Crazy Fdbk is a lot like CFB added to a screen driven tube. Gonna need some excessive V drive.
For Crazy Fdbk, depends on the output tube and the gain setting of the N Fdbk network. You definitely want to use a tube with a low internal Mu factor for lowest screen V. TV Sweep tubes can get you from 3.0 to 4.0 internal Mu readily. Crazy drive typically is running with 0 to +150V grid 2 drive for these tubes. (driver may be supplying the negative half of that swing also)
Then the resistive divider from plate to grid2 (or possibly a grid2 Mosfet follower in there) and then to ground, determines how much of the plate V swing appears at the grid2. This R divider might be set for gain similar to the internal Mu for example. So one might have 1/3 of the plate V swing (so 1/3 of B+ to 50V say ) as negative feedback on grid2. So add that to the basic grid2 drive requirement and you may end up with like 150V plus 150V (and then possibly the negative half swing too) for drive signal.
12AX7, probably not gonna do it.
Gonna need something like a small TV Sweep tube for the driver stage. 6GF5, 6AV5, 6BQ6A, 6SN7, 6S4A, 38HE7, .....
Increasing the R divider ratio (so less N Fdbk) will reduce the drive V requirement of course. A lot of N Fdbk isn't really needed for linearity, since the operating principle is still Crazy drive (but from the cathode side here. ) But to get the output impedance down, one is going to want some N Fdbk functioning. The native Crazy drive output Z is similar to datasheet g2 drive curves, and the Crazy N Fdbk will lower that Rp by the amount of N Fdbk applied.
If one is going to all this trouble to get the output stage rock linear, one is going to want a high quality driver stage too.
I would say that, practically speaking, ordinary "local" N Fdbk(s) back to a driver stage ( so mucho gain is enclosed in the loop) would be a lot easier to do for similar results to Crazy Fdbk. But then a "wire with gain" does have a certain "cachet". (Mt Everest has been climbed here...)
And then if one is going for some kind of ultra low distortion amp (ppm with tubes maybe?), then a perfectly linear output stage does have some strong attractiveness. With driver stage "local" N Fdbk added, and global N Fdbk added to push those extra .0000...% on out there.
Last edited:
I would like to ask if this concept ( Positive Current Feedback simple Zen amp ) can be applied to tube amps too, by taking the positive current feedback from the 10 Ohm sensing resistor at the cathode of the power tubes, and sending it back to the cathode of the driving stage through a dedicated mosfet.
Thanks in advance.
Thanks in advance.