# RF Antenna Matching circuit and PCB design for 868 MHz SigFox device

I'm designing a Sigfox sensor device based on WISOL/SFM10R1 a SigFox verfied module by WiSOL and Atmega328.

I need to understand the parameters for 868 MHz RF Antenna Matching circuit and PCB trace width needed for the design. Following are few design layout and sketches.

I'm attaching a 50 Ω external antenna to uFL connector on board. Currently trace width of track from WiSOL module to uFL connector is 0.6096 mm

Schematic:

Component value here are not correct and thats what I want to know the way to calculate it. I'm not a RF expert its difficult subject for me.

I found one online calculator which I tried to use as a simple guide at https://www.eeweb.com/tools/microstrip-impedance for calculating trace-width. But as per my current parameters I'm getting trace-width of approx 3 mm which does not looks practical.

Parameters:

• Trace Thickness: 17umm (I hope this means copper thickness)
• Substrate Height: 1.6mm (This is probably PCB thickness.)
• Trace Width: ?
• Substrate Dielectric: 4 (I'm using FR4 PCB board so i guess this is ok)

Freq: 868 MHz

I have few questions here:

1. Which calculator should I go for - Microstrip, Symmetric stripline or embedded strip at the link I mentioned above, as im having double sided PCB and antenna trace is on top layer!
2. What will be the way to calculate value of component to match the antenna of 50 Ω
3. Does length of trace also matters?
4. Why I'm getting 3mm trace width? Where I am going wrong?
5. Whats the significance of those multiple vias added around antenna in most designs.
6. Any other suggestion to improve this antenna circuit design which should considered from certification point of view.
• so, small things: 1. Units! It's "MHz", not "Mhz", please. Commented Mar 2, 2019 at 10:36
• 2. these are six questions at once, maybe start with one question that you researched well – for example, if we focus on the first one, it shows you haven't done a bit of research on what "symmetric stripline" actually is, or else you'd already have ruled that out. Same for embedded strip line: If you have a many-layered board, sure, go for that. But you'd probably know if your board has more than 4 layers. Minutes of using google, less time than spent on asking here! Commented Mar 2, 2019 at 10:38
• Re: 4. You're probably not going wrong. On thick FR4, multiple millimeters of trace width for microwave does sound realistic if you're doing microstrips. Commented Mar 2, 2019 at 10:39
• Question 2. is too broad, also, multiple application notes and a large wealth of things you can find online. Commented Mar 2, 2019 at 10:39
• Question 4.: that's probably even in the antenna's data sheet, or try searching for "vias ground plane antenna" on this site. Commented Mar 2, 2019 at 10:40

Your Pi matching network will look like this ( or similar values, that provide a high-Q??? resonance at 868 MHZ)

simulate this circuit – Schematic created using CircuitLab

• Thank you @analogsystemsrf for your answer. Can you please explain about how the values are calculated. Commented Mar 2, 2019 at 19:16
• @ Rajendra If you like an answer, its polite to upvote (or even "accept"; but wait another day on accepting). Consider that 1pFarad at 1GHz has -j160 ohms reactance. Since I don't know your antenna impedance, I simply used 1PF. Similarly, 1nanoHenry at 1GHz is +j6.3 ohms, 10nH is +j63 ohms. Clearly these are not resonant at 1GHz; how about 868? Nope, but you need to adjust the values. And "vias" in series with the two capacitors? And the inductance of that narrow-strip of Ground foil between the two capacitor-Ground nodes? Even PCB thermal-relief will affect the PI-network resonance. Commented Mar 3, 2019 at 12:45
• Thank you @analogsystemsrf for the answer but I already tried upvoting your answer but i don't have much reputation to do that. I'll try to understand your comment and get back here if needed. Commented Mar 3, 2019 at 13:38

Im posting the answer as suggested by WiSOL technical team. I'll update to it after fabrication and RF testing. Component values are as follows:

L1 = 47nf for ESD , does not change RF maching

C4 = 100pf it offers 0ohm impedance.

C9=DNI

@Marcus Müller Imp part was PCB track which needed will be Coplanar Wave and not Microstrip.

So final calculation came as :