@rachel yep, the very slow time to blow was expected, which is why I fitted a 2A polyfuse there instead of 3A which is the limit of the buck converter. The load was 1.2ohms so theoretically 2.75A but the 10mohm current shunt resistor read only 0.022V so actual load closer to 2.2A.

And a reflow on the hotplate did the trick, there was a short between the charge pump cap output pin and the input voltage. I suspect it was near the capacitor itself but it’s sorted out now.