When making a molecule from the SMILES [CH] the carbon ends up with 0 radicalElectrons instead of 3.
(There does not seem to be a problem with [CH2] or [CH3])
I have added a (failing) unit test in ca3b3bc
I notice that ChemRes make a complete hash of these radicals: http://cactus.nci.nih.gov/chemical/structure/%5BCH%5D/formula
But the syntax at http://www.opensmiles.org/spec/open-smiles-3-input.html is clear enough.