PC HDDs' have BLDC motors (Brushless DC), which are not really PWM friendly; they basically have a 3-phase bridge driving the windings, and the drive electronics monitors the speed of the platters.Between a rock and a hard place for sure!
PC HDD's spin fast enough and run on DC which could be regulated with PWM down to 717 RMPs, but you would have the time of your life balancing your sample. This, of course, is a bit of humor. Do not attempt this at home!!!
PC fans also have BLDC motors, but they frequently also have PWM inputs as well as outputs that give feedback as to the motors' speed.
One of the things about a centrifuge is that the rotor is a relatively large mass; the samples are fairly insignificant in comparison. This is important as the rotor comes up to speed; if there was a large relative imbalance, the motor bearings would be damaged to to the unequal force applied, and the centrifuge could start "walking" around; sort of like a washing machine slamming around during the spin cycle, but at a much higher speed.
I think you need a bigger fudge factor; the standard RPM of that centrifuge results in about 22 times as many G-forces as you want!I had that hunch too, as I'm pretty sure I made a couple false assumptions with that math...it was more of an estimate. I would expect it to be less than 3.2 minutes realistically. Lets give it a fudge-factor of ±50%
Care to share what the samples are that you're intending to spin?