The coding of the conditions is fine (the first 4 EVs). Also, the intercept, age and sex (the last 3 EVs) are also fine.

However, you can safely remove all the interactions from the model, that is, EVs 5 through 15 (inclusive), because there is nothing to interact really: the first 4 EVs are already completely separable, such that no interaction can be further interrogated by additional regressors.

About the message with TFCE: I'd change the design first, which will mean various of the contrasts will need be removed. The message may go away then as it may be coming from one of them.

All the best,

Anderson