A single-ended signal is transmitted through a single signal line with the ground plane as the reference. In other words, a single-ended signal is the potential difference between a single conductor and the ground plane during transmission. This requires the reference ground potentials at the signal source and receiver to be basically the same for the signal to be transmitted from the source to the receiver correctly.
I. Single-ended Signals vs. Differential Signals
1.1 Definition
A single-ended signal is transmitted through a single signal line with the ground plane as the reference. In other words, a single-ended signal is the potential difference between a single conductor and the ground plane during transmission. This requires the reference ground potentials at the signal source and receiver to be basically the same for the signal to be transmitted from the source to the receiver correctly.
A differential signal is transmitted through two lines: one carries the positive-polarity signal and the other the negative-polarity signal. The receiver judges and identifies the signal by comparing the voltage difference between the positive and negative signals. The advantage of this is that the receiver can still identify the signal correctly even if the reference ground potentials at the source and receiver are inconsistent.
1.2 Transmission Differences
Single-ended signals are based on the reference ground plane. When a DC signal flows through the reference ground plane, there is almost no potential difference between the source and receiver sides of the ground plane; when AC signals, large current signals (especially high-frequency signals) flow through it, a potential difference will be generated due to parasitic inductance. The magnitude of the potential difference is affected by the signal operating frequency, rise/fall edge slope, operating current, and parasitic inductance of the reference ground plane.
Although both lines of a differential signal reference the ground plane, the two signals float simultaneously (under ideal conditions) when the ground plane floats, and the voltage difference between them remains almost unchanged—since the receiver identifies this difference, differential signal transmission has much lower requirements for the reference ground plane compared to single-ended signals.
When a signal passes through the magnetic field of another signal during transmission, an induced electromotive force is generated on it. For single-ended signals, the induced electromotive force is directly superimposed on the signal; for differential signals, the induced voltages on the two lines are equal, their difference is zero, and thus no impact is caused on the useful signal. This is the reason why differential signals have stronger anti-interference ability than single-ended signals.
II. Advantages and Disadvantages of Differential Signals
2.1 Advantages
- The receiver of differential signals judges the signal by identifying the voltage difference between the two lines, so the accuracy of the reference ground potential has a small impact on differential signals.
- Differential signals have strong anti-interference ability and low inherent EMI radiation.
- In a single power supply system, differential signals can process bipolar signals accurately and easily.
2.2 Disadvantages
Differential signals require the two signals to have equal amplitude, 180° phase difference, opposite polarity, and equal line length of the two transmission lines. Since the receiver compares the voltage difference between the two signals, phase and time delay are critical for differential signals—a problem that does not exist for single-ended signals.
III. Design Rules for Differential Signals
3.1 Tight Coupling Principle
When the two differential lines are tightly coupled, the currents on them are equal in magnitude and opposite in polarity, and the magnetic fields generated are also equal in magnitude and opposite in polarity, canceling each other out. Another advantage of tight coupling is that external noise voltages induced on the two lines appear as equal common-mode noise, and the receiver is only sensitive to differential-mode signals (insensitive to common-mode noise), thus suppressing common-mode noise at the receiver end.
3.2 Equal Length and Equal Spacing Principle
Differential signals must maintain equal electrical length, and the spacing between the two traces must be consistent throughout the entire length. A change in spacing will cause unbalanced magnetic field coupling, reducing the effect of magnetic field cancellation. In addition to increased EMI, a change in routing spacing will also cause signal impedance variation, leading to impedance discontinuity and signal reflection that damages signal integrity.
Routing with the same electrical length ensures that the two signals reach the receiver at the same time. For differential signals of the same length, the two signals are equal in magnitude and opposite in polarity, so their sum is necessarily zero. If the electrical lengths of the traces are different, the signal on the shorter trace will change state earlier than that on the longer one. In severe cases, the driving currents on the two traces will be the same at a certain point; when the two signals are added, the total signal transitions from zero level during the change process. Under high-frequency conditions, this causes the differential signal to flow back to the source through the reference ground plane, forming a loop antenna and radiating outward.
3.3 Controlled Impedance Principle
The impedance of differential signals is determined by the physical geometry of the signal traces, their relationship with adjacent reference layers, and the PCB dielectric. These geometric parameters must remain consistent throughout the entire trace length. Impedance discontinuity occurs when the differential impedance deviates from its standard value (100±15%), which can cause signal reflection due to impedance mismatch and further damage signal integrity.
Formula for characteristic impedance:
3.4 Complete Return Path Principle
For high-frequency circuits, providing a relatively complete reference plane on the adjacent layer can offer the minimum impedance path for the return current, making the magnetic field generated by the signal cancel out with that generated by the return current and minimizing EMI. Crossing a partition will cause the area of the signal return path to be out of control—the magnetic fields from the signal and return current cannot cancel out effectively, resulting in poor EMI radiation. Due to the different differential-mode loops of the differential signal pair, the induced noise magnitudes are also different, and common-mode noise cannot be eliminated effectively at the receiver, leading to degraded signal performance.
