Serial data transfer

from Wikipedia, the free encyclopedia
Comparison of parallel and serial data transmission

Serial data transmissions transmit digital data autonomously on one line (or on a pair of lines). In contrast to this, with parallel data transmissions, data is transmitted synchronously over several lines. The fundamental difference is that serial transmissions do not have to take into account the runtime differences between different lines, which allows much higher clock frequencies. The name serial is associated with misconceptions, since in principle every data transmission works serially. A better name is bit-serial data transmission (as opposed to byte-serial transmission of a Centronics interface ), but this name also arouses false associations, since several lines can also be used in parallel for serial data transmission (e.g. PCI-Express, Gigabit Ethernet, HDMI) and more complex modulations can be used that no longer recognize individual bits (PCI-Express, USB 3.0, USB 3.1, SATA, Ethernet from Fast Ethernet). Various serial interfaces are defined for serial data transmission . These include plugs, voltages, modulations, protocols used and software interfaces.

Nowadays, with a few exceptions (DDR RAM connection to CPUs and legacy interfaces), only serial data transmissions are used. Even ultra-fast AD converters use serial interfaces to output the converted data. In the past (up to the 1990s) serial interfaces were used for slower data transfers (up to approx. 10 KByte / s) over possibly longer distances (several hundred meters), parallel interfaces for faster transfers (up to 1 MByte / s) over shorter distances .

Explanation of terms

Ground (GND connections)
Ground lines are low-resistance connections that are used to compensate for potential differences between two end devices.
For this purpose, a ground line has to have a significantly lower resistance than the sources of the potential difference, otherwise this only works partially.
single-ended (unbalanced)
With a single-ended transmission, the receiver has GND as reference potential (in the hope that it will be the same with the transmitter). The signals are transmitted via pairs of lines, which consist of a shield (mostly GND) and an inner conductor (e.g. data line) as with coaxial conductors. In spite of this, one speaks here of unbalanced transmission because the shield protects the inner conductor from external influences, but not the other way around.
Differential (balanced) transmission
Here the difference signal is generated in the receiver by 2 lines of equal value in order to recover the useful signal. Common mode interference thus stand out. Smaller potential shifts do not disrupt the transmission.
Asynchronous
Asynchronous data transfers transmit individual data words (5 to 16 bits) each individually and independently of one another. It follows that all data sent require synchronization information (e.g. start bit, stop bit, see RS-232 ). The start bit synchronizes the transmitter and receiver for a transmitted data word. If several data words are transmitted one after the other, each data word is provided with its own synchronization information in the form of start and stop bits.
Synchronous
With synchronous data transmission , the data is grouped in blocks and transmitted together. In the case of synchronous transmission, individual start bits are no longer required for each data byte. The user data is grouped in larger blocks or transmitted as a continuous data stream. This makes the transmission more efficient. HDLC and SDLC from the 1970s were early representatives of this transmission technology .

Transmission media

Cable for serial data transmission

An electrical line is usually used as the medium for serial data transmission , but glass fiber, wireless connection (radio transmission) or another medium is also conceivable. Often data is also stored serially, e.g. B. magnetically with magnetic tapes or the hard disk or optically with CD / DVD (only one head per disk surface).

Serial data transmission was always used when the transmission medium is limited (e.g. to as few individual wires as possible) or represents a cost factor. Basically, this is done at the expense of transmission capacity. If the transmission capacity is more important, parallel data transmission used to be an option (see also bus systems ). B. PCI bus . Due to the advances in semiconductor technology, there are now such fast, inexpensive serial-parallel converters such. B. UART (Universal Asynchronous Receiver Transmitter) called that, for example, the cabling effort with parallel data transmission is becoming more and more important. Because with ever higher transmission rates it becomes more and more difficult to keep the so-called clock skew and the crosstalk on the neighboring line small enough with parallel data transmission.

Clock offset

In the case of synchronous serial data transmission, a clock (so-called "clock" or clock signal) can be sent on an extra line in order to signal when a bit is present on the data line. However, the use of an additional line can lead to problems: The clock skew ( English clock skew ) describes a time offset, which the individual signals can no longer be simultaneously arrive at the receiver due to non-identical line parameters. There are runtime differences that have to be waited until the next date can be transmitted. This limits u. A. the maximum achievable transfer rate. On circuit boards with high data rates, attempts are made to minimize the clock skew by using meandering lines. The causes for the clock skew are i. d. Usually of a physical nature and are related, among other things, to the cable length, temperature fluctuations, material defects or capacitive coupling .

properties

In the following, some terms or features are listed that are basically assigned to every serial transmission standard. A distinction is also made between the properties of the physical interface hardware and the protocols .

  • Connector structure, pin assignment
  • Differential (balanced) transmission or non-differential transmission
  • Voltages, currents, impedances, terminating resistors, wavelength (for optical transmissions)
  • DC voltage component, galvanic isolation
  • Unidirectional: simplex, bidirectional: half-duplex, full-duplex
  • Line coding or modulation
  • Self-clocking or with an additional clock signal
  • Number of transmission channels, per direction
  • Hardware or software handshake
  • Transmission error handling : parity , CRC , Hamming distance etc. (see coding theory )
  • Point-to-point connection (P2P) or multipoint (serial bus)
  • Arbitration: multimaster or master slave
  • Real-time capability: e.g. B. required for field buses
  • With data buses arbitration: priority control via tokens , CSMA etc.

There are various standards for serial interfaces that can be used for serial transmission.

literature

  • Karl-Dirk Kammeyer: message transmission. 4th edition. Vieweg + Teubner, 2008, ISBN 978-3-8351-0179-1 .
  • Ekbert Hering, Klaus Bressler, Jürgen Gutekunst: Electronics for engineers and natural scientists . Springer Verlag, Berlin / Heidelberg 2014, ISBN 978-3-642-05499-0 .
  • Kristian Kroschel: data transfer. An introduction. Springer-Verlag, Berlin / Heidelberg 1991, ISBN 3-540-53746-5 .
  • Carsten Harnisch: Network technology. 4th edition. Hüthing Jehle Rehm publishing group, Heidelberg 2009, ISBN 978-3-8266-9418-9 .
  • Bernd Schürmann: Computer connection structures. Bus systems and networks. Friedrich Vieweg & Sohn Verlag, Wiesbaden 1997, ISBN 3-528-05562-6 .

Web links