Fixed bias, and power output.

[leadfootdriver]

Does one affect the other? If I bias a "100 watt" amp with 4-EL34's at 70% idle dissipation, does the amp actually put out 70 watts?

If that's true, could I get an amp the break up faster if I bias at 60%?

[Steve Conner]

To a first approximation, no, the idle power dissipation and maximum power before clipping are independent of each other.

To a more accurate approximation: The hotter you bias the amp, the LESS power output you get, but the better it sounds.

The above only applies to push-pull amps. For SE amps the idle power and output power are related.

[soundguruman]

The output is about the same, biased hot or cool.
You want to bias it hot, it will break up earlier and have more sensitivity. BUT the tubes won't last as long.

[leadfootdriver]

Cathode biased amps are biased at around 100% from what I can tell. I read that it's safer due to the give of the cathode resistor.

With tube life aside, can I safely bias at 100% with fixed bias without damaging the amp?

[J M Fahey]

The output is about the same, biased hot or cool.

No, it´s not.
Output power is AC power, coupled from the tube plates into the speaker.
Plate dissipation is DC power, "lost", not transmitted to the load.
If tubes were not dissipation or current limited, your theory might hold, but in real world power amplifiers it does not.
Here I linked to one set of 6L6 plate curves.
There are many all over the place, I just linked these as a generic example.

Suppose 450V DC idle plate voltage.
It can be seen (in *these* curves) that the plates can swing from 450V to around 50V (interpolated), not much doubt about that.
Now to current:
we see that (in these curves), maximum plate current is 135 mA.
The minimum value? : the idle value we set.
What will the available current swing be? Or delta current if you like.
Simple: Max (minus) Idle.
That´s the audio current swing which will be transmitted to the load.
It´s clear that since we are limited to a Max. value, the higher the Idle current , the lower the available output power, all other conditions kept the same.
Which is the real world case: when we adjust our tubes hotter or colder we don't adjust +B, or output impedance, do we?
As a generic example (using these curves):
If biased to quite cold 10W: Ib=10W/450V=22mA Available current swing=135-22=113 mA
If biased to quite hot 30W: Ib=30/450=67mA Available current swing=135-67=68 mA
Since we are not varying the +B value, nor the Voltage swing, our power will vary *directly* with the available current swing.
When cold biased= 113mA
When hot biased= 68 mA
Reduced Power because of different biasing choices: 68/113=60%
We have now 60% of our earlier power or we lost 40% , take your pick.

Ah !! , but it does not *sound* like I lost 40% !!!
Maybe, ....... probably, ....... but that´s an entirely different affair.
But it sounds much better biased hotter!! Maybe, see above.

[pdf64]

'can I safely bias at 100% with fixed bias without damaging the amp'
Check the PT and OT transformer spec for what continuous current they will cope with; may need an email query to the manufacturer as it's a bit unusual. It's possible that the primary winding of a OT specified for class AB may overheat with a high static dc.

[GainFreak]

Cathode biased amps are biased at around 100% from what I can tell.

Not always. You can cathode bias them at whatever dissipation you want.

[jmaf]

Cathode biased amps are biased at around 100% from what I can tell. I read that it's safer due to the give of the cathode resistor.

With tube life aside, can I safely bias at 100% with fixed bias without damaging the amp?

Class A amps are biased close to 100% dissipation. Cathode biased amps aren't always Class A.

[diagrammatiks]

class A amps will have less plate dissipation as they swing.

The available output power isn't dependent on the bias at all for a class A amp.

Fahley's analysis is more correct...although still simplified over what the actual equations are. No fault on his part...the full set of parameters are just really friggin complicated.

[R.G.]

I have always felt uncomfortable with the whole concept of biasing to a % of dissipation. Especially where there is nothing especially relating % dissipation to the amp's operating conditions.

