STM32 F3 Discovery Board

[Rutger]

Just received an email from Digi-key they are selling a STM32 F3 Discovery Board for $ 10.90.
The board has the following features:


    STM32F303VCT6 microcontroller featuring 256 KB Flash, 48 KB RAM in an LQFP100 package
    On-board ST-LINK/V2 with selection mode switch to use the kit as a standalone ST-LINK/V2 (with SWD connector for programming and debugging)
    Board power supply: through USB bus or from an external 3 V or 5 V supply voltage
    External application power supply: 3 V and 5 V
    L3GD20, ST MEMs motion sensor, 3-axis digital output gyroscope
    LSM303DLHC, ST MEMs system-in-package featuring a 3D digital linear acceleration sensor and a 3D digital magnetic sensor
    Ten LEDs:
        LD1 (red) for 3.3 V power on
        LD2 (red/green) for USB communication
        Eight user LEDS, LD3/10 (red), LD4/9 (blue), LD5/8 (orange) and LD6/7 (green)
    Two pushbuttons (user and reset)
    USB USER with Mini-B connector
    Extension header for all LQFP100 I/Os for quick connection to prototyping board and easy probing

I thought is sounded like a good deal for the price, but hey maybe I am out of touch.
Here is the link to the digi-key website, it pre-order only right now:

http://www.digikey.com/product-detail/en/STM32F3DISCOVERY/497-13192-ND/3522185&WT.z_email=STM_STM32F30912-2_stm32f3_en

Rutger

[ptricks]

The thing about the ST boards for me has always been that way the software libraries are written, not fun at all to use.
Make sure you take  a look at the programming environment before you spend money.

[andyturk]

Well, for less than $15 you can get an F4Discovery with a 168MHz processor, a meg of flash and 192K of RAM. That's a pretty amazing deal for what you get. Personally though, I'm using a hot-rodded STM32VLDiscovery. I desoldered the original F100 mcu and replaced it with a much more serious F103. I would've used the F4Disco, but you can't plug it into a breadboard because it's got dual headers on both sides.

Here's the VLDiscovery hiding behind a bunch of wires:


The board is supported by ChibiOS (a free RTOS) out of the box and everything compiles cleanly with the Yagarto toolchain on OS X.

The F3Discovery board looks pretty good, but it has those dual headers which take all the fun out of breadboarding.

[T4P]

Well you can always do this 


Every single board ST produces actually is like a amazon kindle fire/google nexus 7 approach, the product costs more than the cost price.

Well, if you ever knew the difference between the F3 and the F4 is that the F4 is a whole new monster to the F3, it's quite literally a lotus exige S vs a bugatti veyron

It's much faster and with a FPU That type of F4 is usually Cortex-M4F

Only thing the F2 doesn't have but it's 3/4 the performance of the F4 is the FPU. That's what it lacks

Plus if you buy the disco boards the two boards it doesn't really make any sense to get the F3 if you already know the performance game, of course neither are particularly good at power consumption. Well that's what the L series is for

[poorchava]

I don't know what's the current status, but like a year and a half ago ST libraries used to suck really bigtime. For example there was no function to update PWM duty cycle. If you wanted to change the duty cycle without resorting to pure register math you would have to re-initialize whole pwm module....

[Short Circuit]

Well... software-wise, ST isn't that great, that's right.

But if you consider just the hardware, it's great value for money.
Suppose you need a gyro for a project; this board is way cheaper than a raw L3GD20 (older L3G4200 actually) breakout from sparkfun ($50).
Heck, the board is almost same price as a single piece of the controller (E9.71 vs E8.11 at Digikey).

Personally, I prefer to start projects by designing a board and work from a prototype. But I do know other engineers that start most of their projects
on one of these Discovery boards. You can't beat 10 bucks to get started on SW development rightaway, especially when there's a deadline of some sort.

[andyturk]

Not that I'm am embedded expert, but most of the OEM code I've seen looks like it was written by summer interns.  Yuck.

For an alternative, you guys should really check out ChibiOS: http://chibios.org. It's a very well implemented open-source RTOS (real time operating system) that also has a common API for hardware peripherals (e.g., USARTS, SPI, ADC, etc.). I think the guy who wrote it works for ST, so the STM8 and STM32 support is very good, but it also has ports for AVR, AT91, LPC and even MSP430.

The demo apps compile cleanly out-of-the-box with GCC with specific examples for several Olimex dev boards as well as the shipping STM Discovery boards (i.e., no F3 yet).

It's really good. Check it out!

[T4P]

Yep, ChibiOS works perfectly on the ST boards

[mrflibble]

I don't know what's the current status, but like a year and a half ago ST libraries used to suck really bigtime.

Recent measurements indicate they (ST) have a Quality Assurance department dedicated to safeguarding the Library Suckage Level.

[Badreddine]

Hi poorchava,

You can just need a training on how to use the ST library because I am using the function to update PWM duty cycle TIM_SetCompare1(), TIM_SetCompare2(), TIM_SetCompare3()....
So you can judge that ST library is not good if you don't know how to use it. Perhaps you are familiar with other libraries and ST one is new for you.

