STM32H753BIT6 controller Radiation Emission issue

[Tejas_P_R]

Hi All,

We have design schematic using STM32H753BIT6 microcontroller and as of now we have cleared all EMI/EMC tests except Radiation emission. We could able to see the harmonics of 60MHz crossing the margin.

We have System clock running at 480MHz, LCD Clock at 60MHz, SDRAM clock at 100MHz and USB clock at 60MHz.

We did some of the experiments in finding out the source of the emission using near field probes and spectrum analyzer and got the source point.

USB clock output is connected to microcontroller as input with series termination resistor of 33R. when we disconnect this path to microcontroller, we are not able to see any 60MHz harmonics.

The trials we did to reduce radiations is as follows.

1. We changed series termination resistor 33R to 50R to change rise time and fall time of clock, but this didn't help in reducing the radiations.

2. We replaced series termination resistor with Ferrite bead still there is no improvement.

3. We are not getting harmonic peaks at supply pins of the microcontroller(we monitored using 50R coaxial cable with spectrum analyzer), but still we added bulk capacitors for the supply pins of Microcontroller and USB controller, but still the 60MHz radiation levels are same.

4. We added few pF capacitor near series termination resistor to reduce the radiations on the clock signal, but this also didn't help in reducing the radiations.

The clock signal is routed very well and having proper ground and power planes. there is no split planes observed.

I gone through so many forum queries related to radiated emission but still there is no proper solution for our problem.

Could anyone faced similar issues, please suggest me to resolve it.

Thanks and Regards

Tejas

[AScha.3]

Hi,

why you have USB at 60Mhz ? (afaik it should be 48MHz , for full speed USB) or is it hi-speed ?

+

For test : disable the LCD and its signals (I had most emission just from the TFT connected !)

[Tejas_P_R]

Hi, 

We are using USB controller with 24MHz crystal and is generating 60MHz clock output (For High speed) is connected to microcontroller as input.

We disabled the LCD clock still we observed the 60MHz and its harmonics peaks, after removing series resistor placed in between USB clock out pin of USB controller and Microcontroller we couldn't see any peaks., So by this experiment we have confirmed that the USB clock output connecting to controller is causing the radiation emissions.

[LCE]

How does the clock look on the scope?

Have you tried some termination of that clock line on both ends?

Another serial resistor, or some capacitor (4p7) or AC termination (e.g. 47p / 50R to GND) at the sink?

Maybe it's not only the clock line, but the combined power of clock + strobes + data lines?

[Tejas_P_R]

Hi LCE,

The clock SI is really good, we could monitor the proper clock in scope with rise time and fall time of 2.5ns.

We tried terminating the LCD clock by connecting pulldown resistor to ground, but still we could see the radiations of 60MHz harmonics, when we terminate/remove the USB clock out pin series termination resistor (Isolating the USB_CLKOUT pin signal from USB controller to Microcontroller) we couldn't see any radiations.

We tried with 50R,100R and with capacitor of 27pF,10pF and 3.3pF termination for USB clock out pin but still we couldn't arrest the radiations of 60MHz harmonics.

We tried adding the different series termination resistors to data lines also, but couldn't see any improvements.

[LCE]

Did you see any changes in radiation when changing clock termination?

Maybe it's something else, bad grounding or voltage buffering close to this chip area?

[Tejas_P_R]

No we couldn't see any changes in radiation levels after changing clock termination on USB_CLKOUT pin.

We have very good Power and Ground planes below the controller and USB controller, still we are facing the issue only when USB_CLKOUT pin of USB controller is connected to main Microcontroller.

Looks very strange, these radiations are spreading through the I/O lines of the controller where we connected the wires.