18
\$\begingroup\$

I have received two boards from one of our customers that have the same defect. It is a short circuit between VCC and GND. I was able to trace this short circuit down to the MCU (which is a LPC1114 from NXP) with a thermal camera. These boards were running quite some time at the customer site, until the short occurred.

In the pictures below you can see two different MCUs. One that is still soldered on the PCB and one that was removed from the PCB and supplied with VCC/GND.

Enter image description here

Enter image description here

Edit1: In both pictures, Pin1 of the CPU is on top-right corner (TOP view of the MCU).

My questions to you:

  • Has anyone ever experienced such an internal MCU short?
  • What could be the cause of such an internal MCU short? (Guessing accepted :-)
  • Is it possible that such a short could happen when there is too high temperatures involved?

Edit2: Pinout of the MCU MCU pinout

\$\endgroup\$
6
  • 3
    \$\begingroup\$ Yes, has happened and can for almost any reason. You will want to show schematics how the MCU is powered up and connected to external circuits. \$\endgroup\$
    – Justme
    Jul 15, 2021 at 13:18
  • 2
    \$\begingroup\$ @Justme, "has happened and can for almost any reason". The vast majority of MCUs don't so I imagine you mean "for a small variety of reasons". \$\endgroup\$
    – TonyM
    Jul 15, 2021 at 13:43
  • 2
    \$\begingroup\$ This is a result of shoot-thru (SCR effect and self-destruct) usually from long interface cables and a transient nearby. \$\endgroup\$ Jul 15, 2021 at 13:43
  • 1
    \$\begingroup\$ Stefan Wyss - Hi, Which corner on the MCU is pin 1 in: (a) the top photo, and (b) the bottom photo. (c) Which package is that MCU? It looks like 8 pads per edge = HVQFN33, yes? (d) As commented earlier, will you be able to supply schematics and explain how the boards are used / integrated into the customer's environment? Thanks. \$\endgroup\$
    – SamGibson
    Jul 15, 2021 at 15:11
  • 1
    \$\begingroup\$ I have seen something like this and never figured out the root cause. It was infrequent. \$\endgroup\$
    – mkeith
    Jul 16, 2021 at 4:48

3 Answers 3

29
\$\begingroup\$

Typical cause would be some electrical transient from outside coupled to an I/O pin, which causes latch-up and destruction via the power supply. Proper design of protective circuitry will generally prevent that in situations short of a nearby lightning strike.

Power supply transients could also cause it, only a few volts might be enough, but usually that does not occur.

Momentary reversal of the supply voltage could also cause it.

\$\endgroup\$
3
  • 3
    \$\begingroup\$ Could also be a power sequencing issue causing the latch up. \$\endgroup\$ Jul 15, 2021 at 15:53
  • \$\begingroup\$ Yes, this becomes essential for satellites. \$\endgroup\$ Jul 16, 2021 at 16:57
  • \$\begingroup\$ @PeterMortensen Single-event latchup is almost always a transient phenomena. Once you remove power the latchup disappears, and the effect will not be seen when power is again applied. \$\endgroup\$ Jul 16, 2021 at 19:21
13
\$\begingroup\$

I agree with other answers that an external overvoltage on I/O pin is a likely cause, though an external overvoltage on VDD pin is also a possible cause.

To fix the circuit, it would be useful to know which I/O pin received the transient. Sometimes this can be determined by measuring the internal protection diodes of the chip using the multimeter diode mode:

  1. Measure with the black lead on the VDD pin and the red lead on each I/O pin. Take note of readings.
  2. Measure with the red lead on the GND pin and the black lead on each I/O pin. Take note of readings.

For a fully functional chip, the expected values would be 0.6 V on each pin. Often a damaged I/O pin will show either an infinite or zero diode forward voltage, as the protection diode has burned out before other parts of the chip got damaged.

\$\endgroup\$
9
\$\begingroup\$

To my knowledge, yes, it belongs to a transient from outside. An I/O pin or VDD with a scarce transient current and high I/O pin loads will short the CPUs.

  • In the best of cases, sometimes, it's an internal diode or a (default) latched pin at the output inducing the short.
  • It may remain shorted for undetermined time until the diode or MOSFET burns completely (depending on its maximum running current).
  • If the power source to which the board is connected is a linear voltage source, or an SMPS with scarce current pumps (low frequency switching SMPS or defective switching capacitor or both), the effect of latched I/O pins may have extend to the CPU internals and never recover.
  • If the CPU has low current consumption even if shorted it may be an external component wired to a interrupt pin halting the CPU from start in an indefinite looping state
  • There may be several other temporary VDD to ground running conditions to investigate.

Here are a few more ideas:

  • Try using a precision voltage/current source to power sequence and check the oscilloscope for possible latched pins while it boots
  • Try identifying I/O pins which may be open collectors and take measures. If it's fed high while latched low, you can temporarily reverse the pin voltage and the boot-sequence may jump over the current drain and actually boot
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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