Problem with stm32F446 output on pa6 and pa7

[Simon Andersen]

Posted on January 22, 2018 at 21:58

Hi, I've made a pcb using the stm32F446 mcu, 64 pin package, but I have a problem with it. I have four LED's attached to pa4 to pa7. Two of these work fine, but the ones on pa6 and pa7 only lights up entirely when both are on. I tried to remove everything attached to the pins, and saw on an oscilloscope how the output isn't the supposed 3.3v, unless both are on.

From here, you can see four stages: {both off},{ pa6 on, pa7 off}, { both on},{ pa6 off, pa7 on}. The probe is only measuring pa6.

Furthermore, when programming the chip using VisualGDB and stlink, sometimes it does not go to the programmed area after. When it doesn't, it goes there after a restart. And when it goes into this, pa6 and pa7 are both high. This is not much of a problem, but they might be correlated. I've noticed that the boot pin is floating, as it wasn't needed for the design, could this be causing the problems?

Thank you for any help

Simon

[waclawek.jan]

Posted on January 22, 2018 at 22:04

I've noticed that the boot pin is floating, as it wasn't needed for the design could this be causing the problems?

Yes, definitively.

JW

[waclawek.jan]

Posted on January 23, 2018 at 01:00

... the 'sometimes it does not go' problem, I mean.

The waveform is a different issue and indicates short between the pins.

JHW

[Andrew Neil]

Posted on January 23, 2018 at 01:30

What's with all this taking photographs of the screen of a digital scope?

It can give you a screenshot straight to a USB stick - without all that glare, keystone, etc ...

[Simon Andersen]

Posted on January 23, 2018 at 13:50

When in a lack of a USB stick in the lab, the camera is faster. However, I surely agree with you.

[Simon Andersen]

Posted on January 23, 2018 at 13:49

Hooking boot0 to ground solved the problem of going to the bootloader. But the waveform is still a problem. I've had the pcb under a microscope to check for any shorts, which are not present. Furthermore, I've tried to measure the resistance between the two pins when no power is supplied, and it is 1.5 Ohm. This is also checked on a mcu not connected to anything, where the same resistance was measured.

Obviously pa6 and pa7 have something to do with the bootloader, and I've found that it is hooked to SPI1, pa6 input and pa7 output. However, it should behave 'normal' when the bootloader isn't running, so I don't see what the problem could be.

[waclawek.jan]

Posted on January 23, 2018 at 16:05

Can you please try on a 'known good' board, e.g. a Nucleo?

JW