October 10, 2019
Engineers can no longer ignore electromagnetic crosstalk. They must understand what it is, how to find it and how to correct it.
Electromagnetic (EM) crosstalk is the interference caused by electromagnetic signals affecting another electronic signal. Engineers may also refer to this phenomenon as coupling or noise.
The EM signals causing the interference are known as the aggressors while the EM signal affected by crosstalk is known as the victim.
Crosstalk occurs via two mechanisms:
Engineers developing system-on-chip (SoC) architectures that ignore crosstalk are taking a big risk. Crosstalk can produce electronic design errors that could lead to delays in reaching the market and cost overruns.
To help understand the complexities of EM crosstalk analysis, engineers can contrast the problem with capacitive coupling.
Capacitive coupling is strong in proximity and weaker at a distance. So, engineers can safely ignore capacitive coupling between signal lines that are far apart. In contrast, inductive magnetic coupling cannot be ignored between relatively distant signals.
EM crosstalk is more challenging. First, the symptoms of the problem do not appear in one metric — like timing failure. Instead, crosstalk often manifests as a degradation in some key performance criterion that varies from design to design. Therefore, identifying the issue as crosstalk is the first challenge.
To make matters more complex, crosstalk usually involves unwanted coupling between digital, analog and radio frequency (RF) blocks. Either one can be the aggressor or victim.
EM crosstalk needs to be identified, debugged and resolved differently in different designs. Traditional solutions involve architecture or software tricks that prevent the modes of operation that trigger the problem. However, this is becoming financially and technically untenable as designs have grown in complexity and speed.
To model EM crosstalk accurately, engineers need to analyze and model a staggeringly complex scope of physical structures, including:
Most of these structures have complex physical layouts that require a large mesh to simulate the resistance, capacitance, inductance, coupling capacitance and mutual inductance.
A second modeling factor that increases the size of crosstalk models is that engineers can’t analyze EM crosstalk by limiting the focus to a small bounding box within the design. Analyzing neighborhood victim signals works well when assessing electrical capacitive coupling. However, magnetic fields can travel along large loops, form outside the immediate neighborhood of a victim signal or encircle the whole layout of the chip.
Additionally, it’s hard to limit the size of a model generated by EM crosstalk tools because it needs to include all the nets that contribute to the crosstalk problem and all the nets and structures that might have an impact on the performance of the circuit.
To be useful downstream in development, the crosstalk model must:
These three requirements are hard to meet given the typical size and complexity of crosstalk models.
EM crosstalk is a big concern for engineers because of the demand for electronic systems to increase in bandwidth and decrease in size. This puts high-speed circuitry and high-bandwidth channels in proximity.
As electronics become smaller, crosstalk will become a bigger problem.
Additionally, the continuous increase in internal clock frequencies (5 to 10 GHz) and the increase in data rates (above 10 Gbps) are also fueling the emergence of crosstalk issues.
In short, fast speeds and small electronics create crosstalk; consumer demands are creating SoC trends that make it impossible to ignore parasitic inductance and inductive coupling.
There are many architectural and application design trends that contribute to crosstalk.
For instance, EM crosstalk is frequency dependent. However, engineers cannot analyze EM crosstalk for a simple frequency of interest.
As an example, a clock signal with fast rise and fall times contains significant harmonic frequency components. So, a clock running at 10 GHz has a 5th harmonic frequency component running at 50 GHz.
Those who target on-chip clock frequencies of 25 GHz, however, will have to think about how to safely model the 3rd harmonic, which falls into microwave frequencies.
EM crosstalk can affect signal magnitudes, or noise level. Hence, the impact of crosstalk is further exacerbated by the decrease in signal voltage levels and the increase in sensitivity to noise driven by lower-power trends in SoC applications.
Ethernet, fiber channel and peripheral component interconnects (PCI) can also be sources of crosstalk. To achieve high data rates, these buses employ multiple serial lanes that operate in parallel. For example, a 100 Gbps ethernet can employ 10 channels that are each running at 10 Gbps. When so many high-speed serial lanes reside in a single system, every lane can be a potential aggressor or a potential victim — a true crosstalk nightmare.
Other architectural trends that increase the likelihood of EM crosstalk include:
SoC integration places high-speed digital circuitry, analog and RF blocks close together. This creates many opportunities for crosstalk inside those components and across various blocks.
Most electronic design automation (EDA) tools are geared for a specific design type — such as digital, analog or RF component design. However, crosstalk is not limited by these boundaries. In other words, the types of analyses engineers typically use to design electronic components might ignore crosstalk.
With the advent of advanced technologies and SoC architectures, ignoring electromagnetic crosstalk is risky.
Learn how to use Ansys Pharos to identify and alleviate crosstalk.
To learn more about crosstalk, watch the recorded webinar: De-Risking High-Speed Serial Links from On-Chip Electromagnetic Crosstalk and Power Distribution Issues.