maximum current the tube can support at that B+ without melting down from the power. This bias current times the B+ has to be less than the maximum power rating of the tube.

To get more power out per tube, you use two tubes in push pull. In push-pull Class A, the tubes still get biased to the full output current, but now you get active drive on both polarities of signal swing, so one tube turns off just as the other tube hits its full current. The output power is doubled, but you're still limited to the amount of power dissipation each tube can handle. In fact, the power dissipation per tube is the most pertinent limit, as in general tubes are better at resisting voltage than dissipating power.

To get more power out of a push-pull amp, you go to Class AB. This means reducing the idle current (by increasing the reverse bias voltage) so each tube has a lower idle power. This lets the designer increase the B+ voltage so the tube can do the same peak current, but swings wider voltage range. This makes more audio out, but introduces crossover, the point where the tubes go from both tubes conducting to only one conducting.

To get even more audio power out of push-pull AB amps, you increase the plate voltage but also increase the reverse bias, which decreases the idle current again, but again lets the designer raise the B+ supply, so the peak current of the audio signal stays the same, but the peak voltage out gets bigger. This makes more audio power out.

It pushes the crossover point down into the audio signal near the actual zero crossing poitn of the audio signal. The more you bias toward true class B, an idle current of zero per output tube on average, the more of the tube's dissipating abilty is used to produce audio power and the less that is wasted simply idling.

There is a more or less smooth progression from Class A pushpull through Class AB to Class B where by decreasing the idle current on each tube (lowering the "bias to X% of tube maximum power" thing) you can increase the B+ voltage and spend more of the tube's power dissipation ability on the putting out audio instead of just sitting there toasting.

So bias is RELATED to output power, but it's inversely related, and all wrapped up in how much B+ you can use. If you bias the tube nearly off (low idle current, low % of maximum tube dissipation just sitting there with no signal) you can use a high B+ without the tubes melting down and get more audio power out from the same tubes. As you increase the idle current (decrease the reverse bias voltage), you have to back down the B+ to keep the tubes from melting.

This is a smooth progression of balancing idle power and B+ versus audio power out. The dependency on audio power out and B+ is the big issue. Bias is the "oh, yeah, you gotta touch that up" thing that lets you use bigger or smaller B+ and get out more or less audio power. Or look at it the other way. If you simply must bias to X% of the tube power, this is an iron clad limit on what the maximum B+ can be before the tubes melt down, as you're using more of the tube's maximum power just sitting there instead of on audio swing.

[Merlinb]

Bias does have something to do with output power,

Since your statement might cause confusion by contradicting mine, let me re-write:

Bias (between class A and class B) has nothing to do with output power.

Bias has to do with efficiency, i.e. idle power. This also affects reliability.

To get more power out of a push-pull amp, you go to Class AB.

No. To get more efficiency you go to class AB. i.e. you run the tubes cooler, and so watse less idle power. But you could still bias those tubes to class A, and you would not get any more audio output power, just hotter tubes (possibly destroyed tubes!).

So bias is RELATED to output power, but it's inversely related, and all wrapped up in how much B+ you can use.

Yes, because efficiency and reliability matter just as much as audio power. You have to juggle the three.

[potatofarmer]

You also have to consider the screen voltage and the plate-to-plate impedance of the OT. It's typical in guitar amps to keep the screen voltage around the same as the plate voltage, but larger-bottle pentodes like the KT88, KT120, GU50, etc more typically have much higher allowable plate voltage than screen voltage. From what little I've seen in the HiFi world it seems the more common way to use these tubes is to have a separate power supply for the screens, holding them at 350V or 450V or what have you while letting the plate supply hover in the 600V - 800V range.

Conversely if you have a quad of KT88s with 420V on the plates, you won't get anything near 120W output with only 50V on the screens!

edit: not really a reminder for misters Merlin and RG, so much as us shmoes in the peanut gallery

[R.G.]

Since your statement might cause confusion by contradicting mine

Sorry, didn't mean to confuse people by contradicting you.

If it's any help, you and I are describing different views of the same elephant.

Bias (between class A and class B) has nothing to do with output power.

Yes, that's a clearer statement of what I think you meant. Bias has nothing much to do with output power *at a given B+*. B+ and tube current/power limits set what power you can get out. Increased bias can "use up" some of the useful current range of a tube.

