Just curious if anyone has built/analyzed anything like the original push-pull series feedback IT circuit.
Doing a bunch of spice modelling with some different pentodes, the surface level analysis looks good, but two troublesome issues stand out:
1) The 0.1uF capacitors in the feedback path cause a huge increase in gain at low frequency as the feedback trails off. If the interstage can pass frequencies below 10 hz, feels like things could be problematic. I had to go up as high as 3.3uF in some cases to kill the hump down in the 3-5 Hz range. Certainly you can get caps that size but it feels a little ugly to have both an interstage and also caps large enough that they probably have to be metallized.
2) I have some bifilar interstages I was thinking about using if I did try to build it, but it looks like this circuit is not going to be ideal for inter-winding capacitance no matter what, as the lowest AC voltage you could put on the wires that are physically at the center of the winding would be the voltage grid-to-grid for the drivers. The voltage swing on the feedback dividers is very large so you'd essentially setup the secondary to 'drive' the feedback path while the grids are the ground side of the windings.
Probably the answer is 'Don't do it, this is not worth it.' Since you have really high drive voltage through the IT, you'd basically have to build a driver setup powerful enough to drive an IT at low distortion and enough voltage for big power triodes anyway. You'll have basically built a monster PP driver that could just drive PP 300B's anyway. In theory the series connection doesn't present much load to the IT, but Rp of the driver is still going to dominate distortion through the IT, plus whatever the parasitics I mentioned above are doing.
Doing a bunch of spice modelling with some different pentodes, the surface level analysis looks good, but two troublesome issues stand out:
1) The 0.1uF capacitors in the feedback path cause a huge increase in gain at low frequency as the feedback trails off. If the interstage can pass frequencies below 10 hz, feels like things could be problematic. I had to go up as high as 3.3uF in some cases to kill the hump down in the 3-5 Hz range. Certainly you can get caps that size but it feels a little ugly to have both an interstage and also caps large enough that they probably have to be metallized.
2) I have some bifilar interstages I was thinking about using if I did try to build it, but it looks like this circuit is not going to be ideal for inter-winding capacitance no matter what, as the lowest AC voltage you could put on the wires that are physically at the center of the winding would be the voltage grid-to-grid for the drivers. The voltage swing on the feedback dividers is very large so you'd essentially setup the secondary to 'drive' the feedback path while the grids are the ground side of the windings.
Probably the answer is 'Don't do it, this is not worth it.' Since you have really high drive voltage through the IT, you'd basically have to build a driver setup powerful enough to drive an IT at low distortion and enough voltage for big power triodes anyway. You'll have basically built a monster PP driver that could just drive PP 300B's anyway. In theory the series connection doesn't present much load to the IT, but Rp of the driver is still going to dominate distortion through the IT, plus whatever the parasitics I mentioned above are doing.
Since C is part of the NFB loop, there will be peaking. Instead of increasing the value of C,
the values of R1 and R2 could be increased by the same factor, keeping their ratio constant.
the values of R1 and R2 could be increased by the same factor, keeping their ratio constant.
Fair enough on the capacitor and resistor divider values. I'm curious about how that will interact with the various parasitics. I need a better model of a PP interstage transformer.
I've had astounding success by doing series feedback to the previous stage cathode. It's so simple and just so good. You can do much more feedback if you want to.
Schade's circuit was interesting for showing some curves in a white paper, but it's just pretty complex for what it does. But it could be fun to build too.
Schade's circuit was interesting for showing some curves in a white paper, but it's just pretty complex for what it does. But it could be fun to build too.
