Definitions of unipolar, polar, bipolar, RZ, Manchester codes

Line codes are electrical waveforms used to represent binary data for transmission over a baseband channel. Each code below is shown for the example bits 10110, with bit period $T_b$ and level $V$.

Unipolar (NRZ) code

Binary 1 is sent as $+V$ for the full bit period, and binary 0 as zero voltage. It is simple, but it has a DC component, and long strings of 1s or 0s carry no timing information.

Unipolar NRZ

    1   0   1   1   0  
+V ────|   |───────|   
 0     |───|       |───

Polar (NRZ) code

Binary 1 is sent as $+V$ and binary 0 as $-V$ for the full bit period. There is no DC component for equally likely bits, and it has the best noise immunity of these codes for a given power. Long runs still give no transitions.

Polar NRZ

    1   0   1   1   0  
+V ────|   |───────|   
-V     |───|       |───

Bipolar (AMI) code

Binary 0 is sent as zero voltage. Binary 1s are sent as pulses of alternating polarity, $+V, -V, +V, \dots$, usually half-width (RZ). It has no DC component and narrow bandwidth, and two 1s of the same polarity in a row show an error has occurred. A long run of 0s loses timing; HDB3 and B8ZS solve this.

Bipolar (AMI, RZ)

    1   0   1   1   0  
+V ──|         |─|     
 0   |─────| |─| |─────
-V         |─|         

Unipolar RZ code

Binary 1 is sent as $+V$ for the first half of the bit, then returns to zero. Binary 0 is zero for the whole bit. Each 1 gives a transition that helps timing, but the bandwidth is double that of NRZ and there is a DC component.

Unipolar RZ

    1   0   1   1   0  
+V ──|     |─| |─|     
 0   |─────| |─| |─────

Manchester (split-phase) code

Each bit has a transition at its middle. Binary 1 is $+V$ for the first half and $-V$ for the second half; binary 0 is the reverse. It has no DC component and is self-clocking, which is why it is used in 10 Mb/s Ethernet. The cost is double the bandwidth of NRZ.

Manchester

    1   0   1   1   0  
+V ──|   |───| |─|   |─
-V   |───|   |─| |───|