Shop OBEX P1 Docs P2 Docs Learn Events
STM32 Discovery, and other 32-bit platforms? — Parallax Forums

STM32 Discovery, and other 32-bit platforms?

John A. ZoidbergJohn A. Zoidberg Posts: 514
edited 2014-08-16 21:50 in General Discussion
Hey all,

Anyone tried those low-cost STM32 Discovery evaluation boards yet? They are with the ARM processors there. Or also the mbed ones in the Sparkfun? :)

Comments

  • LeonLeon Posts: 7,620
    edited 2010-10-24 03:49
    I've got a couple of mbeds and LPCXpressos. The former takes a similar approach to the Arduino in terms of software development, with a web-based compiler. The latter uses a conventional IDE (Eclipse) and gcc compiler, and is harder to use. Both make excellent introductions to the ARM Cortex MCU.
  • Heater.Heater. Posts: 21,230
    edited 2010-10-24 04:05
    A while back I picked up a Raisonance STM32 Primer 2 dev kit.

    What a disaster that is.

    1) There is only an IDE for Windows.

    2) If you make a mistake installing the supplied USB programmer drivers they won't work, can't be made to work and are almost impossible to remove. I gave up following the insane removal instructions and had to reinstall my Windows XP. Luckily it was only running under VirtualBox.

    3) It was possible to program the Primer 1 from Linux. This possibility was removed from Primer 2. A fact I discovered only I bought the damn thing.

    The result is that I have a brick gathering dust here and Raisonance will not be getting my custom any more.

    I hope these STM32 kits you are looking at are a bit more friendly
  • John A. ZoidbergJohn A. Zoidberg Posts: 514
    edited 2010-10-24 06:00
    Heater. wrote: »
    A while back I picked up a Raisonance STM32 Primer 2 dev kit.

    What a disaster that is.

    1) There is only an IDE for Windows.

    2) If you make a mistake installing the supplied USB programmer drivers they won't work, can't be made to work and are almost impossible to remove. I gave up following the insane removal instructions and had to reinstall my Windows XP. Luckily it was only running under VirtualBox.

    3) It was possible to program the Primer 1 from Linux. This possibility was removed from Primer 2. A fact I discovered only I bought the damn thing.

    The result is that I have a brick gathering dust here and Raisonance will not be getting my custom any more.

    I hope these STM32 kits you are looking at are a bit more friendly

    I see. Well the company let one choose the compilers when he/she obtain the training kit. I guess I might choose the other ones since the Raisonance is a bit fussy. :)
  • LeonLeon Posts: 7,620
    edited 2010-10-24 06:45
    It worked OK on the Primer 1, it's probably OK on the Discovery. I found it a bit clunky compared with other ARM tools I've used.
  • John A. ZoidbergJohn A. Zoidberg Posts: 514
    edited 2010-10-24 08:24
    Hmm, I'll get the STM32 first because they are accessible here right now and I would like to get my hands dirty on the 32-bit world.

    Also, I believe these things have full floating point operations - any possibility that I could do digital filters and sound synthesis/modellings easily? :)
  • LeonLeon Posts: 7,620
    edited 2010-10-24 08:52
    None of those ARM chips have floating-point hardware, it has to be performed in software. They do have a MAC instruction for DSP operations but won't be as fast as proper DSPs as they only have one memory. Simple audio processing is feasible.
  • John A. ZoidbergJohn A. Zoidberg Posts: 514
    edited 2010-10-24 10:16
    Leon wrote: »
    None of those ARM chips have floating-point hardware, it has to be performed in software. They do have a MAC instruction for DSP operations but won't be as fast as proper DSPs as they only have one memory. Simple audio processing is feasible.

    So which of those microcontrollers in the market which contains these Floating-Point hardware? I'm pretty curious here. :)
  • LeonLeon Posts: 7,620
    edited 2010-10-24 10:34
    You need one with an ARM11 core, but they aren't available on low-cost boards. You don't need floating-point for audio processing, of course. A much simpler and cheaper option would be a Microchip dsPIC, if 16-bit fixed point is adequate. I've got a very simple dsPIC system I designed specifically for speech processing. For really high-quality audio XMOS has some nice solutions.
  • pharseidpharseid Posts: 192
    edited 2010-10-24 16:02
    Any experience with the Beagleboard? I can't think of anything to use it for, but it seems really interesting. 700 something MHz ARM, VLIW DSP, graphics coprocessor. I looked at TI's VLIW DSP's years ago, but I can't remember if this one is floating point or integer. And the ARM, of course. And I'm curious what the development software is like, you have so much going on there, getting everything working together on something, like 3D graphics which I think this is designed for, that's an interesting problem. I remember really liking TI's implementation of VLIW, it seemed more practical than Intels Itanium.