First, a tube has a fixed (notionally at least) power dissipation it can do without melting down. The datasheet for the tube gives you this. This is the power dissipation INTERNAL TO THE TUBE that gets the tube as hot as it's reliably safe to heat it. Notice that the power in the tube is the sum of the products of voltage and current across the plate-cathode, screen-cathode, and grid(s)-cathode. Of these, the plate and screen are the vast majority. And notice that these are a product of volts times amps (and probably some averaging process based on the signal swings). It says nothing at all about the idle bias of the amp.

An amp's bias can be split into four general regions of operation: Class A, where the output tubes never turn off completely. Single ended amps must run Class A or have very ugly distortion. In push-pull, you can get Class A (neither tube ever turns off in the range from zero signal to full/clipping output swing), Class AB where the tubes alternately turn off for part of the signal cycle, Class B, where the idle current is hypercritically **zero** so that one tube turns off **exactly** as its mate turns on and each tube conducts for **exactly** one polarity of signal, and finally class C, where both tubes are biased off for some portion of the signal swing around zero.

Notice that as you move from Class A to AB, to B, to C, the idle dissipation goes from substantial, to smaller, to nearly zero to really, no fooling zero. No one uses class C for audio because the crossover distortion is so bad. Class B even sounds harsh. So the smallest idle bias used in most audio is Class AB. Class AB lets you smooth over the crossover distortion and get more power out of the same pair of tubes.

In Class A, both tubes conduct all the time. Limiting class A has each tube just at the hairy edge of turning off on the biggest possible signal peaks you can have for the power supply voltage B+. Each tube is biased at 1/2 of the biggest output current it can provide to the secondary (as reflected through the OT) all the time, so the tubes each have essentially all of B+ and half the max current through them all the time, idle or no. You then get the biggest output power out of a given pair of tubes by selecting a B+ which lets the maximum tube current flow on signal peaks, and which does not melt the tubes (i.e., bias them at over 100% of their rated power) at idle. Limiting Class A demands biasing to 100%. The subtlety here is that B+ must be limited too, to keep the tubes from dying.

If you want more power out, you raise the B+ and cut back the idle current. This reduces the static/no signal idle power so it's less than 100% of max dissipation for the tube, and then you can have the tube swing a bigger voltage and current under signal conditions and output more power to the speaker. You can only do this because you've cut back the idle power. So idling an amp which was designed for Class AB at 100% of tube dissipation means they will overdissipate under some signal conditions. They will have shorter lives as a result of the overheating.

The more you reduce the static idle bias current, the more B+ you can put on the tubes and the more signal power you can get out of them without overdissipating the tubes. And, because the tube output current v. voltage curves don't sum to a nice linearity near zero, the more crossover distortion you get. This process reaches the notable peak at pure class B, which has each of the push-pull tubes conducting for **exactly** the positive (or negative) half cycles. In this case, the idle bias is near zero current; it may not be truly zero for other nagging circuit reasons, but it's near enough. In pure class B, you reach the highest theoretically linear amplification possible with the given tubes. You get their ability to conduct their max current, and you get the maximum B+ swing, as well as the maximum contribution of each tube to output power possible. You get the smallest idle power. All of your tube's abilities result in signal power out. It just sounds bad.

When you pick an output tube, you pick a bundle of potential. It can conduct X current, it can withstand Y voltage, and it can dissipate Z watts without burning up quickly. Your choices of B+ and bias point are your choices of how you will use that ability to dissipate Z watts. If you *can* bias your amp at 100% of tube dissipation, then you *can't* get the most audio power out of it without dialing down B+ and running Class A, or taking some chances on burning up tubes. You can also bias it to 100% with a too-small B+ in class A and waste some of the available power potential.

If you want more audio power from the same tubes, you can bias it to a smaller fraction of its dissipation potential. And you must, if you want to get more power. If your amp is designed with a higher B+ for your tubes than class A (as it must be, if they are to live), then you can bias it to some % of its dissipation which is less than 100%. You can choose how little of its dissipation you want, but there is a point where if you go over that %, the tubes overheat. Your amp may be designed so tubes biased at 70% live. If so, you can bias them at anything less than that power and they'll also live, but you'll get more crossover distortion.

