STM32

...

• Industrial

  • PLCs
  • Motor control
  • Pumps and compressors

• Smart home

  • Smart appliances
  • Central alarm system

• Personal electronics

  • Keyboards
  • Smartphones
  • IoT tags and tracking devices

• Smart city

  • Industrial communication
  • Lighting controls
  • Digital power

• Medical and healthcare

  • CPAP and respirators
  • Dialysis machines

• Industrial

  • Advanced HMI
  • POS machines
  • Metering
  • Sensors
  • Medical and healthcare

• Smart home

  • Home applicances
  • Whiteware

• Personal electronics

  • e-Bikes
  • EV charging
  • Wearables
  • Audio

• 'Enable wireless connectivity'

• Dual-core architecture—MCU and a radio transceiver

• Industrial

• Smart home

• IoT

Before looking into the product selector, it looks as though I could find suitable microcontrollers in the Mainstream, Ultra Low Power, or Wireless categories—with the two former requiring some external connectivity module to meet my wireless requirements.

Of all the series, the STM32WB looks to be the most initially attractive, as this appears to fully satisfy my connectivity & processor requirements without requiring an external module. I will need to investigate further to see whether it satisfies the other microcontroller (ie peripheral) requirements however.

I have ensured to add a sufficient safety factor atop each of requirements in my applied filters, as I am still in the process of/yet to design many aspects of this project—including the Interface and Power circuits. I have consequently made sure to budget for enough surplus peripherals/GPIOs/resources such that I have sufficient room to grow, in case my requirements grow once I have settled on a microcontroller.

As I am only intending to produce a small handful of these boards (ie only enough prototyping/development boards to produce a final design/revision), and am not looking to commercialise/mass-manufacture these boards, I can afford to adopt a prototyping approach and over-specify my microcontroller, rather than strictly optimising my selection.

I will need to consider my USB Power Delivery controller peripheral—there are indeed STM32 microcontrollers with this feature, but I will need to consider whether it would be in fact better to delegate this to an external IC that is communicated with over SPI/I2C. My primary concern in this regard is my need for two USB Type-C Connectors, where I want to be able to have a Power Connector plugged into a USB PD source, and another Data Connector separately plugged in to my computer. I will need to ensure that this dual-connector application is supported by the microcontroller's peripherals, and that it is implemented safely.

Although full details can be found in USB Power Delivery > Implementation, I have decided to indeed apply a further filter for only microcontrollers with a UCPD peripheral. This will allow me to safely & reliably negotiate for my extended current sink requirements from the source, without requiring an external IC. This additional filter also does not box me in too heavily, as it narrows the available products (without the inclusion of QFN-type packages in the filter) from 521 to (still) 121 options.

I note that the STM32WB wireless microcontrollers I identified earlier are only available in QFN/BGA/CSP packages—so I will either need to settle for an alternative non-wireless enabled microcontroller and use an external module, or accept that I will need to use the QFN package. My primary concern here is that these are not easily hand-solderable, and I am trying to choose parts which I can assemble myself. With that said however, I have already:

1. Chosen Discrete Non-Addressable RGB LEDs (RS-1515MBAM) that will be too hard to hand-solder (not to mention the 735x quantity); and
2. Chosen LED Drivers (TLC5951DAP) with an exposed pad, which I will not be able to hand-solder.

The RS-1515MBAM LEDs are less concerning as they are ordered from LCSC—so I can put them through JLCPCB's assembly service. The TLC5951DAP drivers are more of a problem, as these are not available through LCSC—such that I will need to find access to a hot-plate/hot-air station to assemble them.

At that point, what is to stop me from using a QFN package? LCSC even has STM32WB microcontrollers in stock, so I could also have these assembled on the board for me. I will still stick with my SO-type TLC5951DAP devices however, as I suspect these two-sided packages will be easier to route nicely than a quad-flat package.

With the above considerations, I will include QFN-type packages in my search filter, and see whether STM32WBs will be suitable.

Applying my filters to the STM32WB product selector, I see that there a number of problematic filters that cannot be met—

1. The maximum number of Requirements > SPIs available is 2. I suspect this should be fine however, as SPI is a bus architecture after all—hopefully the LED Drivers will be the only peripheral that I need to talk to without a chip select line.

2. There are no microcontrollers with a UCPD peripheral.

Conceding the above, I narrow the selection down to five microcontrollers—with the only meaningful differences being their packages, flash/RAM, and whether they support Zigbee. These five microcontrollers are:

1. STM32WB35CC
2. STM32WB35CE
3. STM32WB55RC
4. STM32WB55RE
5. STM32WB55RG

Looking further at these microcontrollers, I think that I would rather choose a mainstream microcontroller and use an ESP32 or similar as an external connectivity module. This is due to a number of factors:

1. I would rather avoid the extra development associated with designing my own PCB antenna, and the need for JLCPCB's controlled impedance process.
2. I would rather have a microcontroller in an 'accessible' package/footprint.
3. Wireless connectivity is very much an auxiliary requirement—it is not listed in the project goals nor specifications.
4. By using an external ESP32 module, I can get more exposure & experience with both an STM32 and an ESP32.

With my final filters applied as per the table above in STM32 > Product Selector, I am brought to a final short-list of 121 options.

I see that these options comprise the following series:

I will further narrow my short-list by filtering out options with more than 40 Requirements > GPIO pins, as I will definitely not need more than that. This narrows the options down from 121 to 22—perfect!

At a glance, it looks like the Ultra Low Power microcontrollers are significantly better than the mainstream STM32G4 options—more flash/RAM, better processor, comparable clock speed, at just the cost of less analogue peripherals.

I see that even the option with the most flash & RAM, the STM32U585CI, is available for NZD$11.07—in my quantities, this is not a significant from the cheaper options at ~NZD$7. I will therefore select the variants of each series with the most flash & RAM available for comparison against each other—this reduces the number of options from 22 to 10.

Taking a quick look at ST's product naming convention, I see that they take the form

such that I would expect my final part number to be an STM32abbbcdTe for a QFP-type part.

Looking at the table of options above, I see that the primary difference is indeed the number of analogue peripherals available on the mainstream (G4) vs ultra low power (L5 and U5) series of microcontrollers.