I figured it out, and it is rather strange and consists of two things.
1) .ttf and .eot I received from the font-protein was not good. Getting the original .ttf back from the font site and converting it to .eot yourself was necessary for the correct transfer. I donβt know if this is really a font-protein mistake, because the differences in output seem very arbitrary and strange, see also point 2).
2) Removing the .woff entry from css. Woff was added only for security, I was not sure if it was used by anyone, but it seems that all browsers still display it correctly.
It seems very strange, because in browsers everyone uses the same .ttf (except IE8 and less, which use .eot), but they do it differently. In addition .woff must be removed, why? Although I solved my problem, it still seems like a rather random and strange problem. If someone knows more about this, I would love to know.
Hope this question is helpful to others. font-face is still a long way to being easy to use, and browsers need to be more careful about matching this correctly.
source share