(Without any guarantees from me) the radio is probably fully protected against high SWR. I've certainly abused my radio from time to time, changing bands and transmitting where there's no antenna, etc. It depends on your level of comfort. There's probably much less risk of damage at lower power, the worst case would be a sudden, infinite VSWR while at full power. You could reduce the risk by including a small value attenuator (even a length of cable) to start.
Calculate or measure the input VSWR with one or two capacitors open circuit, this is the most likely sudden failure mode that you can't predict. It's probably not as bad as infinite SWR.
Your VNA S21 measurement will give you some idea of the power that the diplexer will need dissipate, if you calibrate it carefully. Something like this: $P_{diplexer} = P_{in}(1-S_{21}^2-S_{11}^2)$. A watt or two feels OK, 5 watts is quite a lot.
The main failure modes I can think of for the diplexer are:
- overheating inductors, slow
- overheating PCB tracks and dielectric, slow
- overloaded capacitors, heating and sudden failure
- Flash-over, unlikely at just 100 W.
I'd suggest, in order of importance:
Set up instruments to measure reverse power, at the radio side, and forward power on both antennas. Monitor these during the tests.
Test for performance at lower power first, observe for any heating.
Use a thermal camera to observe the board and component temperature rise. Thermal cameras don't work on shiny metal like copper wire or bare copper PCB, so for this test you should ideally spraypaint everything black. But that changes the cooling properties, so use some common sense to derate things a bit.
If you don't have access to a camera, don't bother with a "laser thermometer", their 30 mm spot size is too big to see things getting hot. Just touch everything with your fingers, after turning off the RF.
For lossy things, assume that the heating is linear, so if it rises 2 degrees after a few minutes of continuous transmission at 10 W, then it will rise 20 degrees at 100 W. 20 C is probably the maximum safe temperature rise, considering that it might be in hot enclosure outside. For capacitors, any temperature rise at all means you have a problem and they could fail suddenly if you increase the power further.
Test for all frequencies you might use, and then +-5 MHz on these, to find out if the circuit gets hotter at some frequencies.
Test with the "chosen" antenna port connected to a load of 1.5:1 or 2:1 VSWR. (two dummy loads in parallel for 25 ohms). Ideally you would test at a few different phases of reflection, you can do this by putting 10 or 20 metres of cable between (25 ohm) dummy load and diplexer, and changing frequency a bit.
Bump the board a few times while testing, in different directions, harder and harder, to see if anything is loose, or touches and sparks.
Test with the board at -20 C, straight from the freezer. Test with it warmed to 70 C in an oven. It should still behave well (and check its performance with the VNA too).
If it is getting slightly warm at 100 W, then test at higher power than you expect to use. A factor of 1.5 is plenty, less than that isn't worth very much.
Best would be to test at a higher power, with a safe amplifier. Your local EMC lab will have hundreds of watt amplifiers at these frequencies, designed for infinite VSWR, and dummy loads and so on. If it's a non-profit project, they might be willing to stress-test it for you for free.
