hej,
the arduino is also having some timing issues, the delayMircoseconds don't work so well for delay between 2-10 uS, but the delay for 1 uS is working perfekt. So for a delay with 4 us you should write 4x delayMircoseconds(1), the same for the 8 us delay, the 52uS delay should work on the normal way.
One the other hand you should use the c code for writing the pins. Just copy it from the arduino library, this will save some time.
You can also have a look in the DMX512 code example on the arduino wiki (not on the arduino website, the one on the website is not working!) For DMX you also have to use 4uS delays.
looking forward to your results
tomek