Have you tested that circuit with a square wave generator to look at "ringing" or other artifacts? Three transformers is a lot of iron... UTC transformers are good, but if the circuit will pass clean square wave over the audio spectrum, the integrity of the preamp to transients should be good. The sine wave tests say something, but you need to go a little further. As an experiment, take the input transformer and feed -10 dbm of square wave in, terminate the secondary with a 100 k resistor, put a scope across the output, and look at the wave form. If you have a dual trace scope and can look at the input and output simultaneously, what you may see may surprise you. You may observe improvement by changing the 100k resistor to a lower or higher value. "Tuning" out any "ring" with a low value capacitor may also help. Remember, you have Miller effect at the input, especially a input stage with gain. Miller effectively multiplies the effective capacitance at the grid by the tube section gain, so be careful. Keep the leads short and stray capacitance at a minimum...