0
\$\begingroup\$

I'm just getting into RF design since I inherited a board that uses an STM32WB55C for BLE. For now I want to start getting more familiar with the RF concepts since I don't have a lot of experience.

As I was going through ST's RF Design guide here:

How to develop RF HW with the STM32WB

And also found this post on their community forums:

Characteristic impedance of the RFO pin in the WB55

Alright, so let me tell you what I'm struggling with:

I now know that the characteristic impedance of the RF pin in the QFN48 is 42 + j19, hence why they are using either an IPD or a matching network filter.

In that guide you can see the following values for the matching network:

[Pi Matching Network](https://i.imgur.com/HlBI8Sl.png)

But I ran those numbers via a S-Paramter Calculator and I end up with a characteristic impedance of 62 + j22 instead of the desired 50 Ohm:

[See results here.](https://i.imgur.com/Y1Sgolp.png)

I then used the following calculator to get the values for a matching network for a 50Ohm load

This one here.

And it gives me the following results:

[Pi-pad results](https://i.imgur.com/Qp3aETs.png)

And they don't match the STM guide. If I run the numbers from pi-pad on my Smith calculator I get a proper 50Ohm impedance.

I tried a different tool: SimSmith on MacOS and I noticed something. If I put the target on the left side, and plug in the values from STM32's AN then I do see a value closer to the reported 42 + j19:

[See here](https://i.imgur.com/KTD9EFB.png)

You can see here right under C2 an impedance of 42.46 + j13.28

Now I am thinking I may be doing something wrong, or I have misunderstood how a matching network actually works.

Can anybody help me understand what's going on with the Application Note, and what I am doing wrong?

Edit: I just realized that the order of the components in the pi filter is not the same in that last calculation I did, so it may just be a coincidence that it calculated a number close to 42 + j19.

\$\endgroup\$
2
  • \$\begingroup\$ I just realized that in my last calculation the order of the components doesn't match how they are setup in the STM32 schematic, so I think it may have been a coincidence. \$\endgroup\$ Commented Jun 22 at 22:04
  • \$\begingroup\$ I just realized something: In SimSmith if i set the load impedance to be the WB55 pin impedance and load the values I got from pi-pad I get a source impedance of 50 Ohm. And if I set the load impedance to be 50 Ohm, re-order the capacitors in the pi-filter I get a source impedance that is the conjugate of the WB55 pin impedance (42 - j19). \$\endgroup\$ Commented Jun 22 at 22:23

1 Answer 1

1
\$\begingroup\$

Technically a shunt capacity of 0.6pF (or an equivalent, short 50 Ohm transmission line stub) would already be enough to make the output impedance 50 Ohm:

enter image description here

The component values for the matching network are only valid for the PCB layout shown in the design guide. ST probably assumes an additional 0.4pF parasitic capacity to ground at the input and output of the matching network:

enter image description here

Another simple matching solution would be to load the output with a short 50 Ohm transmission line in series with a 2.7pF capacitor:

enter image description here

\$\endgroup\$
2
  • \$\begingroup\$ In your second scenario, may I ask how you added the 0.4pF parasitic capacitance to the simulator? Are you adding the capacitance to the 2 caps that are part of the matching network? \$\endgroup\$ Commented Jun 23 at 2:30
  • 1
    \$\begingroup\$ Yes, both capacitors in the PI-matching network have a parasitic C in parallel originating from the layout. Additional parasitics could be present. The schematic does not reflect that. If you want to use the matching network shown in the design guide you also have to replicate the entire layout 1:1. \$\endgroup\$
    – Raonoke
    Commented Jun 26 at 12:01

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