0
\$\begingroup\$

I have a problem with a design that uses 2 S9KEAZN64AMLC microcontrollers. I have encountered an issue, where it seems that if the device is turned on with a switch i.e. with a very fast rise time on both the supply voltage and the 5V rail, then the device can fault out. Whereas if the device is powered with a more gentle rise time it always boots correctly.

I thought this was a closed case and to fix the issue I would add capacitance to slow the turn on. This does seem to fix the fault. But I want to understand it more. Not all of the units I have exhibit the fault, perhaps 10% do.

See below plots, where:

  • Blue = 12V supply from the power supply, switched with a relay.
  • Pink = 5V rail powering the microcontroller
  • Green = Non-maskable interrupt (NMI) (PTB4, Pin 10)

Microcontroller operating ok Micro operating ok

Microcontroller not operating Micro not operating

You might then say that it is due to the NMI pin being driven low during start up. And that I should comply with the final point of the Errata about the VDD ramp up.... perhaps.

NMI Errata

However I then tried to recreate the fault, by removing as many decouple/bulk capacitors on both the 12V and 5V rails. This brought the rise time down to 3us, so I was expecting a fault condition, however it never faulted out.

So in summary, I think I have a solution to my problem, increased capacitance on the decoupling capacitors near the microcontroller and on bulk capacitors on 12V/5V rails. However I don;t understand the different result between different boards.

Any thoughts? If you want any more information, drop a comment and I will add information.

Further Info / Schematic Microcontroller Schematic

Power Supply

\$\endgroup\$
1
  • \$\begingroup\$ The first picture shows a dip in the 12V along with an oscillation on the nmi. Could just be probe setup or something more sinister. Placement and spec of your caps might need looking at - track inductance and capacitance vs voltage bias of the ceramic caps. Does the regulator like ceramic caps - ie low esr? Different brands might work differently. As to why two boards act differently - could be a borderline issue, different components etc. Nailing these issues requires a scientific approach. \$\endgroup\$
    – Kartman
    Commented Apr 21, 2021 at 23:51

1 Answer 1

0
\$\begingroup\$

That NMI pin is curious but it might just be a symptom of the MCU rebooting over and over.

Some comments:

  • Nothing in your post addresses low-voltage/brown-out detect, which is strange, as this is the #1 thing to go check. What level have you configured it at and how do you handle it? You have pretty wild contact bounces in the critical area between 2V and 3V. So it is very likely that LVD will kick in at multiple times.

  • When the part eventually boots, what does it tell you was the reset cause? Power-on reset or something else?

  • A larger bulk cap on Vin before the LDO is a good thing regardless of if you intend to connect bouncing things there or not.

  • From the schematic you have provided, we can't tell what goes on with your /reset pin. Do you provide external pull-up or not? What decoupling cap value do you place there and what does NXP recommend? This cap value is often very critical.

  • Another thing to try is to temporarily tie VADC directly to 5V. In case different parts of the MCU get powered-up at different times, it can also result in LVD kicking in and you might end up in something resembling latch-up. I've had some very subtle problems with this in the past on NXP parts.

  • How is your SWD connected and do you always have the part powered when programming? Or do you provide power over SWD? This is a common mistake - 5V supplied backwards into your 7805 when there's no input voltage will damage the regulator.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.