JTM45 style amp with KT66's can't run 6L6's?

[martin manning]

When the load line crosses the Vg1=0 line far below the knee (as it does when the primary impedance is too high), it takes less Vg1 voltage swing to reach the point where the screen current shoots up, and it's much more likely that the screens will be over-stressed.

[stephenl]

When the load line crosses the Vg1=0 line far below the knee (as it does when the primary impedance is too high), it takes less Vg1 voltage swing to reach the point where the screen current shoots up, and it's much more likely that the screens will be over-stressed.

This would only be of concern when operating closer to max operating point, right?

The amp's that I've built with KT66's into 8K primary were designed for about only 25 watts output.

[martin manning]

This would only be of concern when operating closer to max operating point, right?
The amp's that I've built with KT66's into 8K primary were designed for about only 25 watts output.

Yes. With a high load impedance the power stage will clip before Vg1 reaches zero, and as you drive it harder, taking Vg1 closer to zero, screen current shoots up. If the preamp output signal isn't capable of driving the power tubes into clipping, then you are pretty safe. If it is, large screen resistors will limit screen current and lower the Vg1=0 knee dynamically, both of which will protect the screens from over dissipation.

[stephenl]

So on the standard JTM45 w/KT66 -> 8k Prim, would it be a good mod to remove the 470K screen grids and replace with 1K or 1.5K to better protect the screens? Might add a little squish to the PA?

[martin manning]

Yes, I would say so. 1k is a common choice.

[pdf64]

The classic JTM45 uses a 1k shared then 470ohm individual screen grid resistor network http://marstran.com/JTM45.gif

Yes the RI dispenses with the shared 1k but then its HT voltage is lower.

[martin manning]

The classic JTM45 uses a 1k shared then 470ohm individual screen grid resistor

Yes, and going from 470 to 1k individual or increasing the 1k shared to 1k5 will have pretty much the same effect in reducing the peak screen current. As noted above, this is an issue if the amp is played loud or into an attenuator.

Yes the RI dispenses with the shared 1k but then its HT voltage is lower.

It's more about primary impedance. If I recall early JTM45's had 8k and later ones used 6k6. IDK what the reissue has. Hammond's replacement OT splits the difference at 7k3.

[Roe]

The classic JTM45 uses a 1k shared then 470ohm individual screen grid resistor network http://marstran.com/JTM45.gif

Yes the RI dispenses with the shared 1k but then its HT voltage is lower.

yes but on the RI the screen voltage tends to be higher than the plate voltage even at idle

[martin manning]

I've put a simulation together to demonstrate the effect of high load impedance (load line far below the knee) and the value of the shared screen resistor on screen dissipation. It shows that the original 8k Zpri with 470R and shared 1k is risky if the amp is played loud (driving Vg1 to 0), and it will only get worse if the grid signal is increased further. That will cause the screen grid to spend even more time at high dissipation. Increasing the shared screen resistor to 1k5 doesn't seem to have much effect on the output waveform, so I wouldn't expect much loss of character for doing that.

[pdf64]

What HT voltage?

[martin manning]

About 430.