If your amp has an even higher B+ for the same tubes, that condition requires biasing to a lower % of the power tube dissipation as a maximum. You may choose to go lower, but not higher without overheating the tubes.

So the common discussion of biasing to X% is at least confusing, and it leads to questions like this. There are some things which limit how high (as a fraction of the tube's dissipation rating) you can go. You can do anything below that. But biasing to X% dissipation is not independent; you must have a B+ and a tube current/power rating that lets you do that without overheating tubes. As B+ gets higher, the max % dissipation must go lower to not overheat tubes.

And finally, you cannot increase the output power by biasing an already existing amp to a greater or lower % dissipation. That limit has already been done for you by the selection of output tubes, B+, and OT ratio. You *can* change the crossover distortion and self heating by biasing it to lower than max idle current levels. You *can* bias it to 100% of tube dissipation, but since any class B action at all (including Class AB) increases tube dissipation on big signal swings, if it's already at 100% and you push it to more with signal swings, it's going to overheat. Tubes being as tough as they are, this may not be instantly fatal, but it's not good for them either.

So I don't like "bias to X%". It's a simple, tidy, easy to understand concept that hides complexity and makes it easy for the uninformed to get it wrong. If you understand the concepts behind it, it's an OK way to look at things, but you have to understand them first.

[GregS]

The available output power isn't dependent on the bias at all for a class A amp.

Perhaps there's something I'm not understanding but I don't see how this is.

First, in a single ended scenario if you bias to either extreme it would limit the headroom and the output power. If biased really cold the tube would cut off on negative input signal. If biased really hot then the signal would be limited by the grid.

In a push-pull example you could not bias too cold anyway or you'd no longer be in class A. However if you biased too hot you would still run out of headroom and output power would be limited. The negative input signal on one side would not be limited nor would the rise in voltage on that side of the OT primary which would be somewhat less than the HT (to be in class A the tube still has to be conducting so voltage will never reach the HT). However the positive input signal on the other side would be limited by the grid potential and such the voltage on that side of the OT primary could not be pulled down in proportion.

It seems to me that to get the maximum output power the output stage needs to be at least somewhat centre-biased in class A to avoid running out of headroom. Not trying to start an argument here. I just want to be sure my understanding of this is correct.

Greg

[diagrammatiks]

Sorry, what I meant to say was nominal available output power.

if the transformer load and the b+ gives you a nominal swing of 2 watts, biasing an el84 at 100 percent idle isn't going to give you 5 watts of output.

You can get less then 2 clean but you certainly can't get more just by biasing.

[GregS]

Ok. Thanks for clearing that up.

[MWJB]

"Since we are not varying the +B value, nor the Voltage swing, our power will vary *directly* with the available current swing." how do you rebias without affecting B+ voltage, or the negative swing at the grids? I've seen 50-60v variation in B+ depending purely on bias conditions. You bias for tone, not for power output, you might only see a nominal change in max, clean, power output but you will hear a change in tone far before you get to limiting changes in power output. Idle current, unless at the extremes, only has a nominal affect on power output in a fixed bias amp.

Biasing a 6L6 to 30W isn't "hot", it's sabotage in a P-P, fixed bias amp (it is even unadviseable in cathode bias). 10W is coolish (13-16W may be regular), <5W...now that's cold. But you might still see rated W RMS at less than 10W idle dissipation. However, you may well detect differences in tone at increments of far less than 2W idle dissipation.

[R.G.]

"Since we are not varying the +B value, nor the Voltage swing, our power will vary *directly* with the available current swing." how do you rebias without affecting B+ voltage, or the negative swing at the grids? I've seen 50-60v variation in B+ depending purely on bias conditions.

B+ varies with bias level, but it is small compared to the whole B+. I think the comment may have been that it's not varying the *designed* B+ level, not that B+ is not varying a millivolt.

A heavily regulated B+ would show no variation with bias levels and as a teaching aid would let the other issues of power dissipation under varying bias levels show through. Sagging B+ in a real-world amp can obscure other effects when you're trying to separate what causes what for understanding.