QN9083 Won't boot when warm

[joseraffucci]

I'm experiencing a weird problem with our QN9083 boards.  They work fine until they get slightly warm.  Not hot, just warm.  A few degrees over room temperature. 26C-28C (measured internally).  The boards operate when warm but when I cycle power, they won't boot until we cool them down.  We've narrowed it down to something on the MCU.  We can get a board in this state, touch the MCU with a cold cotton swab and it comes right up without any other intervention.

We've got 3 main suspects, none of which have left a smoking gun:

• Power
  • We've scoped the input power and it's clean at the pin.
• Clock
  • I'm delving into the SystemInit() code to see if there's a default setting from the DK board that's somehow not compatible with ours.  I'm still digesting the settings -- a lot of the PMU and VDD registers aren't documented very well.
  • I've but an LED blink routine at the start of main() and it doesn't get hit when it boots warm so whatever the issue is, it's happening prior to my code getting called.
  • I've gotten rid of the XTAL delay in SystemInit().  No change.
• PCB
  • We've got 12 boards all behaving the same.  It could be a BGA issue? 
  • DK board doesn't exhibit this behavior.

Has anyone run into something similar? 

[joseraffucci]

I've got more details now...

We removed everything we could think of from our board except an led. No external clocks, peripherals, etc.  Power is coming directly from bench top power supply.  Board fails to boot at 1.8V-2.1V.  Once we feed it 2.2V, it boots reliably.

We took apart a QN9080 USB Dongle board and repeated the experiment.  Board never failed to boot.  1.8V-3.3V input.  Same temperatures, same led blink code.  Unfortunately this board has a QFN part so it's not an apples to apples comparison.

Suspicions are now with the package itself or with our PCB manufacturing.  Can anybody help me repeat the experiment on a working board with a QN9083 (BGA) part?  Just getting it warm and trying to put it into ISP mode would be enough to confirm the problem.

MCU Info:

CHIP ID:         0xFC009010
CID0:          0x0
CID1:          0x2
CID2:          0x0
CID3:          0x0
CID4:          0x2
MEM_OPTION:    0x1
ADC_OPTION:    0x1
FLASH_OPTION:  0x1
FPU_OPTION:    0x1
USB_OPTION:    0x1
FSP_OPTION:    0x1
MAC ADDRESS:   E0:51:24:00:F1:04

[gerardo_rodriguez]

Hello Jose Raffucci,

Unfortunately, I don't have a QN9083 board to test this, but I will ask a colleague if they can test this scenario with a QN9083 board. The QN9083 parts are qualified to meet the operating conditions from the datasheet (T amb = -40C to +85C, and VCC 1.62V to 3.6V), so I would doubt that this is a problem with the QN9083 itself.

To confirm your issue:

-12 boards failing to boot at 1.8V-2.2V when being powered on at 26-28 °C or higher temperatures.

-The failing boards are able to boot correctly at 1.8V-2.2V only if the temperature is lower than 26C. They can keep operating even if the temperature rises but the problem occurs again after a power cycle.

-Your chip revision is C, by looking at CID4 = 0x2 from CHIP ID.

Some extra questions:

-Do these 12 boards operate correctly at 2.2V or higher and high temperatures?

-Do you have any other QN9083 board that operates correctly at all voltages and all temperatures?

-What is the power configuration of the QN9083? DC-DC mode or LDO mode?

Thanks,

Gerardo

[joseraffucci]

Failing to boot at 1.8V-2.1V.  Works OK at 2.2V and up.

This is our first batch of boards and the only QN9083 boards that I have access to.  That's why I'm throwing out a lifeline to see if it's possibly a packaging issue.

DCDC is bypassed.  We're feeding VDD and VCC directly from power supply.  All register options are the defaults at boot time.