Gain of output valves

[martin manning]

Here is what I have been doing...

Absolute Lpri doesn't matter for these ratios.
As said, leakage inductance doesn't scale with primary inductance.
So the coupling factor can't be constant.

Indeed running the transformer model shown above in isolation with the inductances increased doesn't make any difference.
With the coupling factor already at 0.9995, where can it go?

Please show your "scaled" OT model. There must be an error.

Thank you professor, you are correct ;^) Secondary inductance was scaled by 10 instead of 100. But does this look reasonable? Output power dropped ~5% and THD is 500%. You can see it's clipping a bit, but...

[martin manning]

Input signal reduced a bit.

[Helmholtz]

The large "whale shaped" loop makes no sense.
Increased primary and secondary L should make the loop slimmer rather than wider.
Only explanation could be a very large leakage L.

[martin manning]

The large "whale shaped" loop makes no sense.
Increased primary and secondary L should make the loop slimmer rather than wider.
Only explanation could be a very large leakage L.

This is getting tedious to run. It takes several minutes for each case, then the plot scales and plot reference notes have to be updated by hand, and a separate plot has to be generated to integrate load voltage vs. time to get output power.

Here is a new set of plots, all at the same input signal level:

1) Base, OT modeled as shown above
2) Inductances increased 10x (narrower loop)
3) Inductances increased 100x ("whale" loop)

Last one:
4) Inductances at base level, coupling factor reduced from 0.9995 to 0.98 (wider loop, decreasing it further, to 0.95, makes a "whale" loop)

It is possible to model variable inductances if the right data were available. If leakage inductance is constant, the coupling factor can be set to unity, with a separate uncoupled inductor added to represent leakage. That is another common non-ideal transformer model topology. If leakage inductance varies in some known way, modeling that would also be possible. I have to believe all of this has been done before, perhaps you have done it yourself? At least this exercise shows the effects of increasing inductance and increasing leakage by reducing coupling factor in this model.

Note the THD numbers in previous posts are garbage. The Fourier analysis directive has to be told explicitly what the input frequency is, and it did not get updated when I switched from 400 Hz to 1 kHz. That is now automated.

[Helmholtz]

3) Inductances increased 100x ("whale" loop)

If that includes the leakage inductance (which would make no sense), the wide load loop is no surprise.
Generally a wide loop indicates significant load reactance.
(A similar but weaker effect occurs with a speaker load due to speaker inductance of 0.5mH to 1mH being stepped up by the OT.)

The coupling factor k is defined as k = sqrt( 1 - Lleak/Lpri).
Using the measured values of the Fender OT (Lpri = 172H, Lleak = 7mH) this gives k = 0.99998 at full power and 50Hz.

Obviously the Fender OT is different from the Hammond OT: Higher inductance values, higher Raa (2.2k) and larger winding DCR.
I remember having measured an original '72 Fender Dual Showman OT and the numbers were close to this '65 Twin RI OT.

[martin manning]

OK, some progress. Setting K = 1 (zero leakage) for the 100x Lpri case (Lpri a-a is 250 H) collapses the loop.
Adding the 1.31m Lleak from the Hammond data sheet to the primary (half on each leg) opens it just slightly.

[Ten Over]

I’m running the amp into a 4 ohm load.

Did you ever say what kind of 4 ohm load?

[Mark]

I’m running the amp into a 4 ohm load.

Did you ever say what kind of 4 ohm load?

A resistive load using wire wound resistors.