perfect for a special serving of Carb Sorbet
It begs the question of
why lower temperatures are
less perfect.
I think the bottom end of the carb icing envelope is determined by the amount of water vapour in the air.
1) If you start at an OAT of 25 degC (saturated or near) and cool to -5 degC in the carb, there's 20 g/m^3 of water that condenses out into supercooled droplets ready to form ice.
2) If you start at an OAT of 5 degC (saturated or near) and cool to -25 degC in the carb, there's something like 5 g/m^3 of water that condenses out. That's much less of a problem from an icing point of view, which is why the -25 degC in the carb is less troublesome than -5 degC in the carb.
3) If you start at an OAT of 5 degC (saturated or near), heat to 25 degC and then cool to -5 degC in the carb, there's only about 4 g/m^3 of water that condenses out, even less than in the previous case.
You might argue that a greater proportion of the water in case 2 would spontaneously form ice than in case 3, but I think the processes that go with glaciation are much slower than the time in which the air is cooled in the venturi and then slams into surfaces that can collect ice. You'd expect all the condensed water to be capable of causing ice.