ESP32: Do I terminate an unused RF pin (LNA_IN)?

[caiser01]

I'm designing a gadget for work and I'm looking to use a bare ESP32 chip (probably the ESP32-D0WD-V3). The primary reason to use the ESP32 is so we can reuse part of the code we wrote for another ESP32-based gadget that has already been thoroughly tested and field-proven. The new gadget does not need (and probably should not have) any wireless capability, so the WiFi and Bluetooth features of the chip would be disabled in the firmware.

My question is: If I'm just going to drop the bare ESP32 chip on a PCB, should I terminate the RF input/output pin (LNA_IN), and if so, how? Resistor to ground? Something else? I want to make sure the chip won't be fried by reflected RF power in the event a software developer accidentally flashes firmware that turns on the Wi-Fi or Bluetooth.

Espressif has some basic guidelines on RF matching under 2.5 in this document (https://www.espressif.com/sites/default/files/documentation/esp32_hardware_design_guidelines_en.pdf) but I couldn't find any guidelines for what to do if you're not using WiFi/BT.

[boB]

I didn't read the Espressif link yet but I would probably add a 49.9 Ohm  (~50 Ohms)  load to the unused RF port for termination.

The power level is so low that is might not matter ?  Or the ESP32  already has a self protection mode ?
I guess the ESP32 can transmit almost 1 watt or less.   Probably need a 1 watt surface mount resistor ?
How about 10 499.9 Ohm in parallel ?

I see that there is only one power stage for both Bluetooth and WiFi.  Does this newer unit have a second port ?

boB

[woofy]

I would connect a 36.5 ohm resistor for 35+j10 or 31.6 ohm for 30+j10 from the tx pin close to the chip, to ground. Use a small package 0201 or 0402.

[boB]

Yeah, either value should work.

boB

[caiser01]

woofy and boB, thanks!

I was thinking a simple resistor would do it but wanted a sanity check from the crowd here since I know there are more knowledgeable people than I around here.

[hans]

Or 33R, it would be close enough. S11 would be in the order of -15dB for both package types.

Do note that the ESP32 can transmit +20dBm or so in WiFi mode (100mW). I don't think its possible to accidentally have a high duty on the RF port, but in case it is, it would require a 0603 resistor.

[caiser01]

Or 33R, it would be close enough. S11 would be in the order of -15dB for both package types.

Do note that the ESP32 can transmit +20dBm or so in WiFi mode (100mW). I don't think its possible to accidentally have a high duty on the RF port, but in case it is, it would require a 0603 resistor.

Good info 

I don't think there should be an issue squeezing an 0603 resistor in for this purpose.

Thanks!

[nctnico]

I'd leave it open. No need to potentially dump a lot of power into a load. You'd need to beef up the power supplies as well. Actually, I'd investigate whether you can leave the whole RF section unpowered when it is not used.

[hans]

Sorry to put it, but IMO thats bad advice.

When the ESP32 is transmitting (in this case accidentally), it outputs its full voltage rail to reach maximum output power.
3.3Vpp=2.34Vrms (for its fundamental), in 30R, is 16.5dBm.
The ESP32 claims to transmit 19.5dBm. My guess is because the total power is measured for a 2.4GHz square wave (with its harmonics). I'm not seeing a differential output or switched-cap supply rail.
(Note that 30R, or lower, is needed to reach these transmit powers at 3.3V. You would need 3.9Vrms into 50Ohms)

Anyhow, if that fundamental gets reflected back it would reach 4.6Vrms (above max voltage rating), which can fry the chip.

Just a 33R ohm resistor is also not a "perfect" match, in particular of the +10j inductive part. But its good enough as with S11 of -10dB, only 10% power is reflected.. so -15dB should be only a couple of percent.

[nctnico]

And where is the transmission line that causes the reflection? The wavelength at 2,4GHz is 12cm. Say 8.5cm on a board. You'd need to connect a trace which is way over 1/10th of the wavelength (say over 10mm in length) in order to consider that trace a transmission line (or any form of a resonator when unterminated).

[woofy]

Any change of impedance will cause a reflection, like the open circuit at the chip pin.

[nctnico]

Any change of impedance will cause a reflection, like the open circuit at the chip pin.

Again, only IF there is a transmission line connected to the device. Try to do an experiment like a TDR setup with various cable lengths and various rise times / sinusoidal pulse frequencies. You'll see you won't get reflections below a certain cable length depending on the frequency content of the pulses you are using.