I've had a similar experience connecting a USB to a micro controller development system. In my case the problem turned out to be a momentary surge. Basically, "ground" on one device was not the same as "ground" on the other, likely because of other equipment I had connected to the controller. For me, adding an extra ground wire, separate from the USB cable, solved the issue. Fortunately for your case, the DB-9 shielding is accomplishing something similar, diverting any differential between the grounds to point within the tablet's circuitry less sensitive to such grounding differences. Its good you found this, but I'd dig a little deeper too. For example, is the +5V output from the USB contacting something it shouldn't on your custom D9 (like the +5V from the tablet)? Even if its near the same voltage, you don't want both your ground and +5V source from both machines joined. I'm curious if your tablet is connected to anything else, such as a powered speaker system, which may be picking up an external ground. If so, it would be interesting to measure any voltage (both DC and AC) that exists between the D9 shield (pin-1 I believe) and its mating contact. Sometimes the metal case of a D9 connector (if it is metal) is connected to the shield too. If you see nothing measurable there, do the same with the other pins, and make sure any and all voltage differences you measure are explained and expected. It seems you've solved your problem with the shield, but its worth doing the additional investigation here. You just might spare yourself possible damage to tablet's USB port, or just as bad, other circuits that sometimes share the USB on a tablet (like the battery charging circuit).
