USB without Vcc
Yes, communication should fail - the reason is, devices are not allowed to feed any current into D+ or D- (which is how they communicate) until they observe Vbus (the 5V) high. This is actually guaranteed by the cable in many cases, because the device will 1) only have power from Vbus, and 2) not be connected to D+/D- until after Vbus and Gnd. Obviously 1 does not apply to self-powered devices.
That's assuming the device operates to the USB spec, and many devices do not. Another part of the spec they tend to ignore is that they may only draw 100mA until more has been negotiated. Requests for more are typically rejected only when connected to bus-powered hubs, as they don't have more power to grant. That's also why they tend to have four ports, as the hub registers to draw 500mA, needs a bit of power itself, and then must provide those 100mA on each downstream port.
In all, you'll probably have to modify the device or its battery connection rather than the USB cable to prevent charging. The specifics will depend on your device.
The spec allows a device to expect to have 100mA available.
- Communication should not fail.
- No, no device should pull power from D+.
- I often design devices to use the USB power to power whichever chip I use as an interface. This means that without 5V power my communication chip does not power.
In the end you may just want to try it out and see if your device specifically has a problem. What is the device that you need to disconnect power for? If you have a device that has a primary function of charging a battery they probably designed the device around your power as a charger does not need to function if there is no power.
Also, you will still have ground as a reference, and the differential signal uses differential signaling with an occasional common mode signal but it should work with a ground reference, unless they designed their circuit as I noted above.