Noise Injection via ST-Link Causing MCU Disturbance on STM32H743ZI

[PrashantBhatta]

we are facing after migrating our design from STM32F405ZGT7 to STM32H743ZI.

We have an automotive application where a horn load creates significant EMI transients. With our previous design using STM32F405ZGT7, the system was stable and did not experience any MCU freezing or malfunction during horn activation.

However, after migrating to STM32H743ZI — while keeping the PCB layout, grounding strategy, power architecture, filtering components, and external circuitry identical — we began experiencing intermittent MCU freezing during horn actuation.

After extensive debugging and hardware validation, we observed the following:

The issue only occurs when ST-Link is connected.
When ST-Link is disconnected, the system behaves significantly more reliably.
The noise appears to be injected through the ST-Link interface path.
The horn generates high transient noise despite having local snubbers and filtering.
We have already implemented:
RC snubber across the horn load
EMI input filtering near MCU supply
Proper decoupling near all VDD pins
Ground plane continuity and star grounding where applicable

This behavior was not observed with STM32F405ZG under identical conditions.

We would like clarification on the following points:

Are there any known differences in debug interface robustness between STM32F405ZGT7 and STM32H743ZI?
Is the STM32H7 debug interface more sensitive to ground bounce or transient noise?
Are there recommended protection or isolation methods for ST-Link in high EMI automotive environments?
Is there any application note addressing EMI robustness of SWD/JTAG on STM32H7 series?

We suspect that higher core frequency, different internal architecture, or debug interface implementation in STM32H743ZI may be contributing to this sensitivity.

[AScha.3]

> Are there any known differences in debug interface robustness between STM32F405ZGT7 and STM32H743ZI?

no.

>Is the STM32H7 debug interface more sensitive to ground bounce or transient noise?

afaik no ; but in general : any ground bounce or transient will break the debug connection.
> Are there recommended protection or isolation methods for ST-Link in high EMI automotive environments?

Maybe st-link isol can help:

the ST-LINK/V2-ISOL features digital isolation
between the PC and the target application board. It also withstands voltages of up to
1000 Vrms.
https://www.mouser.de/ProductDetail/STMicroelectronics/ST-LINK-V2-ISOL?qs=19rwF8CI%252BwmRqTUzxHKfrw%3D%3D

btw

Which st-link you using ?

[Andrew Neil]

Welcome to the forum.

Please see How to write your question to maximize your chances to find a solution for best results.

---

@PrashantBhatta wrote:

The noise appears to be injected through the ST-Link interface path..

---

So what does your "ST-Link interface path" look like? How, exactly, are you connecting the ST-Link?

Please show your schematic, and some good, clear photos of your setup.

And what ST-Link are you using?

How to recognize a genuine ST-LINK/V2 versus a cloned one

May be worth noting that no STM32 products are qualified for Automotive use.

[jibon]

Hi,

It is a common phenomena when processor speed is getting higher. I also observed ST-Link issues when ST-Link is connected to the Board with Ribbon Cables. Then I separated the ribbon cable from each other and keep the distance as far as possible from each other. It solved the issues. Since STM32H7 is high speed, It causes more noises over the ST-Link cable. You can try this and see.

[PrashantBhatta]

I am using this cheap stlink tool available locally. The greater problem is that if the MCU goes into the freeze state, even pressing the reset button does not work, I have to do a full power recycle.

[PrashantBhatta]

I am using STM32H743ZIT6 with STM32CubeIDE 1.19.0.
1. ST-Link Connection Details
I am using an external ST-Link (ST-LINK/V2).
The SWD signals are connected as follows:
SWDIO (MCU pin) → 100 Ω series resistor → SWDIO_Out (connector)
SWCLK (MCU pin) → 100 Ω series resistor → SWCLK_Out (connector)
Both SWDIO and SWCLK have:
33 pF to GND
BDFN1C071V ESD diode to GND
USART1 TX/RX are also routed to the same connector with 100 Ω + ESD protection.

So electrically:
SWDIO/SWCLK go directly from MCU → 100 Ω → connector.
ST-Link connects at the connector (SWDIO_Out, SWCLK_Out).
No level shifting.
Shared system ground between ST-Link and board.
Please let me know if this termination (100 Ω + 33 pF + ESD) could disturb SWD signal integrity on H7.
(Note: even if the caps, series resistor and the esd diode are removed, the freeze issue is still there)

2. Horn Driver Section
Driver: VNQ7140AJTR
Horn current: ~2.5 A
Output path includes an EMI filter before going to the horn:
22 µF + 100 nF on control side
Common-mode choke (4A, 70 µH @ 100 MHz)
X1/Y2 capacitors to chassis
TVS diode (26 V)
1N4007 diode
Important observations:
The freezing issue was present even before adding the filter.
After adding the filter, the issue reduced, but still persists.
So the filter improved the situation slightly but did not eliminate it.

3. Power Supply Architecture
The supply chain is:
12V automotive input → LM61460AASQRJRRQ1 → 5V → SPX3940AM3-L-3-3 → 3.3V
Details:
12V to 5V:
Buck: LM61460AASQRJRRQ1 (TI, automotive grade)
Layout and EMI filter strictly followed as per datasheet
Proper input/output ceramic caps as recommended
No overheating observed
5V to 3.3V:
LDO: SPX3940AM3-L-3-3
Output caps per datasheet
3V3 rail feeds STM32H743ZIT6

4. Observed Failure Behavior
When the horn is pressed:
MCU freezes.
Pressing the RESET button does not recover the MCU.
Only a full power cycle restores operation.
This is what confuses me most:
If this were just firmware lockup, reset should recover it.

5. Important Comparison
The exact same board design worked fine with STM32F405ZGT6.
After replacing MCU with STM32H743ZIT6, the issue started.
No major layout change.

6. Specific Questions
Why would RESET not recover the MCU, but power cycle does?
Could H7 be more sensitive to:
SWD line disturbances?
Internal regulator instability?
Could the ST-Link ground reference be injecting noise into the MCU core supply?
Is the 33 pF on SWDIO/SWCLK too aggressive for H7?

The following images shows the SDIO lines connections, HORN filters and Debouncing and vcap filters around the stm32

[jibon]

Can you perform a test? If you are using a Laptop, Disconnect the laptop charger and run the Laptop only on battery. Just to see the result.

[PrashantBhatta]

sure, I will try that and will post if it works.

[AScha.3]

>Why would RESET not recover the MCU, but power cycle does?

Possible a current spike creating a latch-up (only power cycle can change this state) -

or your circuit /decoupling is bad.   -> So show complete circuit around cpu + supply (+ caps).

>Could H7 be more sensitive to:  SWD line disturbances?

I never had a problem with a H743 , same as other cpu.

[LCE]

+1 GND loop

When I connect the ST-Link to my PoE-isolated ethernet-audio front-end and then connect also non-floating signal generator, I see some noise in the audio channels. Classical ground loop.

Check where the horn's power supply current might find a path.