LeSving wrote:
By far the best solution is fuel injection, even a basic mechanical one, and here comes the kicker A standard fuel injection (Lycoming/Continental) is completely free of icing problems. The funny thing though, they all have a butterfly valve and the same manifold pressure for various power settings. The have the same venturi for metering of fuel. It’s only on diesel engines and very modern, direct (into the combustion chamber) digital EFI systems, like that Evinrude system, that you can also take out the butterfly valve. How is this possible without icing problems ? Obviously, the fuel sprayed out of the carb is a key element Get rid of that fuel evaporating all over the place, and the problem is solved.
To me that’s a very persuasive argument.
L J K Setright in his book “The Power To Fly” wrote about the different approaches taken by Rolls Royce and Daimler-Benz with their Merlin and DB601 motors claiming that RR chose the carburettor over fuel injection to benefit from the evaporative charge cooling provided by the fuel in the inlet tract.
Gosh I remember that pompous LJKS from the Bike magazine in the 1970s 
Even fuel injection is not immune from icing but the mechanism is different – it happens around -15C OAT in IMC in the TB20 and the DA42 and probably all other types with that big Brackett air filter at the front and the tapered duct. It has been proved on an instrumented IO540 engine that there is a +10K delta T on the way in, so -15C becomes -5C and the engine stops after some 10-20 mins.
