In this paper we present and evaluate four delay queues designed for application tailored Ravenscar hardware realtime kernels. The properties of the different queues and optimisations of them are discussed and both formal models and actual hardware implementation of the queues are presented. A transformation from timed automata to VHDL is described during the translation of the timed automata of the formal model into the corresponding VHDL state machines. Our study of the queues shows that even though parallelism costs much in terms of chip area, there are system configurations where it is the most space conservative. We also show that the queues meet the timing requirements of Ravenscar and that they can be fitted onto an FPGA.