Space Borg

Technology with a twist (or two)

[dids]

TI EvalBot and Windows, Eclipse style

[Work in progress]

This guide is based on Brendan Robert's guide for linux which can be found here.

I finally got my Evalbot from Texas Instruments, only took them over six months to ship it. Anyway, after I got it, I started researching for examples, guides, tutorials and most importantly: the best IDE that I could use.

Documentation regarding the board is pretty scarce and the most IDE's out there have all sorts of limitations and bloatware.

This is when I decided that I would go throug the minimal steps of setting up Eclipse as your primary IDE on Windows, so you could easily install and configure a proper development environment and get started by writing your first HelloWorld project.

Eclipse CDT - Your first task is to download Eclipse CDT. Unzip it somewhere safe.

StellarisWare - While this step is optional, I highly recommend downloading the example sources from StellarisWare, as they contain libraries (and sample code) that'll prove very useful.

CodeSourcery G++ Lite - This is the toolchain, so you'll definitely be needing this one. Just grab the Windows installer and install it.

GNU ARM Eclipse-plugin - Eclipse wouldn't know what to do with you projects without this. Install the plugin according to the instructions, just make sure you do it the recommended way!

LMFlash - We need to be able to download the application to the board.

After all this is done, it's time to create a new C project in Eclipse. Remember to choose ARM Windows GCC (Sourcery G++ Lite) when creating the project.

We'll be using Martin Hubacek's Nunchuk library for this HelloWorld project, so download that here. After unzipping it, copy the following files to the root of your HelloWorld project directory:

I2Clib.c

I2Clib.h

oled.c

oled.h

startup_gcc.c

main.ld

You're almost there! We will now configure all the project settings so the project knows where everything is and will compile without a fuss.

Open up your project properties and browse to C/C++ Build > Settings, change the following (or make sure they're correct):

In ARM Sourcery Windows GCC Assembler -> Miscellaneous

Specify this for Other flags option: -c -fmessage-length=0

In ARM Sourcery Windows GCC C Compiler -> Directories, the following include paths (-I) should be listed

StellarisWare

StellarisWare/driverlib

StellarisWare/utils

In ARM Sourcery Windows GCC C Linker -> General

For script file (-T), specify the main.ld that you copied from Martin's project

Check Remove unused sections (-Xlinker --gc-sections)

Uncheck everything else

In ARM Sourcery Windows GCC C Linker -> Libraries

Add "driver" under the top libraries (-l) section

Add StellarisWare and StellarisWare/driverlib/gcc to library search path (-L)

In ARM Sourcery Windows GCC C Linker -> Miscellaneous

Set other flags option to this: --entry ResetISR

Add a new C file to your project, call it something like HelloWorld.c and paste the following in it:

#include "inc/lm3s9b92.h" #include "inc/hw_i2c.h" #include "oled.h" void SysTickHandler() {} int main(void) {     init_display();     dispString("Hello World!", 0, 0);     while (1)     {     } }

If it compiles without errors, that's great news!

The final and most important step is the flashing part. We'll be using LMFlash to download the application to the Evalbot. Start off by creating a new text file called manual_build.bat and paste the following in it:

@echo off echo Ready to begin flashing. echo. pause echo. cd Debug arm-none-eabi-objcopy -O binary HelloBot.elf HelloBot.bin "C:\Program Files\Texas Instruments\Stellaris\LM Flash Programmer\LMFlash.exe" --quick-set=ek-lm3s9b92 --verify --reset "%CD%\HelloBot.bin" echo. echo. echo Flashing done! echo. pause

Now, of course, you do need to change the paths to fit your environment, including HelloBot.* and the path to LMFlash.exe.

This batch file is mainly useful for problematic flashing, hence the manual-part.

Create a similar batch file to be able to flash it after compiling, automatically. Call it something like build.bat:

@echo off cd Debug arm-none-eabi-objcopy -O binary HelloBot.elf HelloBot.bin "C:\Program Files\Texas Instruments\Stellaris\LM Flash Programmer\LMFlash.exe" --quick-set=ek-lm3s9b92 --verify --reset "%CD%\HelloBot.bin"

OPTIONAL: If you want to, you can just add the full path to the build.bat file to Eclipse projects post-build, so after building it'll automatically start flashing, which is what I like to do.

