Thanks, oPossum.
You post provided hints which helped me sold the problem, although I did not use your three PWM solution. Because my PCBA was only in the proto stage, I could swap P2.0 with P2.4 (with some microsurgery on 0.2mm tracks). And voila! It all works beautifully. I used Timer0 for a 1ms timer tick, and TA1CCR1 and TA1CCR2 for the independent PWMs. A re-spin was required anyway, so it is easy to swap the tracks.
After I got this going the rest of the coding to vary the duty cycles dynamically was a piece of cake. The firmware part of the project is complete.
If you ever travel to Melbourne, Australia, I will shout you a beer
.
Incidentally, the MPS430G2333's are very nice, but for pin-steering there are newer devices that are far more flexible. Code Composer is OK, but in my opinion IDEs with register wizards like CodeVision/AVR (for AVR's) kills it as far as speed of development goes. But it's free and it works, so I cannot complain.
Most importantly, I can source the 28-pin MSP420G2333 during the global shortages, where as STM is a basket case. This product has the potential of being very high volume, and when STM finally decides to free up supply, I am not interested. If chip vendors cannot supply to the engineering fraternity now, they will lose out in the long run.
Once again, THANKS!