If you try to send the signal (e.g., "1" ) from port 888 for 5 vs 500ms in the TTL I/O test utility, do you find you get the same results? As far as the onset delay, no, I don't think there is a built in way to do that in MediaLab--especially with that degree of accuracy. DirectRT, I think could be rigged to do that depending on the design of your study. ML could approximate it by duplicating the item--fixing it for 200ms and then repeating it with an immediate TTL at the onset of the repeat. Does that make sense? Not a very elegant workaround though.