Categories

PCB Design Techniques for High‑Speed FPGAs

It would be wonderful if high‑speed PCB design were as simple as connecting schematic nodes and as elegant as what we see on a computer monitor. However, unless the designer is new to PCB design or extremely fortunate, the reality is rarely that effortless. Before a design finally works and someone acknowledges its performance, PCB designers face many new challenges. This is precisely the current state of high‑speed PCB design – design rules and guidelines continue to evolve, and if we are lucky
Aug 26th,2026 74 Views

It would be wonderful if high‑speed PCB design were as simple as connecting schematic nodes and as elegant as what we see on a computer monitor. However, unless the designer is new to PCB design or extremely fortunate, the reality is rarely that effortless. Before a design finally works and someone acknowledges its performance, PCB designers face many new challenges. This is precisely the current state of high‑speed PCB design – design rules and guidelines continue to evolve, and if we are lucky, they lead to a successful solution.

Most PCBs are the result of collaboration between schematic designers – who have a deep understanding of how PCB components work and interact, as well as the various data transmission standards that form the board's inputs and outputs – and layout specialists, who may know little or even nothing about what happens when the tiny schematic connections are translated into printed circuit copper traces. Typically, the schematic designer is responsible for the success or failure of the final board. However, the more a schematic designer understands about good layout techniques, the greater the chance of avoiding major problems.

If the design contains a high‑density FPGA, there are likely to be many challenges lying ahead for even the most carefully crafted schematic. Hundreds of I/O pins, operating frequencies exceeding 500 MHz (and possibly higher in some designs), and ball pitches as small as half a millimetre all contribute to unintended interactions between design elements.

Simultaneous Switching Noise

The first challenge is likely the so‑called Simultaneous Switching Noise (SSN) or Simultaneous Switching Output (SSO). Large amounts of high‑frequency data traffic can cause ringing and crosstalk on data lines, while power and ground planes may suffer from ground bounce and power supply noise that affect the entire board's performance.

To address ringing and crosstalk on high‑speed data lines, moving to differential signalling is a good first step. Because one line of a differential pair is the sink and the other provides the source current, inductive effects are fundamentally eliminated. When transmitting data over differential pairs, the current remains localised, which helps reduce "bounce" noise caused by induced currents in the return path. At RF frequencies up to hundreds of megahertz or even several gigahertz, signal theory tells us that maximum signal power is transferred when impedances are matched. When the transmission line is poorly matched, reflections occur, and only part of the signal travels from the transmitter to the receiver, while the rest bounces back and forth between the two ends. How well differential signals are implemented on the PCB plays a major role in impedance matching (and other aspects).

Differential Trace Design

Differential trace design is built on the principles of impedance‑controlled PCBs. Its model is somewhat like coaxial cable. On an impedance‑controlled PCB, the metal plane acts as the shield, the FR‑4 laminate serves as the insulator, and the signal trace is the conductor. The average dielectric constant of FR‑4 ranges from 4.2 to 4.5. Manufacturing tolerances can lead to over‑etching of copper traces, ultimately resulting in impedance errors. The most accurate way to calculate PCB trace impedance is to use field‑solving programs (typically 2D, sometimes 3D), which directly solve Maxwell's equations for the entire PCB batch using finite‑element methods. This software can analyse EMI effects based on trace spacing, line width, line thickness, and dielectric height.

A characteristic impedance of 100 Ω has become the industry standard for differential pairs. A 100 Ω differential pair can be created from two 50 Ω single‑ended lines of equal length. Because the two traces are close to each other, the field coupling between them reduces the differential‑mode impedance of the pair. To maintain 100 Ω impedance, the trace width must be reduced slightly. As a result, the common‑mode impedance of each line in a 100 Ω differential pair will be slightly higher than 50 Ω.

In theory, trace dimensions and the materials used determine impedance, but vias, connectors, and even component pads introduce impedance discontinuities in the signal path. It is often impossible to avoid these elements. Sometimes, to achieve a more reasonable layout and routing, it becomes necessary to increase the number of PCB layers or add features such as buried vias. Buried vias connect only certain layers of the PCB, but while they solve transmission‑line problems, they also increase manufacturing cost. Sometimes there is no choice. As signal speeds increase and board space shrinks, additional requirements such as buried vias become more common and must be factored into the overall cost of the PCB solution.

When using stripline routing, the signal is sandwiched between FR‑4 layers. In microstrip, one conductor is exposed to air. Because air has the lowest dielectric constant (Er = 1), the top layer is best suited for critical signals such as clocks or high‑frequency Serialiser‑Deserialiser (SERDES) signals. Microstrip routing should be coupled to the ground plane below, which reduces electromagnetic interference (EMI) by absorbing some of the electromagnetic field lines. In stripline, all the electromagnetic field lines are coupled to the reference planes above and below, greatly reducing EMI. If possible, broadside‑coupled stripline designs should be avoided. This structure is susceptible to differential noise coupled into the reference planes and also requires balanced PCB manufacturing, which is difficult to control. In general, it is easier to control the spacing between traces that are on the same layer.

