Ideas for a bluepill-like dev board for STM32F7x0

[RuBisCO]

STMicroelectronics recently released a new "value-line" series of MCUs in their STM32F7 family referred to as "STM32F7x0", which cuts down on embedded flash memory (down to 64KB, same as STM32F103C8) to reduce cost.

Boy, did they reduce the cost.

Compared to the STM32F103C8 on the bluepill, which costs ~$3 in single and under $2 in bulk, the STM32F730R8 costs only a dollar more in single quantities and is just over $2 in bulk.
The CPU on the STM32F7 is way more OP and it comes with more than double the peripherals and an insane amount of RAM.

I was wondering if anyone had ideas for, or would be interested in developing a board similar to the bluepill but with the new STM32F730R8 instead of the STM32F103C8. Such a board could probably only cost $8 or less to the end user.

I'm pretty sure this chip has a DFU, so it wouldn't even need a USB-Serial bridge to burn a bootloader.

[madias]

Indeed, the STM32F730R8 (only 64pins) sounds interesting.
https://www.st.com/content/st_com/en/pr ... 730r8.html
A bluepill on steroids
Maybe we will get such a devices in 3-5 years on Ali or Ebay.

[heisan]

That is some serious hardware. Add a QSPI flash, and some QSPI SRAM, and I am pretty sure I could get Linux running on it - then you can use it to host an Arduino build environment .

[stevestrong]

What about pin compatibility with H7x or F4x chips? We have already 2 forum threads on custom bp like board designs.

[mrburnette]

OMG

Hardware junkies are doing their Happy Dance! Mark this post... in a few years some Noobie will be looking for libraries to support all those peripherals... without conflicts. Take a $2 uC and abuse it to the max.

A small economic lesson: engineering time is expensive, etched sand is a fixed cost. The more hardware components jammed into a uC, the more flexible the chip becomes to a systems engineer who would specify the part for projects. Once the software engineers become familiar with the uC, then the next design of a product can leverage that knowledge and reduce the time-to-market. The systems engineer may even specify the part for a project to get the initial product out-the-door quickly and then research a less expensive alternate uC for model #2. The entire idea about uC's that over-deliver on internal circuitry is that the usability of the uC is wide, thus knowledge and code can often be reused across a wide range of products: increasing the overall usage of a single part increases the purchase volume for the part and reduces the price-per-unit.

Sadly, I did not see a dual-core in this uC. This means that "joe-user" would likely fumble with traditional Arduino coding skills trying to drive her/his entire program (sketch) from within the single loop() paradigm. Time to start becoming familiar with a decent RTOS.

Ray

[stevestrong]

Due to the limited integrated FLASH and instruction RAM size I assume this IC is targeting only specific simple applications which run (fast) in a simple main loop.

[heisan]

Due to the limited integrated FLASH and instruction RAM size I assume this IC is targeting only specific simple applications which run (fast) in a simple main loop.

To me it looks bigger than that. I think the internal flash is meant mainly for a bootloader. Main memory is supposed to be on QSPI bus. It has MMU, so you can do demand paging, or it also supports execute in place from QSPI memory.

[Squonk42]

Due to the limited integrated FLASH and instruction RAM size I assume this IC is targeting only specific simple applications which run (fast) in a simple main loop.

To me it looks bigger than that. I think the internal flash is meant mainly for a bootloader. Main memory is supposed to be on QSPI bus. It has MMU, so you can do demand paging, or it also supports execute in place from QSPI memory.

Where do you see that it has MMU? The only one I can find is in "Community" and "Communication", I can only see the standard MPU.

[heisan]

Where do you see that it has MMU? The only one I can find is in "Community" and "Communication", I can only see the standard MPU.

My bad. Misread the spec sheet . Will need to manually load hot tasks into ITCM RAM then. So would be limited to ucLinux instead of a full Linux kernel .

[ag123]

wow, seem like a good next thing to do
this thing is pretty much a *superscalar* processor so it is *fast* (maybe a lot faster than even m4)
my thoughts however are that rather than to use this like a 'blue pill' like board (the form factor is attractive).
is to use a larger board and have either additional sram or even sdram (this is much cheaper and you can get (hundreds of) megabytes on a chip) and even additional flash (this might be in the form of an sd card)
it would be a RPi 'competitor' maybe not directly in terms of speed but that in both the combination of speed and a large number of IO pins
there used to be a generation of mobile phones and organizers that runs the 'pocket pc' type of os, this would pretty much live up to that less that MMU

to have all the peripherals on the pins i'd guess that board would need to be a Zx series to offer that much pinouts, 5x SPI + all the other peripherals, STM probably run out of alphabets (beyond Z) for the chip SKU