OT Protection at Output Question:

[martin manning]

With a safety resistor that is tens of times larger than the nominal speaker load the output is greatly reduced. You don't need wattage anywhere near the amp's rated output power.

[SamRE]

With a safety resistor that is tens of times larger than the nominal speaker load the output is greatly reduced. You don't need wattage anywhere near the amp's rated output power.

So purely for a safety resistor, I could get away w/ a pair of 960Ohm 25W Wirewounds in parallel? 30 * 16Ohm = 480 at W_Rated > 1/3 of peak dissipation (75-80W). Shouldn't overdamp the speaker response, nor should it be completely invisible to the primary right?

[martin manning]

50W would be more than enough. A 25W 470R would do it.

[Helmholtz]

470R is roughly 30 times the nominal 16 Ohm load.
So the reflected OT primary impedance will also be increased by that factor, resulting in an Raa of around 100k.
The huge plate load impedance significantly reduces plate current swing and thus available power.
A quick and dirty sim gave me a 20 times lower power compared to nominal load.
I think you should be fine with a 470R/5W safety resistor.

[martin manning]

In a not so quick and dirty simulation (full 100W Marshall power amp), with signal level that produces ~70W into a nominal 16R load, a 470R load across the output dissipates 6.6W, so I'm seeing less than 10%.

[Helmholtz]

In a not so quick and dirty simulation (a full 100W Marshall power amp), with signal level that produces ~70W into a nominal 16R load, a 470R load across the output dissipates 6.6W, so I'm seeing less than 10%.

Did you consider that at such high plate load the tubes run in full class A, so effective plate load doubles compared to class (A)B?
This effect lowers available output by 2.

[martin manning]

I just replaced the speaker load with a 470R and let it run. The 470R resistor dissipation came from integrating power over the transient. When I plot plate current vs. plate voltage for an output tube I can see that it does not go into cut-off.

[Helmholtz]

So it depends on your OT model.
Can you measure effective plate load per tube with the 470R?

[martin manning]

Nominaly 38.2k.

[Helmholtz]

Assuming the typical nominal 1.7k Raa of a 100W Marshall OT, I'd expect an effective class A plate load of 49.94k per tube with the 470R connected to the 16 Ohm output.
I actually measured Raa = 1.75k with my '69 Super Lead OT, so that would increase the plate load a bit more.

[martin manning]

The effective load varies considerably around the loop, so the concept of a constant impedance lode line, even with a purely resistive load, is a simplification.

[Helmholtz]

Yes, the loop is caused by the primary inductance.
The lower the inductance and the lower the frequency, the wider the loop.
(An ideal transormer would have infinite inductance.)
This means that the load is both resistive and reactive.
The reactive part of the load doesn't cause real power and so doesn't transfer power to the output.

[pdf64]

In a not so quick and dirty simulation (full 100W Marshall power amp), with signal level that produces ~70W into a nominal 16R load, a 470R load across the output dissipates 6.6W, so I'm seeing less than 10%.

Does that take into account the much lower degree of HT sag at OT CT and screen grid supply nodes that may result from the 470R load?
Is it at the onset of clipping, or fully overdriven?
Bear in mind my previous scenario (based on experience!) "Many guitarists just keep turning it up when they can't hear anything.
I suppose worst case scenario might be 5 minutes cranked until they finally admit defeat "

[martin manning]

In a not so quick and dirty simulation (full 100W Marshall power amp), with signal level that produces ~70W into a nominal 16R load, a 470R load across the output dissipates 6.6W, so I'm seeing less than 10%.

Does that take into account the much lower degree of HT sag at OT CT and screen grid supply nodes that may result from the 470R load?

Yes. The simulation has the power supply modeled, including sag due to the HT source impedance.

Is it at the onset of clipping, or fully overdriven?

Onset of clipping, signal level would produce 70W at the speaker if it were connected.

Bear in mind my previous scenario (based on experience!) "Many guitarists just keep turning it up when they can't hear anything.
I suppose worst case scenario might be 5 minutes cranked until they finally admit defeat :lol:"

This would be turned up pretty high. With the safety resistor dissipating under 7W, a 10W safety resistor should be sufficient.

[Helmholtz]

Onset of clipping, signal level would produce 70W at the speaker if it were connected.

Didn't you simulate a 100W (4xEL34) amp?
(Also wondering because the OPs Marshall is a 50W 1987 ( 2xEL34) amp.)