Tom, I'll describe the clamp circuit and perhaps you can simulate it and evaluate it. It's along the lines of many other transient absorption circuits I've designed. What I don't know is to what voltage the tube cathode jumps to so I am assuming 1 kV for now. My model also does not include the IC-746 output. That is modeled for now as 50 ohms to GND. The model does include the AL-572 input circuit. A single half sine is produced when arced into a 50 ohm load (IC-746). With the 1 dB pad the AL-572 input circuit is underdamped and it rings for one cycle.
AMP INPUT----1 dB PAD----CLAMP----ICOM
The 1 dB pad it a T with 470 ohm to GND, 4.7 ohm series, 470 ohm to GND. More attenuation would be great for absorbing the energy but I don't think more than 1 dB or 2 dB can be spared from a 100 watt radio and still drive the AL-572 to rated output power.
The clamp consists of two steering diodes to 1 uF of ceramic capacitors maintained at +120 and -120 volts. The steering diode for each rail is made up of six paralleled strings of three series connected 1N4148-type diodes. These diodes survive 10 amp pulses (a few us) in this type of service. Between each 1 uF cap and its 120 volt battery is a 47k ohm resistor. The negative clamp does not appear to be needed.
That's a good clamp.
I have a patented clamp in a receiving antenna system that uses a single reference voltage. It is a hard clamp with peak voltage set by a single zener. The zener can be fed from a 1-5 mA source, just enough to overcome leakage.
It's really odd the amp won't pop 1N4148's inside the amp but the Pro can't handle the transient.
I'm experimenting with a different approach. I have little gas clamp protection devices rated at 150-175 volts peak. I'm running tests on them now. I'm doing one at the filaments, and across the input jack. It's tough to measure the pulse level because it is so narrow and random.