The HMC659 output is rated 9 volts abs max. ADI says to take this very seriously, but they would. So I have a 1 pF schottky clamp diode (even that 1 pF is annoying) and then the biased-up bypassed 8.7v zener. It might have to dump 50 mA.
I'd test the amp chip for actual breakdown, but we only have one in stock now and they cost $330 each.
This is nominally an RF part, so it's likely it will really swing to at least 2x Vcc. That's something the RF boys tend to do, specify max supply voltage and not max device voltage. I'll eventually try it and maybe delete the clamp.
I want all the swing I can get, but we only need to bias it high and pulse it in the down direction.