how to can I reduce the noise radiated from STM32? Does that noise exists to all ST's controllers?

[panatronic]

I am using STM32F030K6T6. My project is a 433MHz one way data link. The RF receiver blocked from the radiated noise of STM32. So, the distance is reduced between the link. In order to make it work, RF receiver and STM32 have to be at least 5 inch far away but unfortunately this can not be done in real PCB as it must be compact (RF+controller). I need something to reduce the noise radiated from STM32 otherwise I have to replace the ST controller.

[T J]

there are several ways the noise is getting into the RF side.

Ground wire is the worst. did you use a ferrite or two ? inductor filters ? maybe you should ?

you could try to shield the RF board with a Faraday cage. but the ground is still connected.

it is the Switching Power Supply noise that is upsetting the RF side.

Just to prove the suggestion:

Remove all the switchers,

Wire in a Lithium battery at 3.7V, linear regulator down to 3.3V,

you will need a regulator with low dropout, like this one:

https://www.digikey.com/product-detail/en/diodes-incorporated/AP2114H-3.3TRG1/AP2114H-3.3TRG1DICT-ND/4505142

it will get warm to hot depending on the current draw. you may need some PCB room for heat dissipation.

[panatronic]

Thanks T J for your reply.

The pcb is double sided with ground on bottom and top and as many as possible vias between them.

The power comes from a 9V battery with linear low drop regulator.

The RF-module is shielded.

The capacitors on STM32 are the recommended by datasheet + 1nF +220pF on power pins of STM32.

I am not using any inductor.

[T J]

ok, good start..

now you are still seeing the ground noise,

the source is the STM32, I guess too.

the Processor Noise is likely at HCLK or /2 or /3

did you notice that you are almost exactly 9x the processor frequency ?

433/48 = 9.0208

suggest use a crystal that is not close to an integer division of 433MHz...

did you use any ferrites ? these block the current spikes along the ground wire.

I use two ferrites to pass power.. 3.3 and Gnd or 5V and Gnd

Then you need low ESR caps on the rails either side of the ferrites, this will only give you 20-50% improvement in noise.

[Jeroen3]

Can you show the PCB?

[AvaTar]

A MCU is high-speed switched logic, so RF noise is not really avoidable.

I see the following options:

Change the core clock frequency, to minimize EMI noise at 433MHz.

Or, shield the MCU section, i.e. a faraday cage.

Or, switch to a MCU with significantly less power consumption, and thus less EMI.

F***ing forum editor, auto-makes bullet/numbered list with a "-" or number as first char, cannot undo it ... >:(

[panatronic]

I changed the core clock frequency from 12MHz to 24MHz, 16MHz, 8Mhz, 4MHz, 2MHz. The noise still exists.

I tried many types of shields. I think that the noise is radiated from all the pins of STM32.

I quit. That kind of processors can not deal with RF receivers, especially at 433MHz.

Thanks you all for your support and your replies.

[T J]

it is ground noise, and yes, all those frequencies are close to integer multiples of 433.

Do you have a crystal on board ?

can you change the crystal to say, 4.192Mhz to pick a ramdom crystal frequency...

Are you using the cube ?

You can shut down all unused clocks inside the cube...

I add prescallers to every clock so that only the lower frequencies are running through the peripherals.

IIC for example, it does not need to be run from 48MHz, drop it down as low as possible.

same with SPI, UARTS, etc etc.

[LMI2]

It is quite common to have a box of metal around either CPU or RF electronics. I think Digikey and others have some in stock and they easy to get in custom sizes too. How else could a smart phone work?

[panatronic]

Yes , I am using the cube. I don't have any crystal on board because all the pins of STM32 are in use. Although, I have tried some crystals (6.25, 12.00, 13.423, 22.00 ) with no success. Below of 8MHz the software has significant latency.

The last shield I used was "https://www.digikey.gr/product-detail/en/leader-tech-inc/SMS-401C/1798-1206-ND/7203474" with no success.

The only way to get rid of the noise is to place the RF module at least 5 inch away where it start to work fine.

Today, I tried 4 types of RF modules with no success.

When the I2C is active the noise is much more and the RF module can not receive anything. So, I activate the I2C with a 3sec delay after any received portion of data, in order to overcome this issue.

Only, timer 3 is active on receiving mode. All others are disabled.

I conclude that the noise is radiated from all the pins of STM32.