2022-05-21 03:54 AM
I've spent the last 2 years developing a product using an STM32F103, but I've recently discovered that the ones I've been supplied with are fakes, which have a tendancy to blow up :( It seems as if no genuine F103s are available now, so either I have to throw away 2 years of work, or switch to a newer, widely-available STM32 - assuming there is such a thing?
2022-05-21 04:05 AM
Due to the Great Chip Shortage, any suggested alternative would be hard to find... but perhaps try to start with the 'F30x (see AN4228).
"Equivalency" depends on the particular feature mix you've used. One thing is surely different, the 'F1 has a different GPIO structure than all other STM32.
JW
2022-05-21 05:40 AM
CubeMX is supposed to have a means of indicating equivalent parts.
I'm not sure the 64-pins devices have been consistent, the 144-pins ones have been more so.
Not sure availability on the F3's is significantly better, the F1's are still made and consumed in large volumes, it's just that all available production is spoken for.
You might do better looking for parts you can source, and then pivot into those.
2022-05-21 10:41 AM
The STM32G491CCU6 at Mouser looks hopeful.
Moving to another STM32 series is not throwing away years of work. With the knowledge and experience you already gained, moving should take a maximum of few months and you will gain additional features and experience.
2022-05-23 01:03 AM
Thank you for your answers. So my problem isn't due to the F1 being discontinued, but 'just' the global chip shortage. (Although, if that's the case, why can't I order F1's directly from ST?)
As far as I can tell the F3 isn't any more available than the F1.
The STM32G491CCU6 looks really nice - the RNG would be particularly useful for my product, however it only seems to offer pins for the first 4 ADC1 inputs, but my project uses all 8 ADC1 input pins available on the F103. Is there any way around that limitation?
I'm also imagining a scenario where I spend 2 months converting my project to the G4 only to find the stock of those has fallen to zero - it could presumably happpen tomorrow!
Incidentally, I'm assuming the F103's I had fitted by JLCPCB were fakes, but how can I tell for sure?
2022-05-23 03:39 PM
I see 10 single-ended inputs on ADC1 alone, and there is also ADC2 and ADC3.
> Although, if that's the case, why can't I order F1's directly from ST?
You have some strange understanding of the term "shortage"... ;) But you probably can order, just the delivery times will be huge.
> it could presumably happpen tomorrow!
Unfortunately it can... It's just that for the newest parts it's less likely.
2022-05-24 06:31 AM
Regarding ADC1 inputs, you are correct - I didn't notice the other 6 inputs :) In my defence, they're arranged pretty randomly.
I'm still debating whether to try redesigning using the G4 or not. One problem is that I can't find any dev boards, so I'd have to use my product as the dev board. It also looks like I'd need to redesign the PCB layout quite a bit - and probably quite a lot of the code, as I (probably unwisely) went for 'baremetal' register manipulation for speed (and learning) reasons. OTOH, I only used ADC1, SPI1 and GPIOs.
When I asked why I can't order F1's from ST, I meant that they're just listed as "out of stock", with no lead times, unlike some of their newer products.
2022-05-24 01:44 PM
Moving HAL code from F1 to G4 also wouldn't be a simple copy-paste. And as a "bonus", with HAL you get a bloated unreliable junk. GPIO is trivial, ADC is simple and, I don't know about SPI, but newer peripheral implementations are typically easier to use.
2022-05-24 02:45 PM
Well they'll remain "out-of-stock" because anyone waiting from here to the middle of next year for delivery is going to grab anything they can get today.
The trick is to order the parts before you need them, not the day before. The industry is entirely driven by "built-to-order" so no excess inventory is created or needs to be warehoused for indeterminate amounts of time.
There's likely no lead-time given for some because it's probably close to 80 weeks, and it's probably not prudent to get that far over your skis. Most business can't put money down and tie it up for that long. There's a huge bubble, and in all likelihood a recession coming in 2023