Fixed bias, and power output.

[Jazz P Bass]

I for one do not know where this "rule" even came from.
Is this what "sounds better" when you bias an amp hot?

[jmaf]

I for one do not know where this "rule" even came from.
I was taught that in a push pull circuit, if you observe the output graph, a change in grid voltage produces a change in plate voltage.
If you apply, say, a 3 volt ac signal to the grid, the "upper tube will have a change in xx volts.
Now, the other tube will have a slightly different voltage drop.
(look at any output tube graph)
This is in all respects, distortion.
The output is not the same as the input.
Now if the bias is set just into class a/b, that is the "best" the output tube voltage drop mismatch will get.
So, my question is, what is really happening when you bias a push-pull, class A/B amp hotter than "just getting rid of crossover distortion"?
That plate voltage mismatch has to get worse, in my mind.
The upper sine wave is different than the lower wave.
Is this what "sounds better" when you bias an amp hot?

AFAIK all forms of Class AB, AB1, etc, imply some amount of crossover distortion. The way I understand it, the only way for a push-pull stage to not have any crossover distortion is for the complementary pair to work in Class B precisely. Meaning as soon as you cross into negative half wave, the other tube cuts off precisely at that instant and the complementary tube takes over. If either tube continues to conduct during crossover, the mixing of these two swings in the output transformer primary generates distortion.

Now, have you heard a perfectly tuned Class B stage for guitar? It sounds as good as a visit to a hospital. It's like plugging the guitar into a 10k dollar hi-fi stereo, it sounds horrible, clinical...

So a bit of Class AB* crossover distortion sounds good for some reason. I've read tens of opinions on this and I don't have knowledge enough to draw my conclusions yet. Now I don't know that the distortion is what makes the 70 to 80% bias level sound good for guitar, instinctively I think it's the hot tubes that sound good, the usual linear region nearer the positive grid region, etc.

Plus I've had some prejudice against going near positive grid territory, and I stood corrected by kg, Steve Conner, merlinb(even on his book) and other luminaries of this forum. So that's something I really can't comment on but it seems to me that the region near positive grid simply sounds better....

[Jazz P Bass]

AFAIK all forms of Class AB, AB1, etc, imply some amount of crossover distortion.



So a bit of Class AB* crossover distortion sounds good for some reason. I've read tens of opinions on this and I don't have knowledge enough to draw my conclusions yet. Now I don't know that the distortion is what makes the 70 to 80% bias level sound good for guitar, instinctively I think it's the hot tubes that sound good, the usual linear region nearer the positive grid region, etc.
..

Yeah but...
I wasn't talking about "crossover distortion".
Which only accurs at a cold bias. (whatever the heck that means)
Keep in mind the "linear region" that you mentioned, is still a curve.
That is what I am talking about.
When you bias it on a different part of the curve, the upper & lower sine waves are different.
More at a hot bias, less at a cold bias.

[MWJB]

Jazz P Bass wrote: "Yeah but...
I wasn't talking about "crossover distortion".
Which only accurs at a cold bias. (whatever the heck that means)"

Your amp may well make it's rated W RMS even whilst biased cool (>12W?), crossover distortion not being an issue until max clean W RMS is exceeded. Bias hotter and you may delay the onset of crossover, but even biased hot, you will still get crossover distortion at some point. At any point beyond clean W RMS the waveform will distort....whether you go hotter or colder. Crossover distortion is a fact of life with fixed bias, P-P.

[jmaf]

When you bias it on a different part of the curve, the upper & lower sine waves are different.
More at a hot bias, less at a cold bias.

There are no PNP tubes, so the signal fed into the power section has to be inverted to be fed into each of the complementary pair. I know you know this, please bare with me for the sake of this argument. So actually, the two tubes are seeing a signal going up nearer the positive grid region, which is more linear than that terrible region near negative cutoff - down there it's teeth grinding region.

When it's the negative going signal, one tube will cut off and the other will see a positive going signal in direction of that linear region. When it's the positive semi-cycle the other tube cuts off and one tube sees a positive going signal. So both see a positive going signal, there is no upper and lower sine wave, don't know if that's exactly what you were saying, but I wanted to point that out. So for a large signal swing, each tube in the pair is seeing the swing going in direction of that positive region.