You can now plug the Evalbot in with the USB cable provided (use the ICSI port on the board), turn the power on (it might require battery power even though it's plugged in) and install the drivers. You can find the drivers in the StellarisWare directory.

Since you've propably already built the HelloWorld project, there should be a Debug directory containing the ELF file. If so, go ahead and run manual_build.bat and see the magic happen.

If all goes well you've got your first compiled application running on your Evalbot.

Congratulations!

EDIT: Still looking at debugging options, might be able to use arm-none-eabi-gdb --eval-command "target remote COM1" to connect to the board and debug, but not sure yet, will keep trying.

[dids]

Stellaris Evalbot. <3

[dids]

Heck, even though it's 10yo, it's still a dual-core!

[dids]

It's an $800 SBC (Single Board Computer)

[dids]

Damn customs!

[dids]

Chessbot's got moves!

Sorry about the crappy quality, especially with the LCD, had the contrast set too high so the camera picked it up a bit too well.

Basically, I'm running the chess app on my PC while transmitting data over bluetooth to the bot (currently just servo positions and moves), just to eliminate some of the wiring.

Will keep adding new features as soon as I can. :-)

[dids]

First build is done (without gripper), though I think I need better servos. :-/

[dids]

An update on the Chessbot

[dids]

Chessbot

It has begun.

An initial AutoCAD model of the body, think of it as a mockup, though keep in mind I've never even touched AutoCAD let alone know how to use it.

The gripper is still the toughest one to design, so this is a very rough example of what I'm trying to achieve. Still going over a few ideas. :-)

[dids]

Frank's alive, almost!

So, I've been working my ass off on the robot project (dubbed Frank or F.R.A.N.K.), both hardware AND software wise.

Here's a few shots of what it looks like at the moment.

The beautiful case:

And the boring bottom:

Back panel has the interfaces for power and programming:

The innards (not connected yet, some of the guts still missing):

Ton of new stuff coming, like for example a remote controlled wireless camera and a remote controlled laser. Imagine that, a fricking laser that you can move around. ;-)

[dids]

The Tube Clock

Work in progress.

[dids]

Quick tip: Using lynx to "dump" files

I needed to transfer Server 2008 R2 (x64) installation ISO to one of my servers, but SCP/SFTP kept failing (started all over again after around 1GB) so I spent hours searching for a way to get the ISO on the box.

I finally found a way: lynx. Now, this isn't as straightforward as it would seem, because by default MSDN doesn't support wget or lynx.

Basically you need the full URL (the one with the DLL in the uri) and you need to dump that URL directly in to a file, like this:

lynx -dump "url" > filename.iso

Thank you, lynx.

[dids]

Quick tip: Installing Windows 7 from a USB drive

I've been in various situations where I had no access to a CD/DVD-drive, so the only alternative was to do a USB install.

Here's a couple of different scenarios and solutions to them.

1. Regular ol' USB install Let's assume you've got a 4Gb USB drive, as that'll do just fine for this task. Plug the drive to a spare box running (preferrably) Vista/Win7, open up the command prompt and fire up diskpart.

Run the following commands in diskpart, while making sure that the disk # you use is the USB drive you've hooked up to your computer: list disk select disk 1 clean create partition primary active format fs=fat32 quick assign exit

You're almost done! Now just copy over everything from the Windows 7 install media and the USB drive is ready to go. Easy, eh? :-)

2. Installing using two USB drives Now, while that title may sound weird, frustrating or even stupid, this is actually the way I do it most of the time.

The trick here is that you have a smaller USB drive (1Gb in my case) and a bigger drive (2-4Gb), so you'll be booting off of the small drive while switching to the bigger drive on the fly.

Sure, you could just use the bigger drive to do the install with, but in my case I couldn't dedicate that drive just for an installation like this.

