mwhouston;
I have been looking at the design for this amp and I have a concern. I believe that the ATS25/807 may be at risk of thermal runaway using this design. I believe that the appropriate fix is the addition of a screen resistor of approximately 1.7kΩ in series with the transformer UL tap.
Please allow me to explain. The nature of UL operation is such that, at the zero signal condition, the screen voltage is slightly above the plate voltage. This happens because the bias current flowing through the output transformer primary resistance causes a voltage drop between the B+ tap and the plate tap. Because the screen tap is typically 40% of the way "down" the potential well from the B+ supply, the resultant voltage at that tap is higher then the voltage at the plate. For some power tubes this is acceptable so long as the transformer resistance is relatively "small" (with the definition of "small" depending on the bias current, and tube construction). Tubes such as the KT88, 6L6, 6V6, EL34, EL84 etc. can deal with the few volts differential without undue heating.
However, the ATS25/807 is an RF transmitter tube. In this tube, the location of the screen is relatively close to the control grid. This is why in tetrode operation, the max plate rating is 600v and the max screen rating is only 300v. If the screen potential is too high relative to the cathode/plate E-field potential at it's location, the screen dissipation will go up, its ability to modulate the plate current flow will be lost, and the tube will go into thermal runaway. Therefore it is imperative that the screen potential be kept low enough relative to the cathode/plate E-field potential at it's location that the overall tube function is not impaired.
Now, the calculations for the schematic have a 97ma plate current and a 6.8mA screen current as calculated from the published conditions in the STC807 report. I have estimated the output transformer primary resistance at 200Ω (This is the actual resistance of the GXSE10-8-5K, but since I don't have the data for the CXSE25-8-2.5k, which should be lower, I specified what I had). This means that the plate-to-screen differential voltage is 0.6*200Ω*0.097A=11.6v. Assuming the 6.8mA screen current at zero signal, this implies that a resistance of 11.6v/0.0068A=1.7kΩ will drop the necessary voltage to bring the screen back down to the plate voltage level. I don't believe that this level of screen voltage suppression will have a perceptible effect on the UL operation of the tube.
I have attached the modified schematic below. Note that if the actual primary resistance of the transformer is different, the formulas above can be used to calculate the appropriate size of the screen grid resistor.
Attachment:
Modified Schematic.png
This whole concern may be the product of my overly developed sense of paranoia but I believe that a little caution up front can prevent much cursing later on. Please let me know when you've seen this post.