6L6 headroom, power, and saturation

[Diablo]

Can someone explain the effects of swapping various types of 6L6 type tubes into a guitar amp on watts delivered, headroom and saturation (power tube distortion). For example, if I start with a push-pull amp using 5881 tubes (rated for 23 watts max) and bias the idle to 70%, is it reasonable to assume that the amp will deliver around 46 watts to the speakers? Now, if I stick in 6L6GC tubes (rated for 30 watts max) and bias the idle to 70%, does the amp now deliver around 60 watts to the speakers? Does tube saturation start to occur at a higher speaker volume with the higher wattage tubes? What would happen if I lower the bias on the 6L6 tubes so that the idle current is the same as the 5881 tubes? Would the amp now be as loud as with the 5881 tubes but a lot cleaner?

[Chuck H]

Different tubes sound different and also distort different. ie; Groove tubes number system rates tubes on how early they begin to crack up. Same tubes, same brand, different breakup characteristics. Different plate voltages will also effect tube performance. Not just current and watt rating. If you want a "dirty" tube, pick a lower wattage tube that is not known for it's clean headroom. If you want a "clean" tube, pick one that is of a higher wattage rating and is known for it's louder, cleaner sound. Guitar amps tend to run the tubes a little on the warm side. You could probably bias a little cooler and expect an improvement in headroom much of the time.

Chuck

[R.G.]

You're making a classic mistake. It doesn't work the way you have in mind at all.

The power a particular power amp can put out depends first on the power supply it's working from. If the power tubes could magically transfer 100% of the power from the power supply to the speakers, then the power to the speakers would be limited to 100% of the power supply capability. Since that is impossible, the power to the speaker is only that fraction of the power from the power supply that can be converted to audio and not wasted as heat or unused. The idle bias point of the tubes is wound up in this, but not as you think.

A power tube like a 6L6, 5881, or any other has some maximum ratings. For each of them there is a maximum power supply voltage they can work with, a maximum current which they can conduct, a maximum power they can dissipate without burning out, and a minimum voltage they can "saturate" to.

If you exceed the maximum power supply voltage, the tube arcs over and dies. If you exceed the max current, the tube's cathode overheats, gets poisoned, and quits emitting electroncs soon (dying in the process). If you exceed the max power dissipation the tube overheats, and in the extreme case melts its glass envelope (I've seen this) or simply quits. And the tube simply won't get to lower than the minimum voltage under load.

The minimum voltage under load always subtracts from the voltage available from the power supply, being eaten by the tube instead. This is one of those unavoidable losses.

A pair of tubes in push-pull can be biased various ways. They can both be biased completely off without signal; this is Class C, and each one only conducts pulses of current when the signal turns it on. This is the highest efficiency operation, and results in the maximum amount of power supply power moved to the output. It also sounds so ugly that no one does it voluntarily for audio, although it's used for RF power amps. Since the standing current of the tubes is zero, the only heat they make is when they conduct. So in this configuration, you can design the amp to have the maximum voltage possible for the tube, let it hit peaks of maximum current, and as long as the every-other-cycle power doesn't melt it, it can supply max power. They are biased to 0% - and hence useless.

So we reduce the negative grid bias a bit until each tube is barely off. Static dissipation is still zero, but each tube now conducts alternately for a full half-cycle of signal. Audio amps sound ugly like this because the crossover between one tube conducting and the other tube conducting is not linear, and sounds ugly. Since the tubes conduct for a longer part of each half cycle, the max power supply voltage and current have to be lowered a bit by recalculating the output transformer to keep dissipation limited to what the tube can stand. Notice that the no-signal power in the tubes is zero (well, OK, heaters only) and the full signal power is the maximum per tube.

For better audio we make the grid bias a bit less negative so the two tubes both conduct for a portion of the signal cycle - they overlap. If we set the standing dissipation to 10%, 50%, etc of their max dissipation, then we have "biased to X%". Each watt of standing dissipation you bias to is at least partially subtracted from the total amount the tubes can transfer to the secondary. You have used up some of the power transfer capability of the tubes to make them more linear. And because of using up some of their available power dissipation, we must decrease the power supply voltage; otherwise, we have made an automated tube-killer. Less audio power gets out, but it sounds better. This is class AB. It's a whole continuum of bias points from just barely turned on at the same time to both on almost all the time.

In the end, we bias both tubes to be on for 100% of the time and conduct 50% of their maximum current in the absence of signal. Each tube is "biased to 100%", and never fully turns off. This is class A. As you might expect, we have to lower the power supply voltage compared to class c, b, and AB to keep the tubes from burning up. Class A sounds best, but wastes the most of the available power supply power and most of the power tube dissipation capabilities as waste heat.

By now you're appreciating that a power amp design includes certain expectations about what tubes go in it, what currents, voltages and powers they can withstand, and how it will work. While there is a large region of overlap of Class AB bias points, you run some risks if you simply decide to bias to X% and the power amp designer didn't have that in mind when he set up the power supply.

You're also appreciating that biasing to X% is a statement of how much overlap the tubes have and that increasing the X% decreases the available audio power out of the tubes. And that a different kind of tube may have a different compromise of power dissipation, minimum voltage, maximum voltage and power dissipation (or, as I often put it, ability to withstand abuse).

Now to your questions.

Can someone explain the effects of swapping various types of 6L6 type tubes into a guitar amp on watts delivered, headroom and saturation (power tube distortion).

Watts delivered depends heavily on bias and raw DC power supply. For a specific DC power supply voltage:
- the power dissipation ability of the tube tells you how much you can bias the standing dissipation of the tube; since B+ does not change (much) as you draw more current, each watt you bias to leaves less for audio output, and moves you closer to burning out tubes
- the maximum headroom and power out is with the tubes biased just to 0%; this leaves the maxim power supply voltage and current to be transferred and dissipated by the tube. But it sounds ugly.
- the higher the bias percent, the less that's left for the tube to transfer, but the less the no-signal to full-signal power dissipation changes. Just at Class A, the no signal and full signal power dissipations become the same.
- the closer to class A (i.e. the higher percent of standing dissipation you bias to) the lower B+ can be and have your tubes survive; If B+ does not decrease, you must change the output transformer ratios to keep currents down with increased signal. Most people change the power supply.

For example, if I start with a push-pull amp using 5881 tubes (rated for 23 watts max) and bias the idle to 70%, is it reasonable to assume that the amp will deliver around 46 watts to the speakers?

No, as explained above.

Now, if I stick in 6L6GC tubes (rated for 30 watts max) and bias the idle to 70%, does the amp now deliver around 60 watts to the speakers?

No.

Does tube saturation start to occur at a higher speaker volume with the higher wattage tubes?

Maybe. It depends on their minimum voltage under load and their cathode emission characteristics, as well as the bias point and power supply. It's not a one- or two-dimensional problem space.

What would happen if I lower the bias on the 6L6 tubes so that the idle current is the same as the 5881 tubes? Would the amp now be as loud as with the 5881 tubes but a lot cleaner?

Yes. But crossover distortion would come up.

Headroom increases as you dial bias down (i.e. to a lower percent dissipation) and use the increase in power supply voltage you could hypothetically do then. In practice, no one changes the B+ voltage. We just noodle around and endlessly (and incorrectly) muddle about who has the right to do what and with which and to who.

[Diablo]

Thanks R. G. for your thoughts. So, for the executive summary.....set the bias lower and get more power to the speakers and more headroom. The reason I asked the question was to make a decision about which types of 6L6 to select for a 5F6A amp. Stock is a 5881. Your answer helps me to understand that I probably want to stick with a 5881 for this amp.