With the smaller drive plugged in to your computer, do the same diskpart steps as in the first segment. Now, obviously you can't copy all of the Windows 7 installation files to the disk, since there's not enough space, so copy everything except the sources folder to the drive instead.

Create a folder called sources on the smaller drive and from the Windows 7 installation files, dive in to the sources folder and select every file & folder except Install.WIM. Copy those over to the smaller drive and we can move on.

Plug the bigger drive to your box and copy all the installation files (yes, all of them) to the big drive in a folder, just name it something simple like InstallWin7.

Boot from the small drive, select your language and click the Repair option in the lower left corner of the installer. It'll propably tell you that it can or can't fix your startup problems, just ignore those and click cancel a few times until you see a screen that has multiple options, Command Prompt being one of them.

Fire that up and find your bigger drive (in my case it's almost always F:) and do the following:

cd InstallWin7 cd sources setup

Ta-da! You've now booted from one drive while actually launching the setup from another drive, congrats.

[dids]

Snow Leopard on a Hackintosh (on the D945GCLF2D board)

THIS GUIDE HAS BEEN UPDATED TO WORK WITH KERNEL 10.6.3 (Atom-enabled)

So yeah, I finally got Snow Leopard to run pretty much perfectly on my Intel D945GCLF2D motherboard.

The reason I'm using this particular board is because it's cheap as fuck AND it packs a dual-core Atom chip in it, so it's good enough for coding and shit.

This article will explain the process of getting SL (Snow Leopard) installed with this specific board, though keep in mind that these instructions are somewhat universal and they just might work with your hardware.

Keep in mind that you NEED to have Hyperthreading disabled in order to boot OS X. At the time of writing this, I haven't figured out how to get it enabled without having OS X kernel panic. (UPDATE: Hyperthreading works, check at the bottom)

Now for the fun part: I take no responsibility over anything that I'm about to explain in this article. If you're doing this, I hope you buy the retail install media, since they're just a few bucks, I mean come on? Apple also holds the copyrights to their stuff, so keeping all that in mind, let's continue.

First step (optional, see below)

The point of this part is to get a working OS X installation, this time without a DVD/CD-drive, so that we can create the installer for your specific hardware. If you happen to have a working Snow Leopard (heck, 10.4 might even cut it) installation, you can use that to prepare the installer, just download the zip file and proceed with the next part.

First you need to have a working OS X installation somewhere, I had Kalyway 10.6.2 (AMD/Intel) running on the same system that I was going to install SL on, so this was no biggie. All you had to do with that particular distro is choose GMA950 during install, remove all audio and all wifi while also choosing the TimeMachineFix patch.

What you need to do, is have the SL installation media in .dmg format since we'll be modifying that in a bit.

You'll also need a USB drive as we'll be using that as the installer (I used a 1TB external USB drive, heh).

Go in to Disk Utility, partition/format your USB drive and name it something like SLInstall (no spaces, please) and right-click on the new partition -> Restore the SL installation media .dmg file. Now, be sure that the erase option is unchecked, this is important because otherwise it'll change the partition label and we don't want that.

Go ahead and download this zip file and save/transfer it to the computer that you're running Leopard on, it contains everything that you need to get started.

Creating the installation media

Now we're ready to create the installer. Inside the zip file you'll find the Misc folder and the MacLoader folder. Open the MacLoader folder and if you happen to have any extra kexts that you need, place them in the kexts folder.

Now double-click 2_SnowLeoInstaller and follow the prompts. For the path just enter SLInstall, if that's how you named the USB drive partition.

Note: Copy the insides of the zip file (Misc and MacLoader folders) to a safe location, for example to a network share that you can easily access or to a second USB drive, you'll need these files in a bit.

Installing

Reboot the computer and make sure that you've got the USB drive as the 1st boot device in the BIOS.

While booting from the USB drive, just let it boot normally in to the language selection screen.

You need to open the disk utility and format/partition your primary HDD as a GUID partition, otherwise SL won't let you install.

