We have a problem with the nRST pin of a STM32F100CBT6 microcontroller. We have been working on the problem for several days and we don’t find out what is exactly the cause of this problem.
The problem lies in the nRST pin. There is nothing but a pull-up 10KOhms and a 100nF capacitor connected on this pin, but the pin does't go high.
For an unknown reason we see an unstable signal on this pin during some seconds, that fluctuates around 2.5 Volts to 2.7 Volts. After a period of 1-10 seconds the reset signal goes to low and remains there forever. The only way to take the microcontroller out of the reset is to force a 3.3V level onto this pin. When the pin is again released, it behaves well during some seconds (3 to 20 seconds), then it starts the fluctuations during some seconds (20 seconds or more) and finally falls again to reset.
Power supply is applied through a linear regulator (LM1117-3.3) to all the VDD and VDDA pins. We have checked pin by pin that the voltage arrives well to all the supply pins. Decoupling capacitors for each power supply pin pair are 100nF and are located very close to the microcontroller (2-3 mm away). Power supply is stable and does not have any fluctuations. It is also accurate, as we have measured a level of 3.37V with an oscilloscope and two different multimeters.
We have built 2 boards with this microcontroller and we have exactly the same problem.
Sometimes the reset signal stays high during enough time to program the microcontroller. The problems happens with the microcontroller erased and with the microcontroller programmed. In the second case (microcontroller programmed) we use HSE and we can see that the crystal clock works properly (8MHz). However, WE DO SEE THE SAME FLUCTUATIONS superimposed to the clock signal. We have checked the PCB and there are no shorts or couplings between the clock lines and the nRST (although we see that they connect to the microcontroller through contiguous pins).
We have also tried to remove the crystal and to run from the HSI Clock (Hi Speed Internal Clock) but the problem with the nRST stays there.
We have tried to remove the capacitor and the pull-up resistor, and the problem is the same, although the fluctuation levels are a little bit lower
We have the IWDG disconnected and the only thing that is now soldered on both boards is the microcontroller and the regulator (with its decoupling 470uF capacitors and the input and the output of the regulator)
The ONLY way to make the system go out of reset is to make a short-circuit (a 56 ohm resistor) between the nRST pin and the 3.3V voltage rail, but we cannot then reprogram the device (our JTAG cannot make the nRST pin low under these conditions).
We are looking desesperately for a solution. Please help us.