Re-using their design might be no good idea - there are cheaper designs for sure. This new DES cracker would just need to try every possible key - like the one of the EFF already did.
DES was a big standard for encryption, so some people did build such machines, right? Of course did they: COPACOBANA is able to break DES in under 9 days and costs under 10,000 US $ - That was 2006. You could even run a few of them parallel and crack algorithms like DES even faster - or you just use RIVYERA and crack DES in 1 day. You can use more than one machine of them, too.
More informations about the DES cracker and the mechanic behind it can be found in this document.
I'm pretty sure there are even better methods as of 2015.