I find when I perform a RegionDifference on two ImplicitRegions with small dimensions (0.01) then use ToElementMesh the resulting mesh is terrible in V11.1 but perfect in V10.4
Using AccuracyGoal->[one order of magnitude smaller than MaxCellMeasure] and MeshQualityGoal -> 1 helps (both not required in V10.4) but it's still not great
Heya. My Meta post was starred in chat. Cool. Sadly it posted here before I corrected the spelling of "Mathematica" in the title. Sadly this won't be the last time I misspell something. Better a polymath than a polyglot!