[Bored@Work]

Hi poorchava,

You can just need a training on how to use the ST library because I am using the function to update PWM duty cycle TIM_SetCompare1(), TIM_SetCompare2(), TIM_SetCompare3()....
So you can judge that ST library is not good if you don't know how to use it. Perhaps you are familiar with other libraries and ST one is new for you.

Bullshit. A library for a microcontroller that needs special training so it can be properly used is a steaming piece of crap. A library is there to simplify things. It should come with proper documentation and examples, enabling "those skilled in the art" to get started within minutes.

[KedasProbe]

My F3 arrived today. http://www.st.com/internet/com/SALES_AND_MARKETING_RESOURCES/MARKETING_COMMUNICATION/MARKETING_BROCHURE/brstm32f3.pdf
Probably going to use the F37X in a future project (with 16 bit ADC)

About learning to use this board:
Anyone knows some good DAC function generator code that can be used?
Also something like AC multimeter code would be interesting.

[RCMR]

My STM32F4 arrived yesterday and my first comment is:  Bloody USPS!

It was ordered a month ago and when it (finally) arrived, it had a little stamp on it that said "mis-sent to Thailand" -- which explains the very long delay in delivery.

My observations are:

1. if the Thai post office has  had rubber stamps made with this legend embossed upon them then they must get a lot of mail that is actually supposed to be going somewhere else.

2. it is no wonder that the USPS is in huge debt and on the verge of bankruptcy if this is the level of competence in their sorting rooms.

But that aside -- I'll be getting stuck in with some familiarisation and porting some of my code to it over the next week or two.

But, as the basis for projects, I've ordered one of those $32 F3 boards with the stacked 2.8"  touch-screen LCD.  These look like they'd be brilliant for a raft of projects around the lab and as the basis of one-off systems for clients.

[mrflibble]

So you can judge that ST library is not good if you don't know how to use it.

One can indeed judge that the ST library is not good for you if you like your libraries to come with well written documentation.

Perhaps you are familiar with other libraries and ST one is new for you.

Indeed I am familiar with other libraries and this familiarity enables me to make the qualitative statement that the aforementioned ST library sucks documentative donkeybawls.

[andyturk]

... donkeybawls.

If the datasheet sucks really hard, then the library code looks better (in comparison). E.g., TI has pretty good datasheets, so it's sometimes easier to just write the stuff yourself instead of trying to figure out the library API. ST's datasheets aren't quite as good, so it's more of a toss up.

It'd be great to have TI's doc with ST's silicon, but the world doesn't work that way. 

[poorchava]

I have an impression that ST has a general mess in their documentation. I actually happen to be ST-stuff-guy in my company and i tell you: they are a mess. Most of the documentation they have  (other than official datasheet) is a pile of copy and paste . And draft datasheets with 3/4 of the fields containing "TBD" are not strange either .

But indeed their silicon is impressive. Actually i think they produce the fastest M4 on the market right now.

And Discovery boards are nice because ST-link is now officially supported by OpenOCD which means you can program&debug the stuff from any toolchain and IDE you want.

[xDR1TeK]

Best IDE that I can tell is the Keil IDE. For the library, I modded some from ARM and some from ST. I have to make my own template for new projects. Theirs was just too large to carry around.

[McMonster]

Best IDE that I can tell is the Keil IDE. For the library, I modded some from ARM and some from ST. I have to make my own template for new projects. Theirs was just too large to carry around.

We use Keil on embedded programming course on my university and I really don't like it. Primitive, stubborn, causes problems sometimes (for example we always need a full rebuild of the code before this damned thing actually starts outputting and flashing a new binary to the micro). No code completion or as-you-write error detection (it's a major pain in the ass to use ST's peripheral and demo libraries without it), extremely stupid code indentation behaviour. Possibly some of those can be eliminated by settings, but we don't really have time to mess with anything during the lab (which itself is poorly organized) and on a computer and system profile shared by everyone. The user interface in overall feels like it has been developed well over a decade or two ago and not touched since.

I would sell my soul so that they let me just connect the board to my own laptop and use Eclipse with Linaro and OpenOCD, but they refuse every time. Maybe I'm biased a bit, but this makes Keil no less awful.

[KedasProbe]

Where can I find shared (demo) code for this board, or for any other Cortex-M4(F) board?

[alm]

Semi companies almost always offer demo code for their demo boards, for example for the STM32F3DISCOVERY (click on design support and scroll down to firmware).

[KedasProbe]

Thanks somehow I looked over that link.
any other places where you can find shared project?

[krenzo]

It looks like they're giving these away now.  "Register for your FREE KIT": http://www.st.com/internet/evalboard/product/254044.jsp

[T4P]

It looks like they're giving these away now.  "Register for your FREE KIT": http://www.st.com/internet/evalboard/product/254044.jsp

I don't see the text anywhere, mind giving me a hand?

[krenzo]

I don't see the text anywhere, mind giving me a hand?

[poptones]

OK, for free I'll take one!

How do they make money giving them away?

Volume