Problem with noise with addressable LEDs

[JonConesa]

Hi! I'm using Inolux_IN-PI554FCH LEDs with a custom PCB, controlled by an STM32H563VGT6.

When using the LEDs and setting a colour like 255,0,0 (RGB) the noise generated on the data track is negligible. However, when I set a colour that is not 255, the noise on the data track becomes very high, interfering with the operation of other components.

 

The noise measurements have been taken at the points marked in the following image.

 

 

 

Noise measure of colour 255,0,0 (RGB):

Measure at mark 1:

 

Measure at mark 2:

 

Noise measure of colour 255,45,13 (RGB):

Measure at mark 1:

 

Measure at mark 2:

 

 

As I mentioned, when producing a colour like 255,0,0 (RGB) or any variant such as 0,255,0 (RGB), the noise on the track is negligible. However, when producing a color that is not 255, such as the case of 255,45,13 (RGB), the noise on the track is too significant to ignore.

Currently, with this noise, I am experiencing issues with a capacitive sensor (AT42QT1010-TSHR) and its electrode (see image below). It triggers and stays in a high state continuously without touching the electrode (making it equivalent to touching the sensor), and it does not change state until you turn off the LEDs or put them in a state that does not produce much noise.

On the other hand, we have performed tuning that allows us to use the LEDs with the capacitive sensor fully functional. The problem is that we cannot fully leverage the capacitive sensor, forcing us to have to bring the finger too close.

The tuning involves modifying capacitors C2 (Cs) and C3 (Cx). To change the sensitivity of the touch switch, you need to modify capacitor Cs (it is recommended to place it between 2 to 50 nF) and for Cx (between 5 to 50 pF).

 

 

I'm considering changing the LED model, but I would prefer to keep them and reduce their noise.

[Peter BENSCH]

Have you buffered each LED with a capacitor of at least 100nF located directly between the respective VDD and VSS pins?

Have you decoupled the power supply from the touch switch and the digital rest, e.g. using an LC low-pass filter?

Regards
/Peter

[JonConesa]

Thank you for the response. So far, I have only placed 4 capacitors of 0.1 uF for the 16 LEDs.

Regarding the power supply for the touch switch, it only has one 0.1 uF capacitor.

When the LEDs are active, they generate a significant amount of noise in the power supply.

The noise with the LEDs on is at 4.47 kHz. This frequency coincides with that generated by the LEDs, as shown in the measurement image at mark 2, where the signal has a frequency of 4.3 kHz.

Without the LEDs:

Installing an LC filter at such a low frequency is not very optimal. What could be the solution?

 

 

 

 

[Andrew Neil]

Your scope has a USB port on the front for capturing screenshots - that would be a lot more useful that photographs!

 

 

[Peter BENSCH]

Well, I wouldn't call a highly sensitive, capacitive touch switch next to an interfering control signal optimal either. The solution is therefore decoupling, e.g. with an LC filter.

[JonConesa]

HI! 

What type of LC filter I must use? I have calculate a cutoff frequency of 3 kHz (the first harmonic is at 4.47 kHz) with a capacitor of 10 uF, and I should use an inductance of 281.45 uH, I think that LC filter for the touch sensor does not make sense. What do you recommend me?

Thanks and sorry for the inconvenience.