Difficulty Tuning nRF24L01+ Matching Network

[Andrew S.]

Hi there,

After prototyping a device with modules and on a breadboard, I transitioned to a custom PCB with nRF24L01+, matching network from the spec sheet, and 2.45 GHz 50-ohm chip antenna. Other components on the board include an ATMega328P, 3.3V regulator, LiPo charger, USB-micro-B connector, and a 400mAh LiPo battery glued to the back of the board. The device is to be part of a battery-powered remote sensor network.

Unfortunately, I have not been able achieve good performance with this system. Using a spectrum analyzer with cable soldered to the board, at best I can achieve -12 dBm power at the 50 ohm point of the matching network in TX mode at 2.4GHz (channel 0). At channel 127, output power is significantly worse (-24 dBm). A single E12 step in either direction for inductor L1 (8.2 nH in the reference network) reduces output power to -16 dBm.

Admittedly, my layout does not match the reference design and I am using 0805 components, including wirewound inductors. Is it possible to achieve good performance in this configuration? Should I change the value of components other than L1 to try to boost output power? Should I re-design the board to use 0603 or 0402 components and more closely match the reference design before doing anything else? Will adding a Pi network at the 50-ohm point help?

Thanks so much! Andrew

[Øyvind Karlsen]

Hi Andrew,

Adding a Pi network should help, and that is probably the first thing you should do to improve performance. It is also important that you tune your antenna for the use-case scenario, e.g. for a wearable application the ground plane will be different from a free-space situation, which causes different impedance for the different cases. Also make sure that your placement of the chip antenna is in accordance with the datasheet.

Next, using 0805 components will significantly reduce the performance as these components have much larger parasitic effects than the 0603 and 0402 variants. My advice would be to change to 0402 and exactly follow the reference design.

From your results with -12dBm at channel 0 and -24 dBm at channel 127 it seems like your matching network does not target the center of the band. If these results persist after adding a Pi network and changing your components and layout, you might want to re-calibrate your matching network towards the center frequency of the band, depending on which channels you plan to use.

There are also some white papers and blog posts I think you might find useful, these are: General PCB design guidelines for nRF51(different chip, but the theory still applies), Tuning the nRF24xx matching network and Antenna tuning.

If you can upload your layout that would also be helpful in addressing your problem.

[Andrew S.]

Hi Oyvind,

Thank you so much for your helpful reply. I will read the literature you suggested.

I'm also attaching EagleCAD files of the revision of the board which I tested at -12 dBm (Rev. 4) and a revision I've ordered to duplicate the reference design of the antenna matching network (Rev. 6). If anything jumps out at you, please do let me know.

Thanks so much!

All the best, Andrew

flowerfriend_rev6-nosma.brd

flowerfriend_rev6-nosma.sch

flowerfriend_rev4.brd

flowerfriend_rev4.sch

[Øyvind Karlsen]

Hi again Andrew,

I have looked at the RF part of your layout, it looks good. I also have some suggestions for improvement. Add a keep-out area where I have marked it in this picture, this will reduce stray effects to ground.

You must also add a Pi-network in the area just before your chip antenna.

[Andrew S.]

Great! Thank you for taking a look. If I add a Pi network to the layout as in Figure 12 of the antenna tuning guide (two components in parallel on either side of a third component), won't the center component's length affect the path length from the matching network to the antenna? Is that path length critical to obtaining good performance?

Also, to be sure I understand the antenna tuning procedure, I should do this:

1. Assemble a board with all components, including the chip antenna, excluding the nRF24L01+, matching network, and pi network.

2. Short circuit the pads of the series components leading from the nRF24L01+ to the antenna.

3. Calibrate VNA with SMA cable length being used.

4. Solder center conductor of SMA cable to the pad leading from the nRF24L01+ ANT2 pin, solder mesh to the ground plane.

5. In a single port measurement, use VNA to find power/phase/impedance reflected back from antenna from 2 - 3 GHz and especially at 2.4, 2.45, and 2.5 GHz.

6. Calculate pi network components to tune antenna to 50 ohm, solder to board, and repeat measurements and change components until desired impedance is reached.

I was also wondering--the chip antenna in this design, according to the datasheet, claims that it does not need a matching network and is already tuned to 50 ohm impedance. Is the idea that the nRF24L01+ matching network provided does not actually deliver 50 ohm impedance in real-world applications and the pi network is necessary to do further matching? Otherwise, isn't additional tuning unnecessary?

Thank you!

All the best, Andrew

[Andrew S.]

Oyvind,

Would you be able to check out this PCB design with a Pi network added before I order it? I'm afraid the constraints of this project are such that a large LED must be in the center of the board and it can not be larger than a 4 cm diameter circle, which seems to force me to rotate the nRF24L01+ and matching network 90 degrees in order to fit the additional Pi network.

Eagle Files:

www.dropbox.com/.../AACL9CdSofj3B_dMZUQ6Uhc-a

Thanks so much!

All the best, Andrew