STM32 F4 Cortex M4 devices

[jezc]

Hi Mark,

ST have (finally) launched their Cortex-M4 range - I'm hoping to get a dev kit soon & see how compatible all the peripherals etc are.

Obviously the core will run Cortex-M3 code directly but I'm not sure how easy t will be to change the simulator over to use the new device.

Is there any information on this please?

Cheers,
    Jez

[mark]

Hi Jez

Yes, there is a lot of activity in the Cortex M4 market at the moment.
I didn't get the chance to look at the ST device in detail but am assuming that there will be a high compatibility with the peripherals in comparison to the STM32 Cortex M3 range. The only way to be sure is to work through each one though..

Regards

Mark

[mark]

Hi Jez

Please note that the STM32F4xx version (including F1xx, F2xx and F4xx) is now moving along rapidly. There is a developer's document here which may give some more insight into the work: http://www.utasker.com/docs/STM32/uTaskerV1.4_STM32_Developers_Document.pdf

The simulator now supports a large range of parts (F1xx and F4xx should be complete in all packages) and I am working through peripherals on the F4xx board(s) and checking that things are otherwise compatible with the original F1xx devices.

Although the STM32 project has been fairly quiet I do believe that there is a lot of potential; just the fact that ARM reports that almost one in 2 Cortex M3/M4 parts delivered come from ST-Micro shows that the devices are certainly used in many projects. That means that I am pushing to get a good package together with some flexible development environment support - I am looking at GCC, Rowley Crossworks, Atollic, IAR and Keil (with possibly Raisonance and Tasking too, but the last ones are still on the "very tentative" list).

Regards

Mark

[jezc]

Hi Mark,

Thanks for the update - I've got several different F1/F2 dev boards already but only the discovery board for the F4 so far...

We've used the F1 parts for several years already (but only the F103/F105 parts i.e. no ethernet) & generally found them to be a good workhorse for jobs with low I/O requirements & features. So we're very keen to compare the newer parts to our other higher-end parts (NXP/Freescale etc).

When do you hope to have a full release (or even a reasonably complete beta)?

Cheers,
Jez

[mark]

Hi Jez

I am pushing this project and will be making a version fairly shortly which will include quite a lot, but not all.

- The simulator can already adapt itself to all F1 and F4 parts (all packages) - I expect that the F2 pins outs may be the same as F4 but I didn't actually check anything yet.

- The F1 and F4 discovery boards are essentially supported.
- The STM32F107 eval board has been supported for quite a long time now (that means the basics with USARTs, Ethernet, Timers etc.) but the project is now fully up-to-date with the most recent developments and supports Rowley Crossworks, IAR, Keil, Atollic, and stand-alone GCC; these are essentially ready...
- The simulator is presently being used to test the STM32F407 eval board project (adapting peripheral drivers where things have changed between F1 and F2/F4) but I haven't received the HW just yet (it should be underway). Once it is here things can be tuned that need adapting, which isn't usually a huge task.

Probably the next version will not yet contain serial and boot loader projects though. It will also have neither USB nor SDIO (but SD cards via SPI is included). The F4 has quite a lot of new features, like high speed USB and a camera interface. The camera interface is a bit exotic but the SDIO and high speed USB (it also has full speed on a second interface) are two things that should be added asap.

It is a shame that the STM32 project hasn't been very popular up to now and so hasn't been advanced much until recently. But it does have the potential to take off once the basics are all there. It is important to get the next version available as soon as possible (with support for more chips and more development environments) so that it can be used more intensively. The developer's document already shows that the developer's interface makes the device quite easy to control (not forgetting that it is also possible to move projects between different manufacturers and chips with ease) and once this is complemented by an up-to-date tutorial and some guides to starting with the project in the various IDEs (maybe short video guides) I think that new users will in fact be positively surprised at how easy it is to configure for any chip type and start working on real applications (parallel to the simulator of course for maximum efficiency).

A release version will also follow as soon as possible (this is also valid for the freescale Kinetis, which is the main STM32F4xx rival at the moment ;-) so the choice is also there!

Even when not all peripherals are supported, the project will have enough features to be useful in many cases. I am determined to get the Cortex M4 projects extended in terms of capabilities (mainly peripheral support) so if there are project needing anything that is missing these can rely on receiving the peripherals (with simulation support, documentation and support) with short delays.

Regards

Mark

P.S. The only thing about the F2/F4 parts that bugs me at the moment is the large granularity (and non-uniform section size) of the internal Flash. Rather than having 1k or 2k sections as in the F1 parts they have 16k boot blocks followed by 64k and then several 128k sections. This is less than ideal for internal file and parameter systems so the large granularity mode will have to be used (as tends to be used in the NXP project). This is a bit of a shame since it is so much less efficient and fiddly in comparison to small granularity flash....

[jezc]

Hi Mark,

That sounds like very good progress - from memory the pinouts between F2 and F4 are very similar, there's an RFU pin on some of the F2 parts that needs to be tied to Vdd on the F4 parts but other than that all should be the same.

I've historically found some of the ST peripherals to be a bit limited compared to other vendors offerings though they do seem to be working to address this - at last!

With the Ti and NXP M4 parts sampling soon (some NXP flashless parts are available now in BGA packages) it's certainly hotting up in the M4 arena. It's almost like we're spoilt for choice once more :-)

Cheers,
Jez

[mhoneywill]

Hi Mark,

These STM F407 devices look very interesting, 6 x Uart with RS485 support (At least an interrupt that tells you when the transmit buffer is empty), USB Device, Ethernet.

The Eval board looks a good price to £10ish from Farnell

One question you might know the answer to mark as you have been looking at these chips in detail. The timer/conters seem to support a quadrature encoders as far as I can tell you can read up to 6 quadrature encoders on TMR1,2,3,4,5,8 do you think that is correct? Shame as we wanted to read 8 encoders.

These chips look as good as the Kinetis K60

Cheers

Martin

[mark]

Hi Martin

I don't know the timers that well but I have also understood that those timer channels can be used for quadrature encoding.

The Discovery board is very cheap - unlike the STM32F100 discovery board the STM32F407 one has a powerful chip (168MHz) on it with lots of memory (1Meg Flash and 192K SRAM if remember correctly). The board doesn't have any perphersl connectors apart from USB and breakout headers so needs a bit of extra work to connect to other things.

The drawback with this family in comparison to the Kinetis is that its FLASH is large granularity (mostly 128kb sector size) which makes it less attractive for storing data and operating a simple, flexible file system in. The Kinetis's external memeory interface supports SDRAM whereby the ST's interface doesn't.

Cheers

Mark

[mark]

Martin

Seeing as you are in fact quite a Stellaris fan, why not consider the Concerto: http://www.utasker.com/forum/index.php?topic=1683.0 ?

Maybe the DSP in it could be used as a multi-channel quadrature encoder (?) alongside a Cortex M3 with all standard features?

Regards

Mark

[phomann]

Hi Mark,

I'm still looking for an alternative to the LM3s9B9XX chips. The C5 errata stats that they only guarantee 100 flash writes. Totally useless for an internal file system.

I want a chip with an integrated Ethernet PHY and USB. At this stage my option seems to be the LM3S6965 with a FTDI  USB chip.

Cheers,

Peter.