Bias has to do with efficiency, i.e. idle power.

Again, better. In my view, bias has to do with moving the crossover point around, and you pay for that nicety in higher power wasted just sitting there, but yeah, more wasted power. Different part of the elephant.

This also affects reliability.

Yes, heat is the enemy, no matter where it comes from. There are three ways in general to getting more power: generate less waste heat, spread it out where no one part is being cooked unequally, or get rid of it better. But heat/temperature can be dealt with by better cooling, up to a point without making reliability much worse.

No. To get more efficiency you go to class AB. i.e. you run the tubes cooler, and so watse less idle power. But you could still bias those tubes to class A, and you would not get any more audio output power, just hotter tubes (possibly destroyed tubes!).

No. To get more power from the same tube complement, you raise B+. At the same B+, bias is only a minor issue in that you can waste the tubes' ability to dissipate power by biasing them hotter. Increased bias does also remove part of the audio power swing by overlapping the tube conduction, but this is a minor contribution.

As I noted, it's a big and complicated elephant. I realize that my view is tainted by my background as a designer, not a repairer or amp builder from preexisting parts. In design work, everything is a variable, and I have no qualms about considering B+ to be anything I want it to be. A repairer or modifier thinks B+ is only variable by a small amount, as the PT is generally cast in stone already. Only expensive replacement of the PT will change B+. When you're stuck with a single B+ that can't be varied much, the range of options narrows.

Both "no" answers come from different views of the elephant.

Yes, because efficiency and reliability matter just as much as audio power. You have to juggle the three.

All three matter, and yes, you have to juggle. But a given situation may require that they not matter equally.

Efficiency, reliabilty, and audio power are a three way trade-off, and in any given situation one or two may be traded off to get more of a third, or you may try to push one to a limit to get more of the other two. Here are some views of that somewhat different elephant.
- Amp techs are all familiar with a situation where a guitarist brings in his cooked-to-death amp for repair. At some point they say something like "... and it sounded really great just before it burned out. Can you fix it so it does that all the time?" Guitarists (which is why we do this) tend to not care much at all about efficiency - they only want it to be loud enough and get them through the gig. Well, until it dies.
- Once the amp dies, guitarists become very interested in reliability. They don't know what that means, but they now like it. But a guitarist will have only a dim idea of what amp to buy based on reliability, and this is based on their friend's story that they once had a [insert amp here] that lived forever.
- Likewise, a guitarist would sometimes sell his soul, not just his amp, to be able to cut through the mix better.

All these descriptions can be recast at the design for sales stage too. Amps can be and are optimized to many different mixes of efficiency, reliabilty, audio power, and many other factors.

Knowledge is power - and sometimes you can convert that into audio power.

[Merlinb]

As I noted, it's a big and complicated elephant.

Curiously, I would say it is a very simple elephant:

The HT voltage and load impedance together determine the output power (assuming pentodes with sufficiently high screen voltage).

Bias determines class of operation.

If you're designing from scratch, you choose all three to obtain acceptable audio power while ensuring your chosen tubes don't operate off limits.

If you're working with an existing amp where you can't change the HT or load, you can only use the bias to ensure the tubes don't operate off limits. This does not significantly change the output power.

[R.G.]

OK, it's a MEDIUM SIZED elephant...

The HT voltage and load impedance together determine the output power (assuming pentodes with sufficiently high screen voltage).

I see that as the HT/b+ and the tube current/power capability determining the output power, and the load impedance being a consequence of those. The secondary loading and transformer ratio are chosen to allow an impedance on the tube plates to let them be all they can be, as the US Army recruiting adverts say. But, either way. Different faces of the same underlying physics.

Bias determines class of operation.

... and also allows more or less of the output tube voltage/current range to be used within the limits set by a fixed B+/HT. And allows you to cook the amp by running the output tubes up into eating more of the available power heating themselves instead of shoveling it into the secondary load. Same underlying physics.

If you're designing from scratch, you choose all three to obtain acceptable audio power while ensuring your chosen tubes don't operate off limits.

If you're working with an existing amp where you can't change the HT or load, you can only use the bias to ensure the tubes don't operate off limits. This does not significantly change the output power.

Yep. Now we're both looking at the same end of the elephant. Mother Nature wins again. The elephant is as it is, no matter how we perceive it.