The order of bytes is maintained and no structure is imposed by USB. Stream pipes flow data from source to destination. The protocol used to connect between the host and a function endpoint is known as a pipe. The host controller initiates all transfers. Because the SuperSpeed data is routed to its destination, not broadcast, devices that are not the target of the communication can remain in a low-power state. (USB 2.0 did allow for a charging downstream port that was not compliant with the standard but could supply more power.) Enhanced SuperSpeed provides additional power management capabilities. As the standard has grown, the power delivery capabilities of a connection have grown from relatively modest to very substantial: 0.5 W for USB 1.0, 2.5 W for USB 2.0, and 100 W in the new USB Power Delivery specification. Data on the SuperSpeed path is encoded using an 8b/10/b encoding while SuperSpeedPlus uses a 128/132b encoding these encodings provide a more sophisticated form of transition management for clock recovery.įunctions may be self-powered or draw power from USB. The Enhanced SuperSpeed USB 3.1 architecture adds four more lines for the high-speed data which flows separately from the low/full/high-speed data provided by USB 2.0. The stuffed bit is then removed by the receiver. This problem is solved by bit stuffing: a zero is inserted after every string of six consecutive 1 s.
However, a long string of 1 s would result in no transitions, leaving the circuits no information from which to infer the clock.
A clock can be extracted from the data stream at each end of the connection by monitoring the transitions. The data signals use nonreturn to zero inverted (NRZI) encoding: no change in the signals indicates a 1 while a change indicates a 0. Differential signaling is used to improve noise immunity. USB 2.0 uses a four-wire cable to connect the nodes of the network: a power signal V bus, ground, and two data lines D + and D −.