KT88 amp

[Helmholtz]

Fully agree, that's why I typically use it between double quotes.

Now we're in need of a better term. Any suggestion?

[pdf64]

Yes, that's what I called "saturation voltage" (had seen this English term used somewhere). …

Yes, I’ve read that terminology somewhere, but wasn’t keen on it, as it might imply some relationship to operation in saturation mode, which is something else again.

[Helmholtz]

A few more watts will be hardly noticeable with respect to SPL.

See post #56.

[The Dude]

I get that this might be an exercise in theory and a quest for knowledge, but I'd just like to point out that "squeezing out a few more watts" isn't something to be concerned about. A few more watts will be hardly noticeable with respect to SPL. You'd need to double the output power to get even a 3dB increase.

[Helmholtz]

The only apparent explanation seems to be that the anode voltage on their Vg1 = 0 curve is a bit higher, thereby reducing the max p-p voltage swing.

Yes, that's what I called "saturation voltage" (had seen this English term used somewhere). Zollner calls it "residual plate voltage" (my translation).
It is the voltage drop across the tube at max. signal current.

A tube with an increased saturation voltage has a more rounded plate characteristic and increased plate dissipation at full power.

[pdf64]

…As mentioned earlier, you might have somewhat inefficient tubes.
E.g. Prof. Zollner found that Chinese KT66s can produce less output than original GEC types because of higher "saturation" voltage.
Don't know about KT88s, though.

Yes, modern power valves just don’t seem to produce as high an output as old vintage production ones that are still serviceable. Beam types, at least. The only apparent explanation seems to be that the anode voltage on their Vg1 = 0 curve is a bit higher, thereby reducing the max p-p voltage swing.

[Helmholtz]

Maybe I'm missing the context, but I don't understand this statement. Maximum supply current is determined by the PT, is it not?

A PT has a thermal current limit. Doesn't mean it can't supply more current, but will overheat.

For a given B+ the current is solely determined by the load, i.e. the tubes and the PT has no influence. All the PT can do is sag the voltage.

[g1]

Supply current is determined by the tubes, not the PT.

Maybe I'm missing the context, but I don't understand this statement. Maximum supply current is determined by the PT, is it not?

[hylaphone]

Use Nick's calculator to find out. I wouldn't go lower than 4k though to avoid plate overdissipation.

4.2K yields about 10W more

[Helmholtz]

So perhaps an OT with lower primary Z would yield more power?

Use Nick's calculator to find out. I wouldn't go lower than 4k though to avoid plate overdissipation.

[hylaphone]

As your B+ sags by only around 20V at full power, I don't see a supply current problem. Supply current is determined by the tubes, not the PT.

Another question (sorry to ask): Are you sure your screen resistors are actually 47R and not 47k?

Yes they are 47 ohms.
So perhaps an OT with lower primary Z would yield more power?

[Helmholtz]

So this suggests the bottleneck is either OT losses, or plate supply current?

As your B+ sags by only around 25V at full power, I don't see a supply current problem. Supply current is determined by the tubes, not the PT.

Another question (sorry to ask): Are you sure your screen resistors are actually 47R and not 47k?

[hylaphone]

Calculating to 55W.
Using your RMS meter probably is more accurate.

As mentioned earlier, you might have somewhat inefficient tubes.
E.g. Prof. Zollner found that Chinese KT66s can produce less output than original GEC types because of higher "saturation" voltage.
Don't know about KT88s, though.

Today I received a pair of KT90, which are rated for 50W. Tried a few different bias points and still can't get beyond that 22VRMS output.
So this suggests the bottleneck is either OT losses, or plate supply current? I'm just trying to figure out what options are left to reach target power, or even to just squeeze out a few more watts.

[Helmholtz]

You should be able to make up a few watts by increasing the bias voltage (making it more negative).
That allows for more grid swing before grid current clamps the signal.

I don't see how that could increase max. plate current swing before clipping.
Nick's calculator doesn't show a difference.

[uneumann]

Load resistor is 8.2 Ohm.
Measured right at the speaker jack, same 60W before clipping.

So 10% loss of the 70W target is 63W, more or less where I'm at. Does that jive with your estimates?

Do I have any options left to gain 10 clean watts?

You should be able to make up a few watts by increasing the bias voltage (making it more negative).
That allows for more grid swing before grid current clamps the signal.

Edit: never mind - see below - greater grid swing does not lead to greater output power.