IV. In-depth Analysis of the Return Path of Differential Signals
4.1 Analysis of Differential Signal Return Paths
Misconception about Differential Signal Return Paths
Most electronic engineers believe that the main reasons for the strong anti-interference ability and low spatial radiation of differential signals are: the positive signal flows from the source to the load through the transmission line, and the negative signal flows back to the source from the load—the differential signal currents flow in a closed loop, the positive and negative signals have equal amplitude and opposite direction, the generated magnetic fields cancel each other out, and the differential signal return current does not flow back to the source through the reference ground plane.
Analysis of Differential Signal Return Path under Ideal Conditions
In fact, the return path of differential signal currents is the same as that of single-ended signals: the D+ signal current flows from the source to the load through the transmission line and then back to the source through the reference plane; the D- signal current flows from the load to the source through the transmission line and then back to the load through the reference plane.
Under ideal conditions, the differential signal currents are equal in magnitude and opposite in polarity, and their generated magnetic fields cancel each other out; the return currents are also equal in magnitude and opposite in polarity, and their magnetic fields also cancel each other out—thus the spatial radiation of differential signals is extremely small.
When the reference ground potentials at the source and receiver are inconsistent under high-frequency conditions, the positive and negative differential signals at the receiver rise or fall synchronously relative to the reference ground plane, while their voltage difference remains unchanged, having no impact on signal identification by the receiver. Similarly, when differential signals are in a magnetic field, the induced voltages on the two lines are equal, the voltage difference between the differential signals still remains unchanged, and there is also no impact on the signal transmission result—this is why differential signals have strong anti-interference ability.
4.2 Two Design Forms of Differential Signal Return Paths
Return Path Design for Differential Signals in Multi-layer PCBs
In multi-layer PCB design, differential signals usually use the adjacent complete ground (power) plane as the current return path, minimizing the loop area. The magnetic fields generated by the signal current and return current cancel each other out, resulting in minimal spatial radiation.
Return Path Design for Differential Signals in Double-layer PCBs
Due to PCB routing density constraints, it is almost impossible to design double-layer PCBs with single-layer routing and a complete reference plane on the other layer. Ground shielding on both sides of differential signals is usually used as the current return path; the magnetic fields generated by the signal current and return current cancel each other out for the most part, resulting in low spatial radiation.
Form 1: Top-layer routing with bottom-layer reference plane
This routing method seems the same as that of multi-layer PCBs, but there are significant differences in practice. The specific design requirements are: the differential signal routing from the source to the receiver is completed entirely on the top layer, and the bottom layer corresponding to the entire top-layer routing trace of the differential signal must maintain a relatively complete reference ground plane to provide a low-impedance return path for the differential signals.
Design Difficulties:
- It is feasible to keep differential signals on the same layer in double-layer PCBs, but it is extremely difficult to maintain a complete reference ground plane on the bottom layer corresponding to the entire differential signal trace, especially while maintaining product cost advantages and versatility.
- If the differential signal uses the bottom-layer ground plane as the return path in a double-layer PCB, the distance between the top-layer differential signal routing and the ground shielding on both sides must be less than the PCB thickness between the top and bottom layers; otherwise, the differential signal return path may switch to the ground shielding on both sides.
- When the bottom-layer reference ground for differential signals is complete, the connection between the bottom-layer reference ground plane and the source/receiver reference ground must also be complete and low-impedance—this is particularly challenging for the design of BGA-packaged devices.
Form 2: Cross-layer routing (top and bottom) with ground shielding on both sides as the reference plane
To balance cost and performance, double-layer PCBs usually adopt cross-layer routing (top and bottom) with ground shielding on both sides as the reference plane. This design often leads to reference ground plane layer changes, even simultaneous layer changes of the reference ground plane and differential signals. Key control points for this design include the integrity of the reference ground plane, the equipotential between the ground shielding on both sides and the bottom-layer reference ground plane, and the processing of cross-partition routing (a major difficulty).
Design Difficulties:
- The ground shielding lines on both sides of the differential signal traces must be complete from the source to the receiver, so that the magnetic fields generated by the return path cancel each other out for the most part and minimize radiation.
- When the differential signals change layers, the ground shielding lines on both sides also need to change layers; ground vias must be placed on both sides of the differential signal layer-changing vias to make the signal return path change layers synchronously and minimize radiation.
- If the differential signals themselves do not need to change layers but the ground shielding lines on both sides do (due to structural layout, chip pin arrangement), the layer-changing ground lines must return to the main chip through the minimum-area path.
- The ground shielding lines on both sides of the differential signals must be connected to the system reference ground plane through ground vias to form an equipotential body; the arrangement and number of ground vias require special attention.
- The length difference of the differential signal pair itself and the length difference of the ground shielding lines on both sides are important factors for magnetic field cancellation, and the minimum possible length difference between them should be maintained.
About Maxipcb
Maxipcb empowers innovators to turn cutting-edge technologies into reality.
We offer one-stop solutions for design, simulation, testing, PCB manufacturing, component procurement and SMT assembly, enabling efficient development, rapid deployment and risk control across the full product lifecycle.Serving the world in communications, industrial automation, aerospace, automotive, semiconductor and beyond, we build a safer, more connected future together.
