Fixed bias, and power output.

[Steve Conner]

RG, all, thanks for the detailed analysis of the points I carefully avoided!

The way I see it is: If you take the same (PP, fixed bias) amp, and vary the bias up and down, but hold the plate and screen voltages constant, the output power at the onset of clipping won't change. The reason being, at clipping one tube is always full on and the other is off, and varying the bias doesn't change that.

Except as someone pointed out above, there's a second order effect if the plate and screen voltages sag with changes in bias. And as someone else said, if you redesign the amp every time you change the bias, that makes a big difference!

What fiddling with the bias does is mainly affect the gain of the output stage for small signals. Most tubes have gain that increases with plate current, so they have low gain for small signals.

Having two tubes conducting at once doubles the gain and fills in this flat spot in the transfer function. Solid-state amp designers talk of "gm doubling" when the output stage is biased too hot, it ends up with a lump of excessive gain around zero, but I don't think tubes suffer from that.

[diagrammatiks]

was in a face palming arguement with someone over at tgp who claims to run a amp repair shop.

he was convinced that by dropping a pair of 6550s in a el34 amp rated at 50 watts you had to bias the tubes extra cold or they would instantly produce 75 watts of output and kill the ot.

if there wasn't an amp in front of me i would have just pounded my entire face into my bench.

[Jazz P Bass]

So according to R.G's well versed argument, by biasing the amp colder you would have "more" available power at the speaker.
Please don't pound your face, pound the TGP posters.

[jmaf]

was in a face palming arguement with someone over at tgp who claims to run a amp repair shop.

he was convinced that by dropping a pair of 6550s in a el34 amp rated at 50 watts you had to bias the tubes extra cold or they would instantly produce 75 watts of output and kill the ot.

if there wasn't an amp in front of me i would have just pounded my entire face into my bench.

In a Traynor amp for EL34 with -bias applied to the supressor, bias would be shorted to the chassis via the 6550's metal base and then it'd yield a kilowatt and would fry the OT among other things. Have them fellas over at TGP hoodwinked us all ?

[diagrammatiks]

is that a problem with new 6550s?

that traynor guy and his crazy ideas.

i've never used a 6550 though. always seemed like the common man's kt88 to me.

[jmaf]

is that a problem with new 6550s?

that traynor guy and his crazy ideas.

i've never used a 6550 though. always seemed like the common man's kt88 to me.

Well apparently Traynor found that by applying a negative bias to his suppressors his EL34's would last longer and his failure rates were lowered.

It's not a problem, really, it's a feature of some 6550's.

Also some KT66's(Tung-Sol for instance) have pin 1 attached to the metal base. It's specially tricky on 6L6 amps, because if you decide to pop in a pair of Tung-Sol KT66 and you happen to have used pin 1 for a terminal to pin 4(common with Fender) then you'd have the metal base at several hundred volts shorted to the Fender retainers.

[Rhodesplyr]

was in a face palming arguement with someone over at tgp who claims to run a amp repair shop.

he was convinced that by dropping a pair of 6550s in a el34 amp rated at 50 watts you had to bias the tubes extra cold or they would instantly produce 75 watts of output and kill the ot.

if there wasn't an amp in front of me i would have just pounded my entire face into my bench.

It's scary how much some amp repair people don't know...

In practical terms, the most common stumbling block in trying to drop 6550s into an amp designed around EL34s is that EL34s will tolerate much higher grid-leak resistor values than 6550s, values that can cause 6550s not to hold a stable bias. For example, this is why you can't plug 6550s into a Dynaco ST-70 power amp (270k grid-leak resistors) without modifications.

[Steve Conner]

Well, there is a grain of truth in what the guy said. If you have an amp with cheap transformers that already run hot at idle, changing to 6550s will make it run hotter still, unless you bias them at the same current as the previous tubes were running. And that will be "cold" for a 6550, since you all like to specify bias as a percentage of rated plate dissipation, and the 6550 has more of that than the EL34 or whatever.

Also the tube swap will get you maybe 70W from a 50W amp if you're lucky.

Only a grain mind you.

[Gtr_tech]

Well apparently Traynor found that by applying a negative bias to his suppressors his EL34's would last longer and his failure rates were lowered.

It's not a problem, really, it's a feature of some 6550's.

