Impedence matching required for very short PCB trace?

Question

I have a custom PCB using 50 ohm impedance antennas for GPS (1.575 GHz), GSM (2.4 GHz), and LoRa (868 MHz). Currently using an impedance matched board with matching track widths of 6 mil. One track goes through a capacitor (matching manufacturer's reference design). My U.fl connectors are VERY close to the pins on the RF chips, or in the case of the one with the capacitor, which is adjacent to the RF pin and U.fl connector on the other side. By "very close", I mean between 2 and 3 mm.

Since the RF pins are much wider than the impedance matched track anyway, and the capacitor pads are also wider, is there any actual benefit in using impedance matched tracks for these short lengths? Indeed, is it better to just match the width of the pins on the RF devices, all the way to the U.fl connector? I'm using a 4 layer design with layer 2 as the dedicated ground plane, 3 as power, and 1+4 as signals with no ground fills. All RF signals are on layer 1. I'm keeping all "noisy" signals well away from the RF side. I have tried to find some clarity on this but maybe missed something. My design seems to perform well for all 3 RF devices, but I'm always hoping to improve things if possible.

Answer 1

That is a good question indeed, but the answer really depends on how much power you are willing to lose to the transmission line (TL) effects.

Different people use different rules of thumbs for the length of the line at which they start to worry about TL effects. I personally use 0.1\$\lambda\$ (10%) for signals and 0.01\$\lambda\$ (1%) for power. In a nutshell, for signals you can tolerate more voltage reflections as your receiver probably has a wide margin as to what it would interpret as a '0' or as a '1', thus the 10% benchmark -- think RS-485 for example . For power, however, as the TL exceeds the 1% benchmark, your 50 Ohm antenna doesn't look like 50 Ohms anymore and thus you have an impedance mismatch. See this question I posted some time ago and the answers for more background information on this.

In this situation, you care about power delivered to an antenna so I would still keep the impedance-controlled lines since at your worst case scenario (at 2.4GHz), the TL exceeds 0.01\$\lambda=1.2\text{mm}\$. By keeping your impedance at 50 Ohms, then the TL length becomes irrelevant, at least in theory.

Where did I get the 1% rule for power?

References: Fundamentals of Applied Electromagnetics 6/E, Ulaby|Michielssen|Ravaioli