NRST / Reset Pin in STM32H7 more ESD sensitive than in STM32F4?

flyer31 · ‎2019-04-23

Hi,

I have a new PCB here which somehow has latch up problems on the NRST pin quite recently. Usually then the NRST pin would just have zero DC resistance (or typically 1 Ohm) to GND. And then of course also the oscillator (8MHz XTAL) will not start working and chip keeps in reset (this is then also the error message of ST-LINK).

I use an external 10kOhm resistor against 3V3 (VCC) and then I have an about 5cm long PCB trace to the 4pin ST-LINK Debug connector... .

In the last 3 weeks I had about 5 times such a latch up problem with this PCB (STM32H750 processor, 100pin LQFP, "Y" version - this is still the startup version I think according to errata sheet...).

Now I added a 100nF capacitor to the reset line against GND. Hoping that it will be fixed now (Nucleo H743 also has such a cap, and in the datasheet they recommed it if a reset button is installed ... I do NOT have a reset button in my PCB, but maybe this cap anyhow somehow much better ... just I never had to do this with STM32F4 PCBs ... and also I never had any STM32 latchup problems before ... this is my first STM32H7 PCB...).

Uwe Bonnes · ‎2019-04-24

NRST is designed to work with the chip internal pullup and an external 100 nF capacitor. With your setup with no capacitor and the external 10 k pullup, reset timing is screwed. However this should not cause breakage. Are you sure there were to ESD discharges or external voltages above 4 Volt involved?

flyer31 · ‎2019-04-24

This new PCB also has a 230V rms test circuit on board ... this is new to me ... but it is all isolated nicely with isolated supply for the 230V part (>2kV isolation voltage, 8mm creepage paths...), and I am not doing nasty things on the 230V side ... . And both circuits have no connection to PE (just standard Y decoupling capacitors...).

So due to this I cannot answer this for sure for now. Now I anyway added the 100nF capacitor ... the next PCB will be much more standard, without 230V ... when I do this (and if I have the courage to again try without 100nF cap), I can tell you more... . (but possibly I will better use the 100nF cap now always, to avoid this problem..).

PS: Meanwhile I added a 300uH - 500mA 12x12mm SMD choke in the 24V supply line for the CPU voltage supply, and now the problems seem to be fixed. This PCB is a 10A 24V DC PWM supply for a lamp ... I have no real experience such high current PWM so far ... the latch-up always happened, when my 24V supply was without current control (current control set to max limit of 11A - somehow then the 24V supply on switching on made crazy things). ... After I recognized this, I inserted the choke, and now fine (but I also added the 0.1uF in the reset line))... . I think this probably was due to this missing choke ... in "usualy PCBs" I always would use a current controlled choke for the complete PCB ... but current controlled choke for 10A would have been a bit large of course :).