From what I've observed, if they only get ASCII they write out the diacritics phonetically. As an example, https://en.wikipedia.org/wiki/Chris_Huelsbeck interchangably uses Hülsbeck and Huelsbeck, but never Hulsbeck