If you bias both tubes hot, both of them will work with the signal swing pushing the grid closer to cathode potential, which is the sweet sounding region as far as commonly agreed.

Of course guitar signals are not sinusoidal, they're terribly assymetrical so one tube will always work harder depending on the timbre.... I don't know if I'm being terribly redundant and obvious, I know you know most of this.

[R.G.]

I thought the main reason for 60-80% biasing in fixed bias was that at under 'full signal' conditions the tube is conducting more on average, so you want to allow for a bit of 'headroom' in the dissipation capability.?? Or did I misunderstand something?

If you bias to 100% of static dissipation on your output tubes, you'd better be running Class A, and limiting signal so neither tube ever turns off. It is a matter of headroom, but not the way most people think of it.

Imagine a tube amp running Class B. That is, one tube does all the positive swings, the other all the negative swings. Ignore that this sound bad for the moment. At idle, the tubes dissipate *zero*. They're both off; they both support full B+ with no current flow, so voltage times current is zero too. When you apply a signal, they start conducting alternately. Now the power dissipation in the tube rises, as there is a time every cycle where the tube is letting current pass and has some voltage across it. In Class B, the idle dissipation is (in a perfect world, which this is not) zero, and it rises with increasing signal to some point. Beyond that, it backs off a little, so class B amps have a peak in output device dissipation below full power.

This sounds too ugly to do, so the standard approach is to turn both tubes on a little. There is an overlap around zero signal where both polarity tubes conduct. This is class AB. The power dissipation at idle is now not zero, but some amount caused by B+ times the idle currents. The idle power is what it is, and in general, total dissipation goes up as signal increases, then backs off some as you near full power. This is the combination of static idle power plus the power generated by signal in class B.

As you keep increasing idle current (and power dissipation), the amount of power wasted in the output tubes from idle increases, and the amount of increase over the idle power decreases as there is less and less Class B swing where one tube is off while the other is on. They spend more percentage of the time with both of them on. When the idle gets up so both tubes run all the time, there is no Class B increase any more; you're running Class A all the time and wasting a pretty constant power no matter how big or small the signal is.

So running a tube at 60-80% of its possible dissipation limit means that you're reserving the additional 20-40% of dissipation for signals above the idle power. It's what lets you have a bigger output power than Class A. It's also what lets you increase the B+ some to get that bigger power without overheating the tubes.

[Steve Conner]

Some thoughts on crossover distortion...

A transistor output stage in a hi-fi amp is very linear, because the transistors run as emitter followers, which implies lots of local feedback. Well, not in RG's favourite Thomas Vox amps, but that is beside the point for now.

So, if we look at the transfer functions of the top and bottom halves, they are almost perfect straight lines, with just a little wrinkle near zero. With a small amount of added bias, they fit together almost perfectly. Using a distortion analyser, you can see a minimum when the fit is right. Once set up like this, the output stage has only about 0.1% distortion even before global feedback is applied.

In a classic tube amp, the entire load line of the power tube is one big "little wrinkle". In a push-pull pair they don't fit together anything like as well as transistors do, and there is no distortion minimum. I was once playing with some Russian Tung-Sol 6550s and you could clearly see this on a scope, the bias affected the shape of the output waveform, not just a little glitch near 0.

It just works out that the more bias you throw at it, the less distortion, or the less of the non-euphonic kind at any rate. All serious tube hi-fi amps were Class-A and had some sort of local feedback in the output stage too.

Also as pointed out, when you overdrive a classic output stage, it charges up its coupling caps with grid current and temporarily biases itself colder. That's why as MWJB says, the crossover distortion will always appear with enough overdrive. Unless you've done the Paul Ruby mod!

[Jazz P Bass]

There are no PNP tubes, so the signal fed into the power section has to be inverted to be fed into each of the complementary pair. I know you know this, please bare with me for the sake of this argument. So actually, the two tubes are seeing a signal going up nearer the positive grid region, which is more linear than that terrible region near negative cutoff - down there it's teeth grinding region.

