We have a product using an ESP32 MCU on a custom board. When performing transmission tests in 11b mode (1 Mbps, 20 dBm output), with our custom firmware, we cannot pass the wifi masks as per this figure:
When we do the test with our firmware, the side lobes are too high and do not pass the -50 dB required at 22 MHz separation.
To discard problems with our hardware, we tested our firmware on a nodemcu board with an ESP32 WROOM-32U module. There we have exactly the same problem with the lateral lobes. The funny thing is that when we use the Espressif RF-Tool with the nodemcu board, the results pass the mask, but not with our custom firmware. So there must be a configuration difference.
Some information about the test setup:
* The nodemcu board is directly wired to the 50 ohm spectrum analyzer input (using the u.FL connector in the WROOM-32U module).
* Our firmware is built using esp-idf v4.3.1 (commit 2e74914051d14ec2290fc751a8486fb51d73a31e).
* For this test, the firmware configures the radio in AP mode and uses the esp_wifi_80211_tx() function to send packets in 11b mode at 1 Mbps (20 dBm power).
This is the spectrum we get on the nodemcu board when using the RF Tool:
It is compliant with the WiFi mask. But this is the spectrum we get when transmitting with our firmware on exactly the same nodemcu board:
The tests in the photographs are for channel 7, but we get similar results for all the channels.
Why does the nodemcu board pass the test when using RF Tool but does not pass it when using our custom firmware? Could this be because of a misconfiguration? What could be causing these big lateral lobes? Is there any guidance on how to configure the WiFi radio on the ESP32 to guarantee a proper WiFi mask?