A timed loop is always more accurate. That's what is was specifically designed for.
Your idea about the case structure is flawed, you need to ensure that the wait executes in parallel to the rest of the code, otherwise your loop time will be the sum of the wait time + the time it takes to execute the second sequence frame. Remove the sequence structure and see of the loop rate improves.
A wait in parallel will NOT slow down the rest of the code!! All nodes without data dependency will execute at the same time and the current frame wil finish once all nodes in it have finished. If you can ensure that all the other stuff takes less than 1 ms, the wait will be the last to finish and exclusively determine the loop rate. 🙂
Of course you should know that on a multipurpose OS, a 1ms loop rate cannot be guaranteed. You would need to go to a RT system for deterministic behavior.
Message Edited by altenbach on 08-28-2007 09:47 AM