Though the JCA20H is _only_ 20watts, you have to take account of how loud 20 watts is, how you define 20 watts (some define it at 1% THD, some at 5% THD, it makes a big difference) and how efficient your speaker is. I decided as it's a simple mod to put in a triode switch which simply drops the output power to about 10 watts. However I found that the triode mod has develops a much smoother tone which I like. A fly in the ointment is that I also unbalanced the phase splitter at the same time, so two variables at the same time.
However looking at the spectrum for Pentode and Triode shows that the harmonic balance shifts and that the change in tone is not just perception and that switching a EL84 output stage between Pentode and Triode is not only an output power change. (unless the phase splitter modification has also contributed)
The initial test setup was to keep the input stage as clean as possible and then drive the output stage in pentode configuration to 5% THD into a 8ohm resistive load.
The normal Pentode spectrum (500Hz sine input) looked like this.
|Pentode 5%THD into 8 ohm.|
When switched to Triode mode the THD increased by 1.2% and the upper spectrum changed structure.
With the input gain set at maximum and clipping in the pre-amp you can see how the different operating modes modify the tone. In Pentode mode the dominant harmonic structure from the overdrive is clear.
In Triode mode again the upper harmonic structure produced from the preamp is modified.
In the final version I fitted R7 and R5 (470R) as 1k5, 5 watt. C-8 Helps keep the supply stiff, R-6 brings the phase splitter value from 82k to 50k which skews the output balance and allows the output stage to generate second harmonics (ongoing trial) C-6 reduces the power stage bandwidth to about 10kHz. The depth mod is from the web to boost low frequency.
|JCA20H HSG output stage.|
I did play with the idea of having SW1 as two separate switches so I could run one valve in pentode and one in triode in lieu of messing with the R46 value in the phase splitter but I didn't get around to it. You do loose some output power in changing the R46 value, but it is negligable.