Should you power the 3.3V plane with the USB internal regulator ?

[Arte]

Some chips like the STM32H750x series can generate 3.3V from 5V via a USB internal regulator.
See: https://wmsc.lcsc.com/wmsc/upload/file/pdf/v2/lcsc/2304140030_STMicroelectronics-STM32H750IBK6_C730200.pdf, @VDD33USB

From what I understand, the 5v is provided at VDD50USB and the resulting 3.3V is then made available via the VDD33USB pin.
Should you save a buck converter by connecting VDD33USB to the other power supplies (VDD, VDDLDO, VDDA assuming your analog accuracy requirements at low) ?
Or will you run into problems I don't foresee ?

[hans]

The VDD33USB pin supplies the USB internal transceiver. (On smaller packages its connected to VDD)

The VDD50USB pin is available on the larger packages as a 3.3V regulator input.
However, the regulator is disabled by default. The chip boots up in bypass mode (VDD33USB = power input), and you can enable the regulator to then supply that pin from VDD50USB.
That means that powering the STM32 won't work as the regulator won't get enabled.

I think the only selling point for having this regulator, is so you can power the MCU at VDD=1.8V and have USB operate in its own power domain of 3.3V. There are no specifications on this regulator, so its anyone's guess if it can deliver say 100mA for the STM32 part itself.

See 6.4.7 of the reference manual
(and perhaps AN5419)

[ajb]

I think the only selling point for having this regulator, is so you can power the MCU at VDD=1.8V and have USB operate in its own power domain of 3.3V. There are no specifications on this regulator, so its anyone's guess if it can deliver say 100mA for the STM32 part itself.

yes, the datasheet also says that the 3.3V regulator is intended "to supply the internal [USB] PHYs" when VDD is not 3.3V.  If it were intended to supply anything beyond the chip it would hopefully have some sort of specifications in the datasheet.  You might be able to power a small amount of other circuitry, but it's probably not worth the potential disruption to the PHYs if you're using USB, especially in a higher speed mode, and of course the usual caveats of working outside the datasheet parameters apply.

Should you save a buck converter by connecting VDD33USB to the other power supplies (VDD, VDDLDO, VDDA assuming your analog accuracy requirements at low) ?

If you truly need a buck converter in the sense of a switching converter (not just a 'step down' in voltage), then the built in regulator wouldn't be a great replacement anyway, since it's a linear regulator, and therefore its efficiency and power capacity will be much lower.

[Arte]

Got it, thanks.

[pcprogrammer]

The datasheet in this case states that the 3.3V for the USB peripheral can be obtained by either connecting the 5V from the USB connector to VDD50USB pin or by connecting 3.3V to the VDD33USB pin. So I interpreted it as being the supply for just the USB peripheral, and not as an option to supply 3.3V to other devices.

It does need a decoupling capacitor on the VDD33USB pin either way.