ESD and MCU freezing

[Lucast]

Hi everyone,

we're using the STM34F407ZG microcontroller in our BMS board.

We have some issue during contact 8kV ESD applied in some screws above the boards. The ESD are applyed on all the screws but the problematic one are just 3 screws, the closest screws to the board (MCU).

The effects of the ESD discharge are two:

1. during the battery standby mode, the ESD shut down the board. The MCU has a "keep supply" pin, a PP output that is used to maintain on the power of the board. If that pins goes low (or floating), the board power shut down. I think that a reset on MCU could leave floating that pin and shut down the board.
2. during the discharge mode, the ESD freeze the board. The board is still powered and the main contactor is closed. To maintain the power on the board, the "keep supply" pin must be high; to maintain the contactor closed, the pin "contactor" must be high (it is a PP pin, with no SW pull-up/down resistor but with an HW pull down).To reset the BMS we have to disconnect and connect the main connector back. After the reset it works again.

The microcontroller has an active IWDG that seems does not reset the board in the point "B" (freezing).

1. Could, an ESD, reset or freeze the MCU (and the IWDG) in this way?
2. Are there some useful documents or MCU registers that explain or confirms this phenomena?
3. Why the IWDG does not works?

Thanks

[LCE]

From my experience, an ESD pulse can do anything to a digital IC, from resetting to destroying.

I would not waste time on checking the MCU, but work on the PCB layout and circuit protection, with the focus on case - ESD protection - internal ground connections.

Think like: "I'm a high voltage, how and where do I get through this device as quickly as possible?"

;)

[Lucast]

It is not so obvius the coupling method of the ESD. It is possible the coupling mechaninsm is by means of an RF field induced by the discharge.

Since it is not a destructive phenomena, I think is a good "starting point" to put all the unused pins in push-pull mode, with a pull-down internal resistor, drivingh them to low.

[LCE]

When you have not taken care of that from the beginning (of mechanical device, circuit and PCB design), and that in a very meticulous way, it is always not so obvious or simple...

And "not destructive" is nice, but not enough for CE or FCC.

I've been through many EMI-certifications, and I also learned the hard way. But there was never any damage done by any ESD tests, for that the design has to be terrible indeed.

But usually you can't save a design by changing MCU GPIO settings.
The ESD pulse might travel over supply and / or GND connections, so any case-PCB connection must be chosen a) carefully, and b) somehow decoupled.

 

[Tesla DeLorean]

I suppose if the whole chip latched-up the IWDG could be impacted, but how are you kicking it normally? Do you toggle an LED when you kick it, so you can see that it' has stopped internally?

Do you have an external watch dog? Any way to interrupt the MCU supply?

[Lucast]

---

@Tesla DeLorean wrote:

I suppose if the whole chip latched-up the IWDG could be impacted, but how are you kicking it normally? Do you toggle an LED when you kick it, so you can see that it' has stopped internally?

Do you have an external watch dog? Any way to interrupt the MCU supply?

---

I cannot be sure that the IWDG has kicked. I can suppose sometingh because once the MCU goes in reset, the pins goes floating and the DC/DC of the boards lost its enable control that is maintained on from the MCU itself (the DC/DC's enable when the MCU is off is made through a momentary switch).

I have not any ext. watch dog. The only way to interrupt the MCU supply is throught the MCU pin itself. If that pin is floating or 0V, the BMS shut down. If the MCU does not work, as happened after the ESD, the method to shut down the MCU supply is to remove the power from the board unplugging the connectors.

 

[Lucast]

I can replicate the ESD test in my lab now.

I see that with ESD compliance to IEC 61000-4-2, at level of 1,8 kV the MCU start to:

• block itself but the IWDT reset the board when triggered;
• sometimes, most frequently with high level ESD, the MCU freeze and the IWDT does not trigger. The output remains stable to some position. I try to reset the MCU by pulling low the nRST pin but it does not work. The only way to rest the MCU is to remove the power;

I tried to apply different solutions. The working one are:

• to force each unused pins of the MCU with push-pull/open drain in high/low position or with pull-up/down resistors;
• apply some ferrites on the cables of the board. Close to its connectors;

Now the board is immune to ESD up to 7kV but I wanto to reach the immunity up to 8kV.

Is there a way to identify the pins that are sensibles to ESD and leads to latch-up/reset ?

Thanks