Also some KT66's(Tung-Sol for instance) have pin 1 attached to the metal base. It's specially tricky on 6L6 amps, because if you decide to pop in a pair of Tung-Sol KT66 and you happen to have used pin 1 for a terminal to pin 4(common with Fender) then you'd have the metal base at several hundred volts shorted to the Fender retainers.

But....the tube retainers won't fit around the larger 6550 base, so the shorting is a non issue. You have to bend the "tube traps" down flat in order to get the tube into the socket.

[jmaf]

But....the tube retainers won't fit around the larger 6550 base, so the shorting is a non issue. You have to bend the "tube traps" down flat in order to get the tube into the socket.

But then the tube traps can spring back up shooting a hot 6550 out of the amp, possibly hurting somebody.

[Gtr_tech]

But then the tube traps can spring back up shooting a hot 6550 out of the amp, possibly hurting somebody.

Very unlikely (if the retainers are properly "smashed down"), but the rivets can always be drilled out and the retainers removed if its a concern.

[R.G.]

So according to R.G's well versed argument, by biasing the amp colder you would have "more" available power at the speaker.

People thinking about power amps tend to forget that power amplifiers are really power supplies with some amount of circuits added to let the power out to speakers in a controlled manner. There's a huge amount of elaboration in the amplifier circuits to control the power sent to the speaker in a hugely refined, just-so manner. But if the DC power is not available from the PT/rectifiers/filters, it's not going to go to the speakers.

Mother Nature sets the Rules: (1) You can't win (i.e. energy is not going to be created or destroyed in your power amp); and (2) You can't break even (i.e. you're always going to have some losses and waste of the power you have). Class D amps do the best job of converting the power supply into audio at the speaker, getting the majority of it there. Analog amps are much worse, involving lots of DC wasted as heat. What you waste as heat in the amplifier can't go to the speakers.

Class A is very wasteful. Most of the power supply output is wasted as heat in the output devices. This is so bad that getting rid of waste heat without melting down the output devices is the central issue in Class A amplifier design. Class A is so wasteful primarily because both output devices (for push-pull) are on all the time, so they're both making waste heat.

Class AB is better in terms of waste heat because one polarity of output devices is off for part of the time. There is a continuous range of how much better it gets as you bias from conducts-all-the-time (i.e. Class A) to almost-never-overlaps (i.e. limiting class AB, or true Class B).

If you have a fixed power supply B+ voltage, then SE Class A will produce the least audio power from that voltage; push-pull Class A will produce more audio and less waste heat; Class AB will produce even more, and Class B will produce more. More importantly, as you go from A to AB to B, less of the available power is wasted as heat in the output devices, so they run cooler.

If you're stuck with output devices that start to fail when they dissipate more than for instance 30W each, then picking Class A means you get the least audio out. This is because you would adjust the power supply DESIGN to give you the DC power to just barely not overheat the output devices. For the same B+ level, going to Class AB produces the same voltage swing on the output devices, but less heat. That "less heat" lets you redesign the power supply to increase the power supply voltage and raise the power dissipation in the output devices back up to their maximums again. But because the power wasted in the outputs is a smaller fraction of the DC power, the AC power out for the same power tube heating is now much larger.

This process of biasing colder and raising the power supply voltage until the power tubes get just as hot is how you get more power out of a particular pair of output devices. The only limit on this is when the cold bias and crossover make it sound horrible. This gets really objectionable a bit before you get to real Class B. There is therefore a big advantage to a manufacturer to flirt with the edges of bad sound and overdissipation to be able to advertise X watts.

Once the power supply voltage is chosen, dinking with the bias produces only minor changes in power output. Given that the audible difference betwen 50W and 70W is almost unnoticeable because of the oddities of the human ear, messing with bias should be for tone (more accurately, freedom from crossover distortion) and reliability (that is, not overheating) reasons, not power output reasons.

[MWJB]

When using larger based octal tubes it makes more sense to use underslung, spring-clip retainers, rather than the bear trap type. Sovtek do make a 6550 that drops into a 6L6/EL34 socket (assuming the socket is wired for 6550 operation). EHKT90 is also a drop in replacement in a 6L6/EL34 socket. In both instances heater current capacity should be confirmed as adequate for running the bigger tubes.

[Steve Conner]

But then the tube traps can spring back up shooting a hot 6550 out of the amp, possibly hurting somebody.

How do you think my old "Toaster" amp got its name?

[tubeswell]

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?