Make sure to connect the motor using the bipolar parallel schema as per the datasheet.
Additional Information:
I also purchased these same motors(3a drive, not the 8.8) nearly a decade ago, they are now being installed on their third life as a CNC wood lathe with C axis spindle control. I never had any perceivable problems wiring them to the diagram, but in recent R+D I discovered a strange thing. I use LinuxCNC, and while going through setup and trying to tweak in the settings for the motors/drives, I found a motor wire identification guide for steppers on the LinuxCNC wiki that gives a different result for which wires go to which terminals on the drives. I have not tested both wiring schemes since I am in the middle of the build and have not yet reached that point. This could be one factor, if the wiring diagram is not actually correct. I always used bipolar parallel. As stated above, I never detected any problems in thousands of hours of run time over two separate machine configs. The other thing I learned recently is that microstepping actually reduces STEP torque by significant amounts, though I can't find any info about its effect on HOLDING torque, if any. 1/16 microstep reduces STEP torque to ~10% of motor rated torque, which can have a significant effect on heavy cuts/fast feedrate moves. I have definitely overdriven my axes by not programming conservatively enough, I do a lot of aluminum routing and hoggy work on hardwoods. There are many other factors in machine builds(resonance, mechanics, etc) that can eat away at the motors ACTUAL real life torque capability vs. its nominal rating. Lot of good info on stepper principles on the LinuxCNC website. My current build has actual ballscrews rather than leadscrews, It will be interesting to see what the performance differences are compared to my other builds. Keep the chips flying!