I worked at company in SBIR phase 2 and one quirk I found strange was that the money could be used for R&D not things deemed “production”. Endless 3D prints were ok, random software features were ok, tooling for injection molding was not. It wasn’t a particularly onerous requirement, but I think reinforced the tendencies of our former academic founders to spend a lot of time on research and none on development.
Part of it is figuring out how to classify "production" as R&D. Well we need to test a prototype and in order to that we need it made as the same was as production due to structure property process relationships.
Or manufacturing defects is going to be a huge source of bad thing so we need to do some research into the manufacturing process.
