ESP32-S3 - Radiated Emission Problem Driving a QSPI LCS

[mattiadsael]

Hi everybody,

in the past months I've been working on a project that involves an ESP32-S3 (in particular, model ESP32-S3FH4R2 with integrated PSRAM and FLASH) and a QSPI TFT color display (resolution 400x400, controller ST77903).
In this projects I'm using ESP32 just for driving this display are reading an I2C sensor, I don't use any WiFi or Bluetooh features.
The QSPI display is driven with a CLOCK frequency of 20MHz. The display works really good and I'm really satisfied with the results.
Unfortunately I'm facing a problem with radiated emissions and the results seems very strange: I would have expected to find odd harmonics up to a certain frequency (for example 220 MHz, the 11th harmonic oh 20 MHz), but what I face are a lot of harmonics in the whole range of 1GHz. You can see the graph attached (purple line is the graph to consider, yellow line is the measure without device - I use a TEM antenna which is not fully shielded).

I've tried different software workaround, for example:

- decreasing to minimum value the pin drive strength;
- setting all the unused pins to INPUT PULLED UP
- disabling wifi-bluetooth components

results improved but not as desired.

In my previous project I achieved a lot of help by using the SPREAD SPECTRUM CLOCK. Most of microcontrollers have this feature that enable to have a varying frequency clock (+/- 1), avoiding frequency peaks in the EMI RAD results.

Has anyone ever tried to achieve SPREAD SPECTRUM CLOCK on ESP32-S3?

Thank you in advance for your cooperation.
Best Regards
Mattia