What are you all switching to when your usual chip is out of stock everywhere?

I swapped families mid-project about a year back on a motor driver board, so I can tell you what the pain actually looked like. Short version: if your firmware is mostly C with your own logic and you're not deep into vendor-specific libraries, the port is days of work, not months. If you're using a vendor HAL everywhere, it's more like a couple weeks of grinding through peripheral init code.

The toolchain side transfers over better than you'd hope. Anything with an ARM Cortex-M core works with the same GCC setup and the same SWD debug probe, so your compiler, debugger and flashing workflow basicly carry across. What doesn't carry is the startup code, the clock tree setup and the peripheral registers. Timers are the big one for motor control, since the PWM arrangements and dead-time registers are all family specific. Also check the ADC specs carefully, because the sampling behavior and calibration routines differ more than the datasheets suggest.

If you want the lazy path, look at the GD32 parts, which are pin compatible with a lot of the popular STM32s. People have literally dropped them onto existing boards and had things mostly work, though I'd verify the flash wait states and the USB if you use it. The catch is documentation, which is thinner, and long term availability isn't always clear. For a demo deadline that might be exactly what you want, but for production I'd think harder.

The other realistic option is Renesas or TI, and those have solid availability right now, but then you're doing a real port. Renesas in particular has some nice motor control peripherals and thier tooling has improved a lot.

My actual advice: buy a cheap dev board for whichever candidate part you like this week, get your motor spinning on it before you commit to any layout changes, and only then cut the board. A weekend of testing beats guessing.

Been on the wrong end of this twice now, and the second time I got smarter about it. What actually saved me was a cheap logic analyzer sitting between the motor driver and the MCU, comparing the PWM output from the old board and the new one side by side. Timing quirks you'd never spot in the datasheets just show up right there on the screen, and I caught a dead time issue in about an hour that would have burned a week of debugging later. Also worth checking stock across a few distributors directly before you fall in love with any part, some of the availability numbers out there are stale.

Before you commit to either path, it's worth pushing your distributor on the actual allocation situation rather than the quoted lead time, because in my expirience those 40 week numbers are often soft and the real answer depends on where you sit in their priority queue. I've also had luck checking a couple of the authorized second sources directly, since sometimes a sibling part in the same family with a slightly different flash size or package is sitting in stock and runs your binary unchanged. Annoyingly the stock snapshots on the big distributor sites can lag by days, so a phone call still beats the web page for anything you're about to design around.

One thing I did on a similar deadline was split the problem. Bought a small batch from a broker at a stupid markup, just enough boards to hit the demo, and kept the redesign as a seperate task with no pressure on it. Cost me maybe an extra hundred bucks all in, way cheaper than rushing a port and shipping something half tested. Only caveat is you have to be careful where you source those broker parts, I stick to ones that will take the parts back if the date codes look off. Then once the dust settled I did the family swap properly on my own schedule.