Critical guidelines for RF and microwave PCB Design
A few decades ago, there wasn’t much demand for RF and microwave circuits. They were difficult to design into the architectures of the time, and so costly that only mil/aero projects could afford them. But today RF circuitry is crammed into a large variety of commercial products. Most of these are handheld wireless devices for medical, industrial, and communications applications, plus applications in a variety of fields are migrating from desktop models to become portable communications units. Not only is RF becoming more ubiquitous, but microwave circuitry as well, both capturing very high frequencies (VHF) and ultra-high frequencies (UHF).
Printed circuit boards (PCBs) now encompass much more than pure digital or mixed-signal technologies, and the PCB layout designer faces many more challenges when designing sub-assemblies with high frequency RF and microwave.
The RF frequency range is typically from 500 MHz to 2 GHz, and designs above 100 MHz are considered RF. The microwave frequency range is anything above 2 GHz. There’s a considerable difference between RF and microwave circuits versus typical digital and analog circuits. In essence, RF signals are very high frequency analog signals. Therefore, unlike digital, at any point in time an RF signal can be at any voltage and current level between minimum and maximum limits.
Standard analog signals are assumed to be between DC and a few hundred megahertz. But RF and microwave signals are one frequency or a band of frequencies on a very high frequency carrier (Figure 1). Unlike digital signals associated with one voltage or one current, RF and microwave signals operate on a frequency.
Figure 1: RF and microwave signals are one frequency or a band of frequencies on a very high frequency carrier.
RF and microwave circuits are designed to pass signals within a certain band. They use band pass filters to transmit signals in a so-called band of interest. The signal within a range of frequency passes through this band range, and the rest of the frequencies of the signal are filtered. A single band can be very narrow or very wide and carried upon a very high frequency carrier wave.
Issues with PCB design and RF/microwave
For the most part, PCB layouts that include RF and/or microwave circuits are difficult to design. However, regardless of their difficulty, the rule of thumb is to start with the basics and follow the laws of physics. Up front, the PCB designer needs to understand that microwave signals are highly sensitive to noise. The possibility of incurring ringing and reflections must be treated with great care.
When working with very high-speed digital signals in the gigahertz range or 10Gb per second, for example, the PCB designer has to follow certain guidelines and rules. When RF and microwave enter the layout, the PCB designer must have the same mindset, yet multiply that mindset many times simply because RF is far more sensitive than very high-speed digital signals.
Second, impedance matching is extremely critical for RF. Digital signals -- even if they are very high-speed -- have a certain tolerance. But for RF and microwave, the higher the frequency, the smaller the tolerance becomes. For example, the PCB designer must keep it at 50 ohms -- 50 ohms out from the driver, 50 ohms during transmission, and 50 ohms into the receiver.
Third, the return loss must be minimized. This loss is caused by signal reflection, or ringing. The return is the path taken by the return current (Figure 2). For example, take a single-ended signal going from the driver to the receiver. Obviously, there has to be a return signal from the receiver to the driver. If it’s a single-ended signal, then the return typically takes the path of least impedance.
At very high microwave frequencies, the return signal takes the path of least inductance. Ground planes underneath the signals are good at providing this path. Therefore, there should be no discontinuities in the plane underneath the signal all the way from the driver to the receiver. However, If there is a cut in the ground or the ground plane doesn’t exist underneath that trace, the signal will still somehow find its way to the driver. It will go through power planes, through a PCB’s multi-layers or through some other route - it will definitely find a return path. But such a path won’t be ideal, and thus will cause reflection and ringing since it will no longer be an impedance controlled signal (Figure 3).