1
\$\begingroup\$

I am working on a device based on the STM32WB55xx line of microcontrollers.

In the latest revision of the PCB, I altered the antenna filter and chip antenna components to better match the device documentation. However, the device is now failing to connect to a phone over Bluetooth, while the previous version connects just fine. I'm looking for feedback on the schematic design, so we can try to determine whether our bug is in hardware or in firmware.

Here is the original Bluetooth antenna schematic:

Original schematic - BLE connects

And here is my new antenna schematic:

Updated schematic - BLE fails to connect

Here's the microcontroller datasheet guidance on the RF subsystem:

enter image description here

Finally, here is an excerpt from the ST AN5165 reference document, which specifies the antenna filter specific to the 100-ball BGA package type we're using:

enter image description here

Is there anything I'm missing at the schematic level? I think I have implemented ST's guidance correctly. If the schematic looks okay, I intend to post a follow-up question focused on layout.

On page 40 of AN5165, the document reads:

"As a conclusion, the IPD reference MLPF-WB55-01E3 can replace the RF output network of the STM32WB for the QFN packages (an antenna filter is still needed) for a 2-layer PCB."

Does this imply that the output filter from the microcontroller and the input filter to the antenna are two separate filters? I assumed that there should be one filter between the microcontroller and the chip antenna, but now that it's not working, I'm revisiting my assumptions. I'm noticing that the original schematic has the Murata bandpass filter chip, and then an additional pi filter before the antenna.

Do I need to add an additional filter stage between the ST IPD chip and the antenna? Or does my schematic look okay as is?

\$\endgroup\$
5
  • \$\begingroup\$ Your schematic looks correct. The original design had a separate filter (perhaps for FCC compliance) and a matching network. The updated design with an IPD incorporates both functions in one component. Show us your layout with feedlines. Is the antenna on the PCB or connected with coax? \$\endgroup\$
    – Mark Leavitt
    Commented Mar 18, 2021 at 11:41
  • \$\begingroup\$ It's an SMT chip antenna mounted on the PCB. No coax, but I did route the RF connections with a 50 ohm controlled impedance. I was planning on posting the layout as a separate question, but if it's more appropriate to combine the two, I can add layout information to this question. \$\endgroup\$
    – Chris Fernandez
    Commented Mar 18, 2021 at 14:19
  • \$\begingroup\$ Can you do a more diagnostic comparison beyond "failed to connect to a phone over Bluetooth"? You could install a BLE utility (like Nordic nRF Connect) on the phone to measure RSSI of advertisements, comparing your two designs at the same distance, etc. \$\endgroup\$
    – Mark Leavitt
    Commented Mar 18, 2021 at 15:22
  • \$\begingroup\$ Yep, we have done that. The device is not advertising. No BLE signal detected by any of the BLE utility apps we've tried, regardless of distance. \$\endgroup\$
    – Chris Fernandez
    Commented Mar 18, 2021 at 17:24
  • 1
    \$\begingroup\$ in that case I'd look elsewhere. Other than a dead short, it's hard to screw up the antenna so badly that no signal is detected within a foot. \$\endgroup\$
    – Mark Leavitt
    Commented Mar 18, 2021 at 21:27

1 Answer 1

Reset to default
1
\$\begingroup\$

This question is approaching 2 years stale, but I received a message from someone to explain what I did to resolve this issue, so I am sharing that now.

We were able to track the non-operating Bluetooth antenna to a particular Bluetooth event labeled HCI_HARDWARE_ERROR_EVENT.

We reached out to ST for information on what causes this event. We were told that the cause is usually instability in the HSE which can be caused by ground currents or poor layout.

So, we then opened a new ticket with ST, asking them to review our layout. I shared my Gerber files, and spent a few days updating it to attempt to match their recommendations. Note that we had a challenging layout constraint of trying to use the BGA package with just 4 PCB layers. Typically, 6 layers are used for a BGA with a pitch this tight.

In the end, ST recommended that we switch to the STM32WB5MMG package instead, due to the difficulty of routing this design with 4 layers. The STM32WB5MMG package is significantly larger, so that option would not work for us. We ended up switching away from ST to a competitor's microcontroller family instead.

Here is the link to our ticket with ST (ST community account required to view): https://community.st.com/s/case/5003W000008FEAwQAO/hse-xtal-layout

\$\endgroup\$

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.