I removed the front end Nuvistors in a SK2-V to create a K2-W upgrade with similar performance to the K2-W but twice the output voltage swing at +/-100V. This is what the original K2-X and K2-XA were designed to do. After removing the Nuvistors, I have not change the compensation in the remaining stages. The existing compensation is probably excessive without the added gain and phase shift of the Nuvistors.
I am calling this experimental version SK2-X. Bob Pease says that the "X " was the convention of eXperimental, at GAP/R. Just like the K2-X was an experimental version of the K2-W.
This modification can be done to the SK2-V in a temporary fashion. I used an adapter socket at the 12AX7 as shown in the pictures below. I jumpered several wires straight through, and brought out the 12AX7 grid inputs to the grid socket connections at the Nuvistor (6CW4) sockets. I tied an extra 1Meg resistor from the 12AX7 cathodes to the Nuvistor socket cathodes to bring the 12AX7 bias current to the same current as the removed Nuvistors. You will notice in the photos that I replace to defective ceramic capacitors in this module with two new mica caps.
I tweaked up the positive feedback around the 12AU7 with an added 22Meg resistor between the cathodes. The open loop gain for +/-100V is 33k and for a smaller swing of +/-50V it is 66k.
You can find my original measurements of the SK2-V here.
This schematic is also available in SwitcherCAD format as SK2-X
My first try at tweaking the positive feedback was with a 32Meg resistor, which was insufficient.
Using a 22Meg resistor was optimum to set the amount of positive feedback.
These same curves shown in the time domain. Note that the looping and stepping at the triangle peaks is due to pole-zero behavior in the SK2-X response.
No need to make any internal modifications to the SK2-V. All the modifications are on the Vector socket adapter under the 12AX7. You can also see the 22MEG trim resistor on the socket extender of the 12AU7.