3
\$\begingroup\$

I've been dreaming up a small gadget featuring a NRF51822 MCU as a centerpiece and I'm a total noob at RF-voodoo. I want to use a chip antenna. My question is this: Do i need this matching network From the Datasheet under reference design

From the MCU/Transceiver or this one From the Datasheet under reference design

From the chip-antenna datasheet or both?

I found some talk online of mixing the two together somehow but that was with a different configuration on the output of the MCU. Can someone help me or point me in the right direction?

Link to the MCU datasheet

Link to antenna datasheet

\$\endgroup\$
1
  • \$\begingroup\$ The matching network is typically designed once the return loss can be measured on an actual PCB sample using fancy equipment (a VNA). Unfortunately "real life" happens and your transmission line won't be exactly as designed, due to issues such as manufacturing tolerances. In other words, these are tuned to their specific applications. As a first pass, you can leave the impedances to ground open and the series impedance short. Atmel AT16802 shows a worked example of tuning a microstrip: ww1.microchip.com/downloads/en/AppNotes/… \$\endgroup\$
    – BB ON
    Aug 10, 2018 at 19:34

1 Answer 1

2
\$\begingroup\$

Technically, yes, you need both. The MCU one goes next to the MCU, and the antenna one goes next to the antenna.

The network mentioned in the MCU datasheet is always required, because it converts the double ended output from the MCU into a single ended output for your antenna, and basically gives the MCU a 50 Ohm output.

The matching network for the antenna is so that the antenna can be precisely matched with the feedline and deliver the maximum amount of power into the air. Since the exact impedance of the antenna and the feedline will be affected by many things, like nearby components and casing etc, this is typically not set in stone during the design phase. As BB ON mentions, what usually happens is that the circuit designer puts the space for a pi matching network on the board, and then the board is assembled and put in the casing. Then a signal is put through the antenna, and the reflection is measured. Depending on the exact characteristics of the reflection, the pi network at the antenna is modified until the reflection is as small as possible.

If you're not going into production and the range isn't critical, then you can probably get away with not doing that final step and just putting the component values from the reference design. Your efficiency and range will suffer, but the device will probably still work.

If your feedline is very short, you might be able to skip the antenna one and just use the MCU one, as explained here. Note that you still need the full MCU one that you talk about in your question.

TL;DR yes you need both

\$\endgroup\$
1
  • \$\begingroup\$ Ditto. Very good answer. I always add the extra pi network, with whatever parasitics on the board and even though your antenna may claim to be 50 ohms, I've found you always get better results to use a VNA and fine tune. \$\endgroup\$
    – Leroy105
    Aug 10, 2018 at 19:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.