    -phar
  • LeonLeon Posts: 7,620
    edited 2010-10-24 16:16
    The OMAP chip used on the Beagleboard has a fixed-point DSP.
  • pharseidpharseid Posts: 192
    edited 2010-10-24 16:49
    You're reminding me how lazy I am. After I first noticed the Beagleboard, I downloaded the data sheet for the OMAP processors, but I don't think I ever looked at it. As I remember the TI interger VLIW DSP's, they have ops like left most ones and barrel shifters, so they probably aren't that bad doing floating point. So now I'm wondering when they do 3D graphics if they do everything in integer or FP. But as we've now established, I'm too lazy to find out which. In any case, if you really need a lot of raw horsepower, you might look at the Beagleboard.

    -phar
  • Heater.Heater. Posts: 21,230
    edited 2010-10-25 00:23
    Leon,
    The OMAP chip used on the Beagleboard has a fixed-point DSP.

    That can't be right.

    The BeaglBoard uses an OMAP3530 processor which according to the TI support forum has a hardware floating point accelerator, VFP, and a SIMD (Single Instruction Multiple Data) accelerator processor, NEON.

    These unit share the same registers so you can use:
    NEON:
    Integer and fixed point
    Single precision floating point

    or VFP:
    single or double precision floating point

    at any moment.

    See here: http://e2e.ti.com/support/dsp/omap_applications_processors/f/447/t/30265.aspx

    The IGEP board uses the same devices http://www.igep-platform.com/index.php?option=com_content&view=article&id=46&Itemid=55

    The later has an excellent temperature spec. so I'm hoping to get them into use for our companies products.
  • LeonLeon Posts: 7,620
    edited 2010-10-25 04:31
    I thought the OMAP3530 had a TMS320C64x on the chip, which is a fixed-point DSP. It might have a floating-point unit on the chip as well, but I don't think that can be used with the DSP.
  • pharseidpharseid Posts: 192
    edited 2010-10-25 05:59
    I believe NEON and VFP are extensions to the ARM. That would make sense that you do transforms in floating point and maybe lighting on the DSP. My memory is that on the TI integer VLIW DSP's you can break the carry links on the ALU's and do SIMD type computations of various different sizes, so you could do all three RGB components simultaneously. The little blurb I read on the graphics unit was pretty vague, it might not do much (or anything) computationally from the standpoint of the 3D pipeline. The intended target I think is hand-held games and phones. That's an impressive amount of computing power for a phone.

    -phar
  • John A. ZoidbergJohn A. Zoidberg Posts: 514
    edited 2010-10-25 08:20
    I have noticed some of the ARM processors which equipped with those Codecs - I guess it's probably for cellphones, gadgets and stuff which requires audio/video interfaces.

    Right now, I'm thinking of some cool ideas that I could use a cheap Cortex M3 with. :)
  • pharseidpharseid Posts: 192
    edited 2010-10-25 14:49
    If you need another cool idea, you might consider implementing a scanning tunneling microscope. Apparently, there are quite a few ad hoc designs floating around the physics community, but the only solid design I've seen comes out of Germany, seems unnecessarily expensive and requires a computer with an ISA slot. There are two super-cheap designs that use those cheap piezoelectric actuators that go into little buzzers, but one is unfinished and the other uses a storage scope. So the field really needs an inexpensive standard design, maybe something you could just plug into a USB slot.

    -phar
  • John A. ZoidbergJohn A. Zoidberg Posts: 514
    edited 2010-11-04 06:53
    I just got my STM32 Discovery training board. However, there are lack of documentation about how to activate the timers, or even general purpose I/O and stuff, unlike in those of the Microchip C Compilers.

    Anyone have any ideas on starting to work on the board other than just lighting up two LEDs? How about the PWM?
  • Heater.Heater. Posts: 21,230
    edited 2014-08-16 21:50
    I just stumbled across this ancient thread and though it deserved an update. Perhaps everyone knows this already but just for the record:

    Back then I was whining about how awful the Raisonance STM32 Primer 2 dev kit was. That thing is still at the bottom of my junk pile somewhere.

    How times have changed. I now have an STM32 F4 Discovery board and it is great. Very cheap, very easy to program from Linux.
    http://uk.farnell.com/stmicroelectronics/stm32f4discovery/stm32f407-usb-otg-discovery-kit/dp/2009276

    I also have a couple of Espruino boards. That's tiny, has and STM32 F4 and runs JavaScript. (Which also runs on the above Discovery board and many others)
    http://www.espruino.com/

    Then the is the MircoPython board. Which I have not seen yet but also looks great.
    http://micropython.org/

    These devices are pretty amazing, floating point support, tons of I/O, enough speed to run JS and Python.



    No, I'm not suggesting we all switch from Propellers to ARMs. They are different beasts. Horses for courses and all that.
Sign In or Register to comment.