Differential Pair Nets

Differential pair nets are routing patterns with a certain parallel clearance apart on which reversed signals flow at the same time to send information.

Generally, two signals are represented as +/- or POS/NEG and keep sending stable signals. Differential pair nets are employed to resolve SI noise, return path, PI, EMI issues that appear often in high speed signal nets. Differential pair nets take care of return path and impedance issues by themselves by being side-by-side with uniform distance apart. For common signal trace, designers consider impedance at design stage. However, for differential pair nets, designers must consider coupled odd impedances. In other words, a differential pair net needs to be considered as one coupled signal pair. Generally, two signals are represented as +/- or POS/NEG and keep sending stable signals. Therefore, two signals must keep coupling status at design stage. This item checks the coupling status of paired signals. Near the component pins, due to the pin pitch, the distance may not be able to satisfy the differential pair distance criteria. The Pin Escape option is available to exclude a region near pin locations.
  • Item: Input item name.
  • Net: Select Differential Pair Net Group.
  • Filter: Enter a filter to choose differential pair net from selected net group. After entering a base net and pair net add them into the list by clicking Add Filter.
  • Coupling Ratio (%): Ratio (%) of parallel length that satisfies the differential pair separation condition with total net length. Lower ratio will be a fail in the report.
    • Ratio Fail: If result ratio is lower than this value, violating regions are displayed in yellow and acceptable regions are displayed in cyan.
  • Separation Tolerance (%): Tolerance (%) to measure an allowable distance between nets in a differential pair.
  • Separation: Define parallel segment separation criteria. If you do not enter any specific value, the DFE uses the longest net’s separation by default.
  • Each Layer: Define different separation values for different layers.
  • Layer No: Define layer number. If you do not enter any specific value, the DFE gets layer number from PCB property by default.
  • Separation (Each Layer): Define different separation values for different layers.
    • Separation: Required separation for each layer.
    • Use the separation of longest segment pairs: The DFE uses the longest net’s separation by default. (per layer)
  • Pin Escape: Enter a radius of circular region around pins to be excluded for the rule check.
  • Component Keep IN/OUT: Define test region. For example, you can test inside or outside the breakout region of CPU with different test values.
    • Component Group: Select required component group which is used for defining the test region.
    • Region: Define the required test region.
      • IN: DFE tests the inside of the breakout region of target component.
      • OUT: DFE tests the outside of the breakout region of target component.
    • Range (COC+distance): Enter the distance value to define breakout range. The COC (Component Overlap Check) plus this distance value is considered the test range.
    • Target Layer: Select the required test layer.
      • All Layer: Test all layers.
      • Component Place Layer: Test only component placement layer.
    • Different Rule Setup in Special Region: Assign additional test region for breakout region testing.
      • Extended Region from COC: Enter the distance value to define additional range. The COC (Component Overlap Check) plus this distance value is considered the additional test range.
      • Allowable Values: Assign additional allowable test value for this additional region.
      • Value: Add additional allowable value.
      • Range: Add additional allowable range.
      • Deviation: Add additional allowable value with deviation using this option.
      • Deviation (%): Add additional allowable value with deviation (%).
    • The test range and test values will be:
      • Case1: Component Keep In region test


        Figure 1.
      • Case2: Component Keep Out region test


        Figure 2.
      • Case3: Component Keep Out with extended region test


        Figure 3.
  • TP Component: Specify used test-point component group.
  • TP Separation: If test-points are used as pair, specify the distance between test-components.
  • Nut Shape Anti-Pad: Checks whether the shape of anti-pad form a closed nut shape.


    Figure 4.
  • Except Layer: Select exception layers to exclude Nut Shape Check.
  • Except Serpentine: DFE does not check serpentine section of net.
  • Except Arc: DFE does not check ARC section of net.
  • Display Separation & TP Distance Result: PollEx DFE report will show segment/TP separation information.
  • Use Edge-to-Edge Separation: Checks edge-to-edge separation.

    Default: Center-to-Center

A Differential Pair Net usually keeps a minimum 80 percent of total net length parallel. The pair nets may be allowed to have a distance tolerance in parallelism when a net go around via or other objects on board. In this case, it is recommended to make those distance tolerances less than 5 percent.


Figure 5.

To design a Differential Pair Net, you must pay attention to keep the continuity of power and ground planes above and below of the differential net. Do not make branch stub to make test point. In other words, impedance for overall routing must be kept the same for the pair nets.

Not only for narrow side differential pairs but also for broad side differential pairs, the rule explained above needs to be applied in design.


Figure 6.

Reference Material

Parallelism will cause impedance discontinuities that will directly affect signal quality.

In this case it also contributes to the trace-length mismatch and will cause an increase in signal skew.


Figure 7.
  • Intel - High Speed USB Platform Design Guidelines