@Tubelab_com : what is the suggested maximum Zout of the driver to keep a reasonable performance?
Not sure if you are referring to the source driving the UNSET board, or the driver tube feeding the output stage, but the response is similar. Each is looking into R102 or R112 (1 meg) in parallel with the Crss of the mosfet (pretty small). Unless the driving impedance (Zout) is quite large the loading effect of the next stage is minimal.
The input impedance of the entire amp is set by R101 which can be nearly any value. A small resistance, say 10K may be hard for a tube preamp to drive, but no issue for a phone or CD player. A large value, say 1 meg can make RF and noise pickup on the input wiring a problem. I have seen an amp oscillate if the input is left open with a large value R101 and messy wiring in my test amp. I set this value arbitrarily may years ago on the TSE at 120K simply because I had a box full of mil sped low noise 120K resistors. Pick whatever value works best for your application.
Great! Indeed I agree it works in substitution of half a 12ax7: I use it in the recovery stage. I’m out with friends, I will weite you something I noticed on biasing the FETSET when back home.
George, this week, while playing with different tubes in UNSET configuration, I was thinking about the impedance of the voltage divider from a to g1: I remember you suggested 300 kOhm as a total value, but how it sounds by keeping the same ratio and changing the impedance (so EG having more current flowing through it)?
The feedback resistors connect from the plate of the output tube to ground. The plate of the tube will swing from near zero to 2 X B+ in normal operation and much higher if the amp is driven to clipping into a speaker near resonance. The resistors must be large enough to handle the power AND be rated for this voltage to live without failure. Do not place them directly on the surface of the PC board. Space them 1/8 to 1/4 inch above the board to avoid heat build up and possible arc over. The 300 K number was again picked because I have a lot of 3 watt 150 K resistors rated for several hundred volts. I also have several other value 3 watt resistors. I did some testing with multiples of these resistors and 300K + 12K to 33K has seemed the best so far, but a serious controlled test has not yet occurred. If the resistor value gets too low the parts may be subject to heat related value change over time. Too high and grid current can upset the feedback. Many vacuum tube models do not correctly model grid current, and this can vary considerably from tube to tube. LT spice is not to be trusted in cases where grid current can play an issue in the results.
The first use of UNSET / CSD technology was in a push pull amp. That little amp is shown next to a Tubelab board for size comparison. The SPP makes 15 to 20 WPC. The push pull CED based amp makes 70 to 80 WPC!
The original breadboard for the CED is the green perf board seen to the right of the completed and operational push pull amp. It is being driven by a hacked up driver board that I never released. You can see the flying feedback resistors in the spring wire connectors around the perimeter of the perf board. The grid to ground resistors look like 100 ohm resistors in the picture, but they are really 20K. The orange bands are brownish due to some abuse, but theparts are still good.
I have not had the time to go back to this breadboard for more testing.
