Atollic TrueSTUDIO Lite: Is this a misprint?

[stm239955_st]

Posted on October 09, 2011 at 18:32

Hello everybody,

Potential new user of ST chips here, feeling somewhat silly for asking this. However: Is the information provided by Atollic in their feature comparison document, v2.2, accurate? I am in particular referring to the number of code breakpoints supported, stated as being One (yes, 0x01). Surely this must be a mistake? My reason for asking is that some time ago I started development on an educational/hobbyist application, which - if successful - will eventually be released to the world as open source code. Originally I had started on the PIC32MX platform, yet some kind soul suggested I changed that decision in time, lest I get locked into the products from a single manufacturer. Additionally I was informed that the other manufacturers were also keen on getting hobbyists to use their evaluation boards, once they saw the success of projects like Arduino (made possible in part due to AVR Studio). However right now I am worried someone has been pulling my leg, and that I may just be making embarrassing noises/wasting your collective time. As far as I can tell, nobody makes a freely available tool chain for any of their Cortex M3 or M4 MCUs, which allow full development of code. All the toolchains have more or less ridiculous restrictions applied, never mind the advanced code management features of the professional packages. I mean, *one* breakpoint?! Somebody please tell me I am wrong. Kind regards from Denmark. Frank. #cheapskate #crippleware

[rmteo]

Posted on October 09, 2011 at 19:15

Looks right to me.

[Andrew Neil]

Posted on October 09, 2011 at 20:07

Try a Search of the forum - this kind of thing has been mentioned before!

The Atolic Lite does seeme to be very heavily crippled in a number of ways; eg, it also can't do printf, IIRC. At that doesn't just mean that printf is not provided, so you have to write your own - they have totally locked-out the possibility of using pritntf at all!

[stm239955_st]

Posted on October 09, 2011 at 20:49

Thanks for the replies, gentlemen.

Andrew, I did try to search for this, but most of my hits just had 'DispForm.aspx' as the subject line to the results, making it rather difficult to weed out the relevant postings. I did manage to find this, though: ''I also have this doubt. Real time debugging ( non intrusive) is very important for software developement. There is no debug restrictions for Lite Version acording to Atollic True Studio documentation. Maybe there is diference between JTAG and SWD (ST LINK) interface'' which made me wonder about the feature comparison document. Sorry about the noise. It seems I am not among the intended audience for the Discovery boards, a shame really. The 'F4 looks very interesting for my project, to say the least. They have even had the foresight of adding a bunch of ground pins, helping to reduce issues with ground bounce and cross talk (maybe). Take care. Frank.

[Tesla DeLorean]

Posted on October 09, 2011 at 20:57

As far as I can tell, nobody makes a freely available tool chain for any of their Cortex M3 or M4 MCUs, which allow full development of code.

Because, surprisingly, there is no money in doing so, and you can't make up for it in volume. WinARM builds code with GCC, works fine, debugging might be fun. There's CooCox, might be worth a look. How many breakpoints do you actually need? If you need half a dozen, there are probably better ways of attacking the problem. Heck, Rowley is only $150 for a personal copy, it's not as if you're talking $2500-5000 a seat requirement. A Keil evaluation will get you to 32KB, there are probably 128, 256 or 512KB variants, depending on what you're real needs are. Although I could see a market of a $25 Mix/Zorland type C compiler/debugger for ARM, if the market wasn't a lot of wet blankets dragging on support for elementary programming and hand holding. It worked in the 80's in large part because people would read books and manuals to figure things out, and that calling tech support was time consuming, and phone calls cost real money. http://www.rowley.co.uk/arm/ Atollic's lack of printf/scanf, hex generation and projects seem like a massive own-goal in terms of providing a viable evaluation platform. I'm not sure I'd count MIPS out as a valid architecture, it's in more things that you think, it was a smart choice for Microchip, who clearly didn't want a me-too ARM offering. Again, things like GCC exist for those who can't spend any money on supporting the tool-chain developers.

[infoinfo980]

Posted on October 09, 2011 at 23:01

Free toolchains exist (minus the debugging hardware) and don't need to be crippled, but you need to be prepared to invest time and effort in getting it all working. Unfortunately exact versions of tools are important in getting it all working.

I'm currently using, all on Windows 7 x64:CodeSourcery G++ Lite (2010q1 release - later versions have a critical gdb server bug)Eclipse Helios 3.6 w/CDTGNU Arm Eclipse pluginOlimex ARM-USB-TINY-H hardware debuggerOpenOCD (0.5.0-dev-00852) running under CygwinThis setup works perfectly for me. Programming, single-step debugging including interrupt handlers and memory inspection is seamlessly integrated into Eclipse. It's a little more clunky than, say, Visual Studio on the PC but not enough to annoy me.In addition to the toolchains you'll need to become more familiar than you'd want to be with things like linker scripts and the vital startup assembly code and system initialisation 'boiler-plates' that every program you write will need.

