How best to proceed with mixed ANOVA follow ups

I have a bit of a problem. My data is non-normal and violates the assumption of spherity, but I have one between subjects factor (only 2 levels) and one (and a second in some analyses - 4 levels and 2 levels respectively) within subjects factor. I have done mixed-model anovas so far, because of the mixed design rendering Kruskall Wallis and Friedman tests inappropriate. I intended simply to confirm any significant results of the ANOVA with Mann-Whitney U's, as its the group differences I'm interested in across each level of the within-subjects variables. This will then control Type I error I think...

I'm wondering whether it will be sufficient to just use the Mann-Whitneys as follow up tests as well, or whether it would be best to use T tests? It seems a bit futile to do both, but inconsistent with the original use of parametric statistics just to do the Mann Whitneys.

Any advice would be appreciated. I'm really lost

Thanks all :)