When it's the negative going signal, one tube will cut off and the other will see a positive going signal in direction of that linear region. When it's the positive semi-cycle the other tube cuts off and one tube sees a positive going signal. So both see a positive going signal, there is no upper and lower sine wave, don't know if that's exactly what you were saying, but I wanted to point that out. So for a large signal swing, each tube in the pair is seeing the swing going in direction of that positive region.

If you bias both tubes hot, both of them will work with the signal swing pushing the grid closer to cathode potential, which is the sweet sounding region as far as commonly agreed.

Of course guitar signals are not sinusoidal, they're terribly assymetrical so one tube will always work harder depending on the timbre.... I don't know if I'm being terribly redundant and obvious, I know you know most of this.

Who Boy!
Thankyou Jmaf for straightening me out.
I do believe I was thinking class A.
"There is not a PNP tube"

[jmaf]

Who Boy!
Thankyou Jmaf for straightening me out.
I do believe I was thinking class A.
"There is not a PNP tube"

I think it's clear with what I said that I wasn't correcting you at all. Won't bother you again. Take it easy.

[Jazz P Bass]

Bother away, please.

[Chuck H]

This is yet another great discussion of the too often gray area of "bias". But, just for the OP's sake, I think some more layman descriptions could help. So (at the risk of a poor analogy needing correction) here goes...

Bias is the power that flows through the tube when it's not amplifying anything. (notice I didn't say "when it's not doing anything") The more power used to idle the tubes, the less power is available for amplification. I hope this makes clear that "biasing to 70%" does't mean you get 70% of the tubes power and "biasing to 100%" doesn't mean that you get 100% of the tubes power. This is why biasing to a higher % of the tubes power actually makes less power available for amplification.

The input signal level will affect how much power the tube will TRY to make. If there is more input signal level than the available power can amplify the tube will clip. This is why tubes that are biased colder (like class B as described) can take a larger input signal level and amplify it. Because the tube spends half it's time making almost no power it can spend the other half of it's time making maximum power. Unfortunately, as also noted, class B doesn't sound good for guitar amps.

I get the impression that your after maximum power. If so then what you would do is bias the tubes as cold as possible without creating "crossover distortion". Crossover distortion is a generally bad sounding artifact that happens when the push of one tube doesn't meet evenly with the pull of the other tube (as in push/pull output) and it sounds like a can of bees. A cold bias can result in this type of distortion. So if you bias to a lower % of the tubes power and hear a mosquito buzzing around in your tone you'll know you need to bias to a higher % of the tubes power. Once you reach the point where the power tubes will clip, and the crossover distortion level is tolerable, that will be the most power your amp can make as limited by other things that become too technical for the moment. That said...

I get the idea that what you want is maximum power.?. So bias as described above or tell your tech to do it, and...

Get some effecient speakers. Speaker effeciency is a great way to bump your rigs volume (which of course sounds like more power, and actually is) If the speakers you have now are rated at, say, 98dB efficiency, and you get some new speakers that are rated at 101dB efficiency, the volume increase would be the same as DOUBLING your amps wattage. Good, efficient speakers are the best way I know of to get the most volume from your gear. You can go through tube types and ratings (GrooveTubes rates tubes for relative clean-ness) and all manor of biasing and never get the volume increase that you can get from more efficent speakers.

OK... Who's going to correct my plain speak explaination. All critique accepted, BUT, you must do it in similar plain speak that wouldn't confuse a novice

[jmaf]

Bother away, please.

Just gimme a few hours and I'll be back with my usual tragically pedantic self

Honest - didn't mean to correct you at all. I'm not RG or Steve Conner,(added Chuck H ) basically I was thinking out loud and realize I overtyped - but only after I did. My bad.

[leadfootdriver]

How much does the Output Transformer have to do with power output? Does the amp put out less power if the OT has a high primary impedance compared to one with a lower PI?

[JoeM]

Here's a chart from a 6L6 data sheet:

[diagrammatiks]

everything?

it's the other half of the load line.

b+ on one end. voltage dropped across the ot load on the other.