[stm239955_st]

Posted on October 10, 2011 at 07:18

Clive,

Maybe I didn't express myself clearly enough. My mistake was apparently to believe MCU manufacturers beyond Microchip and Atmel were interested in having hobbyists use their products. My sources claimed 'some people' were somewhat unhappy about these two corporations 'stealing all the thunder'. Some background may be in order: I am working on an Open Source project, which - among other things - is intended to lower the bar for entry to hobbyists and experimenters in using a MCU in their DIY projects. Here in particular meaning radio amateurs, 'hams', who wish to integrate an advanced MCU in their DIY radio projects of all levels of complexity. This without having to start from scratch when it comes to programming the MCU. At one end of the complexity scale one use case would be to buy a pre-programmed MCU board with the key digital components, then build the required analog circuits around it. (Part of my project is also to suggest relevant building blocks for the other parts of the radio, meaning the analog parts.) Another use case would be someone interested in SDR, Software Defined Radio, helping with debugging or expanding the DSP algorithms in the core project. For a project like this to succeed, or indeed for any hardware based Open Source project to 'go viral', the following items need to be available: *) Readily available and fairly inexpensive prototyping boards, with all or at least most of the MCU I/O pins brought out to 0.1'' headers. (Yes, I can solder SMD parts. This is for the benefit of other users, who may be less experienced.) Hardware debugger support a requirement.  - No PCI-E nor silly micro connectors accepted.  - No overpriced 'evaluation boards' with tons of integrated peripherals, which are worse than useless for this application. *) Unrestricted and free (as in beer) tool chain. Meaning hit download, install, compile my code, program, go! I don't care about the politics, can't use those for anything. If a manufacturer wants hobbyists to use their products for 'serious' projects, like Arduino, then said manufacturer need to put their wallet where their mouth is. Where is the download link, please?  - Yes, you can roll your own tool chain from Open Source tools, as Andy demonstrates. But if a single hobbyist can do it, then why can't the major manufacturers do the same, complete with ironing out any bugs and kinks, and integrating their own debugger?  - I can afford the commercial packages, but that is completely irrelevant for my project. Asking people to shell out hundreds of (your favorite currency) before they can participate in development is a complete non-starter.  - With 30 years of programming experience I am emphatically not going to accept an artificial limit of a single breakpoint. I would look like a fool if I recommended such a package for debugging in a multitasking/multithreaded environment. I might accept 5 or perhaps even 3 breakpoints at a stretch. But *one*?! A friend of mine did a virtual :headdesk: when I told him… ---- I eventually disqualified Microchip, because:  - No obvious upgrade path if the project needs more oomph for DSP. (Cortex-M4 looks *really* tasty in that department…)  - No inexpensive boards with integrated hardware debugger that I could find, you need to buy the (not too dear, granted) PICKit 3 for that.  - While I didn't do a formal analysis, it seems the ARM architecture on average does more useful work per clock cycle, compared to the MIPS architecture. This counts double for the CM4 parts with dual access SRAM banks for DSP data arrays. Additionally the ARM MCUs are generally available at higher clock speeds as well.  - The free tool chain from Microchip doesn't support optimization, you have to pay for that. Andy, I appreciate the hints, and I will be saving them for a rainy day. Unfortunately, as you may appreciate now, I don't really have a choice. I will probably have to wait and see where Atmel takes AVR Studio when it comes to support for their SAM3 parts, then hope somebody fabricates a suitable prototyping board. Thank you for your time, all. Frank.

[Andrew Neil]

Posted on October 10, 2011 at 08:31

Why do electronics hobbyists have to be such cheapskates?

A woodwork hobbyist would need to spend money to buy woodworking tools; Amateur musicians pay significant amounts of money for their instruments - easily comparable to the cost of a Keil licence etc, etc,...

Edit

: In fact, ''ham'' radio equipment isn't cheap - is it?

[Andrew Neil]

Posted on October 10, 2011 at 08:59

''MCU manufacturers beyond Microchip and Atmel''

  By ''Atmel'' I take it you mean Atmel's AVR specifically? I don't think that Atmel provide free tools for their 8051 or ARM lines - do they? And there's the key: PIC and AVR are both proprietary, single-source architectures. Therefore, tools for these architectures are useless for any competitors products;  so giving them away free just increases the lock-in and contributes directly to sales. Because ARM (and 8051) is a common architecture across all manufacturers, there would be no way for, say, ST to give away a free STM32 toolset and prevent it from being used with other manufacturer's Cortex-M3 chips. The only differentiator is the debug interface - which is exactly why that's what you have to pay for even in the toolsets using the ''free'' GCC compiler and ''free'' Eclipse-based IDE. See:

http://www.8052.com/forumchat/read/183310

A few ARM manufacturers have special arrangements; eg,

• TI have a free CCS licence that's locked to a TI Stellaris devkit;
• NXP have a deal with CodeRed for the LPCXpresso boards.

If you want an example of an ''easy access'' route for hobbyists, take a look at

http://www.mbed.org

See also:

http://www.8052.com/forumchat/read/183309

[Andrew Neil]

Posted on October 10, 2011 at 10:33

''radio amateurs, 'hams', who wish to integrate an advanced MCU in their DIY radio projects of all levels of complexity. This without having to start from scratch when it comes to programming the MCU''

Why this assumption that one should be able to just drop in an MCU without any preparation? Surely, the choice is the same as with the radio gear:

1. You build it from scratch.

   This means that you have to invest considerable time (and money) in learning electronics and, in particular, RF design, and acquiring the necessary skills, tools, and equipment.

2. You buy it ready-made

   If you don't want to bother with having to learn all that stuff, and buy all that kit.

Why should working with microcontrollers be any different.

Of course, you have a laudible goal - I just don't get this assumption that microcontrollers should somehow be special in being ''free'' of leanring & investment.