Pristine STM32L431 PA6 turned to hard output

[waclawek.jan]

When switching on a new, unprogrammed STM32L431, PA6 turns hard to output high.

This is SPI1 MOSI.

AN2606 should say this out aloud.

I know the 'G0 have this too, and AN2606 says this for them, although quite quietly and confusingly in a footnote.

There really ought to be a clean trail of documenting such thing - which may be fatal to new hardware and/or production tester - from RM through AN2606.

JW

[Imen.D]

Hi @waclawek.jan,

"SPI Tx (MISO) is handled by DMA. On the bootloader start-up after SPI initialization as soon as the bit DMATx enable on SPI CR2 register is set to 0x1, the MISO line is set to 3.3 V." => This note is available in the AN2606 for STM32G0, STM32G4 and STM32L412xx/422xx, but not for STM32L43.

I escalated this feedback to add this note for STM32L43x (via Internal ticket number: 173607)

Thank you again for the continued feedback; it is much appreciated! :)

Imen

[waclawek.jan]

Thanks, @Imen.D .

Jan

[Imen.D · Accepted solution]

Hi @waclawek.jan,

I would inform you that this issue is fixed in the new release of AN2606 (Rev 63).

Thank you for your important contribution.

[waclawek.jan]

Thanks, Imen.

Jan

[elu]

We too came across the issue you describe. (Thanks by the way!)

We have a design where the output of a comparator goes to PA6. With a blank MCU, the output of the comparator tries to drive down PA6, creating a current consumption of over 70mA through that pin! The VCC of the board drops by 0.3V !

The footnote in AN2606 Rev 64: "On the bootloader start-up after SPI initialization, as soon as the bit DMATx enable on SPI CR2 register is set to 0x1, the MISO line is set to 3.3 V." is definitely not the best way to warn the designers.

Such PIN behavior should be documented clearly instead of the currently convoluted explanation that tries to hide this horrible hardware bug.

[waclawek.jan]

I agree that this should be documented in the DS as a footnote to given pins (here PA6) to the pins table. Alternatively, the DS should contain at a suitable place (maybe below the same pin table?) a warning pointing to AN2606 for STM32 models where the bootloader is started in pristine chips (as consequence of the PEMPTY mechanism or similar).

However, it's not a hardware bug as such.

One way to tackle it is to power it up while holding NRST low before the first programming.

JW