VOS (Regulator Voltage Output Scaling) bits correct on STM32F427?

[David Ashman]

I am having issues with my STM32F427IIT6 Rev 3 running properly at temperatures higher than room ambient.  Clocks are set up for 168MHz with PLL from HSI source, no over-drive, and regulator is on.  I found that by setting the PWR->CR VOS bits = b01 or Scale Mode 3 the uP works as expected at temperatures up to 85C.  When PWR->CR VOS bits = b11 or Scale Mode, 1 the CANbus would periodically freeze and uP would not work well at temperatures just above room ambient.  This is opposite to the datasheet. Table 17 on Page 95 of the datasheet DocID024030 Rev 10 states that for 168MHz, I should be using PWR->CR VOS bits = b11 or Scale Mode 1... which is opposite to my findings.  Can anyone verify that the datasheet matches or doesn't match the Rev 3 silicon?  Has anyone had similar issues operating at higher temperatures?

Thanks!

[waclawek.jan]

VOS does nothing else just changes the core voltage, see datasheet, and this can be observed on the VCAP pins. Higher operation frequency requires higher voltage.

I'd look for the source of problems at the power source, power and ground lines/planes arrangement, at the decoupling capacitors and at the VCAP capacitors. Some

ceramic capacitor materials are notoriously voltage- and temperature-dependent.

Also, try a "known good" design, i.e. some of the related NUCLEO or DISCO boards.

JW

[David Ashman]

Thank you @Community member​  . I verified that the Vcap voltages match the Scale Modes as expected. My decoupling caps are always quality low ESR X7R type. I now believe my issue is with the HSI oscillator drift at high temperatures causing CAN errors. I've measured +2.5% HSI oscillator drift above room temperature; which will impact the CAN bus error rate. After using the trim bits to dial in the HSI oscillator, the CAN works well at high temperature. Since CAN timing is critical, I'll have to use an accurate external oscillator on the board respin.