Thanks for those workarounds Don. Does seen strange to JBridge it but if it works it works. Not sure if I'm willing to shell out for JBridge just for this though (or maybe I should bite the bullet and see if it helps with Cakewalk's tape sim crashing as well).
However, I did receive a reply from Waves this morning, and this is what they said (hope they don't mind me sharing):
"We were able to reproduce this issue on our test machines.
Like you describe, this happens when SuperTap is placed after a VSTi in Sonar X2.
Our QA has acknowledged this as a bug.
We will be in touch with CakeWalk, and work with them to solve this.
A fix will be issued either by Waves or in a future iteration of Sonar, depending on which code is causing this.
Many thanks for reporting.
We will be working on a fix for this this ASAP.
As for the rounding of entered values, this is by design.
You'll also notice that the slider moves at a certain resolution when moved by mouse.
If you want to get a finer ms resolution,
you need to increase the (internal) plugin BPM when it is un-synced (manual mode) from the host clock.
Supertap resolution goes from 40BPM up to 1200BPM.
Explanation:
you have a time resolution of 8ms under 120BPM clock,
you can double the BPM (240) to get 4ms resolution, 480BPM gets you to 2ms and 960BPM will get you to a 1ms resolution.
Hope this helps
and thank you again for reporting this bug and for your patience."
So hopefully we'll see a fix in due course. At least I now know why the plug was rounding ms values as well.