Biasing by Percentage of Maximum Plate Dissipation

[martin manning]

A one-page document capturing methods and measurements required to bias a tube power stage.

[pdf64]

Nice work Martin, are you thinking of uploading it to a linkable webpage?

Regarding the OT shunt method, I think that describing it as 'potentially inaccurate' would be a useful addition to its downsides. That would be justified in the scenario that the current meter's resistance is significant in relation to the OT half primary winding resistance. eg my Fluke 75 300mA sockets have a 6 ohm resistance, 79ii 40mA sockets a 10.7 ohm resistance, 189 400mA sockets 2ohm resistance.
Hence it might easily be the case that the parallel path provided by the OT half primary (ie parallel to the current meter) carries 5 perhaps even 10% of the anode current, and the meter reading would hence be somewhat under the actual anode current.

Regarding the example dissipation limit of 30W for 6L6GC, did you consider converting it to its design centre equivalent (eg about 26W), being that the EL34, EL84 and 6V6GT limits are under the design centre system?

Regarding the example case of a 5881, I can't recall a 25W limit being published for this under any rating system (though I guess the design maximum limit would be about that).

[Phil_S]

Martin,
Notes on the shunt method are refreshing, and more importantly, succinct and very much to the point. Many years ago, I decided that I was not adequately trained or confident enough to be using the shunt method -- frankly the downside aspect scares me. I reasoned it wasn't necessary because Ohm's law says you can measure the voltage and use the R value of the transformer. I never confirmed this was appropriate and correct. I was very pleased to see what you said about this.
Phil

[martin manning]

Thanks for looking! I've revised the 5881 Pa Max to 23W, and added the caveat "inaccurate" to the shunt method for Ia.

On the Max levels for Ia, I'm not sure we know enough to convert between systems. There is an Ei data sheet that shows 6L6GC Design Center at 19W: https://frank.pocnet.net/sheets/084/6/6L6GC.pdf

[Helmholtz]

A general remark:
Just focussing on the plate dissipation (PD) percentage may be misleading.
Tube performance directly varies with idle current but not really with the PD percentage.
E.g. if an amp was designed for 5881s, there's no technical reason to increase the idle current when using 6L6GCs, even if this means a significantly lower PD percentage.

[martin manning]

Seems like everyone does it that way, though. What technique would you recommend?

[GAStan]

I suggest putting this in the FAQ section.

[Helmholtz]

Seems like everyone does it that way, though. What technique would you recommend?

Sorry, no simple rule - just a general consideration.
What I mean is that a 6L6GC doesn't need a higher idle current than a 5881 just because it has a higher PD limit.
5881 and 6L6GC share the same performance data, so when running at the same idle current the performance should essentially be the same. Limiting values are safety data, not performance data.
So when the original 5881s were biased at e.g. 16W plate dissipation, also 6L6GCs can be biased to 16W in the same amp.
Probably not detrimental to bias the 6L6GCs higher than the 5881s, but there's no need regarding performance..

BTW, I like your pdf document.

[martin manning]

Thanks! IMO the main thing is to avoid idle bias that is too hot, and %Pa Max is the best way to determine a nominal upper limit. It can be lower, but not so low that crossover distortion is produced. That can be evaluated by sound and or using a scope. The thing to avoid is trying to bias sonically or visually on a scope without checking the Pa, and of course the ultimate test is verifying that the plates do not overheat in use.

[pdf64]

The thing is that there's a perception that below maybe 60% in fixed bias AB, the valves are literally too cold to operate properly and will be liable to crossover distortion.
Whereas Fender note to idle 6L6GC at 13W eg in a Super Reverb RI https://el34world.com/charts/Schematics ... ematic.pdf
So the 60-70% target range may be needlessly narrow and potentially wasteful of valve life?

... On the Max levels for Ia, I'm not sure we know enough to convert between systems. There is an Ei data sheet that shows 6L6GC Design Center at 19W: https://frank.pocnet.net/sheets/084/6/6L6GC.pdf

I think that EI info can't be taken as representing 6L6GC, as apart from the title, the limiting values etc are identical to 1st generation 6L6/G/GA/GB.
Whereas the 5 ply anode used in the GC increases its dissipation by at least 25% http://n4trb.com/AmateurRadio/GE_HamNew ... No%201.pdf

[Lauri]

Class AB fixed bias amp should never be biased to 70% of maximum plate dissipation if control grid circuit maximum resistance is exceeded. It will pretty much guarantee a short life. There's a reason why in the old days fixed bias amps were usually biased to somewhere around 40% to 60% but these days most people working on tube amps have almost zero understanding of tube circuits.

[martin manning]

Maybe it would be better to suggest 50-60% plate dissipation for Class AB.

[nuke]

I was going to post that is entirely possible to design an amp that uses the output tubes conservatively for maximum life. For instance, Fender used 6L6GC pairs for amps that made 40, 50 and 75 watts. There's really no need to bias all of them to such a high idle dissipation.

The real dope is to employ the load lines to compute the regions of class B and class A operating lines, and derive the operating points.

It would be nice if the amp makers would leave a note in the schematics what they expect the idle current to be for intended operation. But, they didn't leave us many clues, did they?

The hotter you set it, the more tube life and power you waste doing nothing particularly useful.

Also, I see nothing wrong with the using the ammeter shunt method. I don't agree that it is less accurate, because if you use the voltage drop across the primary coil, you need an accurate resistance measurement of the wire.

The ammeter shunt is a tiny fraction of the coil resistance, paralleling the ammeter across the transformer introduces the slight error of the two resistors in parallel, but it isn't all that much. Go calculate it, the error is really small.

You have to use reasonable care with either method. Measure current or voltage, you still have meter probes connected to dangerously high voltage. You also can't forget to move the test leads back to the voltage inputs on your meter... but be a grown up around high voltage already.

[Helmholtz]

Also, I see nothing wrong with the using the ammeter shunt method. I don't agree that it is less accurate, because if you use the voltage drop across the primary coil, you need an accurate resistance measurement of the wire.

The ammeter shunt is a tiny fraction of the coil resistance, paralleling the ammeter across the transformer introduces the slight error of the two resistors in parallel, but it isn't all that much. Go calculate it, the error is really small.

The OT primary of my 100W Marshall measures 15R/17R.
So using a mA meter having an input resistance of 6R will give an almost 30% low current reading.

[pdf64]

... I see nothing wrong with the using the ammeter shunt method. I don't agree that it is less accurate, because if you use the voltage drop across the primary coil, you need an accurate resistance measurement of the wire.

A decent meter can easily and accurately provide that data.

The ammeter shunt is a tiny fraction of the coil resistance ...

Please see the 2nd post in this thread, in which I have provided data demonstrating your assumption is incorrect.