Decoupling and Bypass Capacitors

Another important aspect of determining whether the PCB will perform as expected is controlled by adding decoupling and bypass capacitors. Decoupling capacitors help reduce the inductance between the power and ground planes and help control the impedance of signals and ICs across the PCB. Bypass capacitors help provide a clean power supply to the FPGA (providing a charge reservoir). The traditional rule has been to place decoupling capacitors wherever they are convenient for PCB routing, and the number of decoupling capacitors is determined by the number of FPGA power pins. However, the extremely high switching speeds of FPGAs have completely broken this convention.

In a typical FPGA board design, the capacitors closest to the power source provide frequency compensation for load current changes. Large decoupling capacitors are used to provide low‑frequency filtering and to prevent the supply voltage from drooping. Voltage droop occurs because the regulator's response lags when the circuit starts up. These large capacitors are usually electrolytic types with good low‑frequency response, covering a frequency range from DC to several hundred kHz.

Each FPGA output transition requires charging and discharging of the signal line, which demands energy. The function of bypass capacitors is to provide local energy storage over a wide frequency range. In addition, small capacitors with very low series inductance are required to deliver high‑speed current for high‑frequency transients. Larger, slower capacitors continue to supply current after the high‑frequency capacitors have depleted their stored energy.

The large number of current transients on the power bus adds complexity to FPGA design. These current transients are often associated with SSO/SSN. Capacitors with very low insertion inductance provide local high‑frequency energy and can be used to eliminate switching noise on the power bus. Such decoupling capacitors, which prevent high‑frequency current from entering the device power supply, must be placed very close to the FPGA (within 1 cm). Sometimes many small capacitors are connected in parallel to provide local energy storage for the device and to respond quickly to changing current demands.

In general, the routing of decoupling capacitors should be absolutely short, including the vertical distance through vias. Even a small increase adds inductance to the leads, thereby reducing the effectiveness of decoupling.

Other Techniques

As signal speeds increase, transmitting data reliably across the board becomes more difficult. Several other techniques can be used to further improve PCB performance.

The first and most obvious method is careful device placement. It is common knowledge to design the shortest and most direct paths for the most critical connections, but do not underestimate this point. If the simplest strategy can achieve the best results, why bother adjusting signals on the board?

Almost equally straightforward is to consider signal line widths. At data rates of 622 MHz or even higher, the skin effect becomes increasingly prominent. Over long distances, very thin traces on the PCB (e.g., 4 or 5 mils) will significantly attenuate the signal, behaving like a poorly designed low‑pass filter with attenuation increasing with frequency. The longer the backplane and the higher the frequency, the wider the signal traces should be. For backplane traces longer than 20 inches, trace widths should be 10 to 12 mils.

Typically, the most critical signals on the board are clocks. When clock traces are too long or poorly designed, they can amplify jitter and skew downstream, especially as speed increases. Routing clocks across multiple layers should be avoided, and vias on clock traces should be avoided because they add impedance variations and reflections. If an inner layer must be used for clock routing, the layers above and below should use ground planes to reduce delay. When using an FPGA PLL, noise on the power plane increases PLL jitter. If this is critical, a "power island" can be created for the PLL, using thicker etching in the metal planes to isolate the PLL's analogue and digital power supplies.

For signals operating above 2 Gbps, more expensive solutions must be considered. At these high frequencies, backplane thickness and via design have a significant impact on signal integrity. Backplanes with a thickness of no more than 0.200 inches perform better. For high‑speed signals on the PCB, the number of layers should be kept as small as possible to limit the number of vias. In thick boards, the vias connecting signal layers are longer, creating transmission‑line stubs in the signal path. Buried vias can solve this problem, but at a high manufacturing cost. Another option is to choose low‑loss dielectric materials such as Rogers 4350, GETEK, or ARLON. These materials can cost nearly twice as much as FR‑4, but sometimes they are the only choice.

There are also other FPGA design techniques that provide some flexibility in I/O placement. In critical high‑speed SERDES designs, SERDES I/O can be isolated by reserving (but not using) adjacent I/O pins. For example, a 3×3 or 5×5 BGA ball area can be reserved around the SERDES Rx and Tx, VCCRX#, and VCCTX# pins and ball locations. Alternatively, if possible, entire I/O banks adjacent to the SERDES can be reserved. If there are no I/O constraints in the design, these techniques can bring benefits without adding cost.

Finally, one of the best methods is to refer to the reference boards provided by the FPGA manufacturer. Most manufacturers offer source layout information for their reference boards, although special requests may be required due to proprietary information concerns. These boards typically include standard high‑speed I/O interfaces, as the FPGA manufacturers use them for characterisation and qualification of their devices. However, keep in mind that these boards are often designed for multiple purposes and may not exactly match a specific design requirement. Even so, they can serve as a starting point for creating a solution.

Summary

Of course, this article has only touched on some basic concepts. Any of the topics covered here could be the subject of an entire book. The key is to understand what the goals are before investing significant time and effort in PCB layout. Once the layout is complete, redesigning can consume substantial time and money, even for minor adjustments such as trace width changes. Do not rely on the PCB layout engineer to make decisions that will meet the actual requirements. The schematic designer must provide ongoing guidance, make informed choices, and take responsibility for the success of the solution.

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.