Proceed with the install as you normally would, it's pretty straightforward.

After the installation is finished it's time for a reboot. While rebooting you'll see the Chameleon screen, be sure to press a key to interrupt the automatic booting and select your newly created Snow Leopard installation (not the installer, but the actual installation partition) and boot in to that.

Post-install

Alright, we're almost done! Copy back over the Misc and MacLoader folders from your backup location to the newly installed SL box.

Copy the kexts from the MacLoader/Kexts folder to /System/Library/Extensions (remember to copy Misc/SleepEnabler.kext to the same place aswell) and fire up terminal and do kextcache -v 1 -a i386 -a x86_64 -m /System/Library/Caches/com.apple.kext.caches/Startup/Extensions.mkext /System/Library/Extensions and you've got all the necessary kexts enabled.

Open up MacLoader and double-click 3_MacLoader and follow the prompts. Once done, do NOT reboot.

Open up terminal and execute sudo su. Now you need to mount the EFI partition in order to fix the resolution problem (it's locked in to 1024x768 by default). You do this by creating a temporary folder (I named mine EFI, for simplicity) and mounting the EFI partition to that. You can find your EFI partition by executing: diskutil info / | grep Identifier If diskutil comes up with for example disk1s2, then you'd mount disk1s1 instead: mount_hfs /dev/disk1s1 /Volumes/EFI

After this you just need to copy over the DSDT.aml file from the Misc folder to the /Volumes/EFI/Extra folder and open up com.apple.Boot.plist (still in the EFI directory) in an editor of your choice.

Make sure that the Kernel Flags section has the argument arch=i386 appended to it, if not, add it.

Unmount the SLInstall volume (the USB drive) and unplug it.

Reboot and you should be automatically booted in to Snow Leopard with working audio, networking and have the ability to change the resolution.

Final words

There's still a few quirks here and there, mainly the Hyperthreading which has to be disabled in order for SL to boot. Hopefully someone will figure out how to fix that, since turning off HT is like running a dual-core system with just one core enabled (as far as I understand it).

Any and all improvements to this guide are more than welcome, so fire away!

PS. If you happen to have problems changing the resolution, mount the EFI partition again and modify the com.apple.Boot.plist file in the Extra folder. Add a   by the name of Graphics Mode and a   that looks something like 1920x1200x32 (so widthxheightxdepth), save and reboot.

Update!

Just noticed something, Apple removed support for Atom devices in the latest kernel, so yeah, you're pretty much f*cked if you try to install the latest updates. However, I did manage to fix the situation, that's what this last bit is all about. :-)

Important: After installing a new kernel update (ie. 10.6.3) do NOT press "Restart", but instead fire up Terminal and upgrade the kernel (the whole copy-over-sleepenabler shebang) like normal, you can find the commands at the bottom.

I've updated the zip file, which now contains all the files you'll be needing, mainly just the patched kernel.

Now, move mach_kernel on the / of your main HDD to for example mach_kernel_old. Copy over mach_kernel from the Misc folder to the root of your HDD. Fire up Terminal and do chmod 644 /mach_kernel && chown 0:0 /mach_kernel and you're almost done.

Leave terminal open and copy the SleepEnabler.kext from the Misc folder to the /System/Library/Extensions folder (overwrite/replace if necessary). While still in the terminal, do kextcache -v 1 -a i386 -a x86_64 -m /System/Library/Caches/com.apple.kext.caches/Startup/Extensions.mkext /System/Library/Extensions and wait a while.

You can update to 10.6.2 now (the patched kernel, mach_kernel, is the patched 10.6.2 kernel) via Software update and reboot. Now, I messed up the process, so you might have to update first, then decline the reboot and then copy over the patched kernel. In my case, I had to boot off of the installation media and copy over the kernel that way, works both ways, I guess.

UPDATE: Hyperthreading can now be enabled, works like a charm!

10.6.3 was released!

Getting 10.6.3 wasn't as tedious as some would think, it was actually pretty straightforward.

I've updated to ZIP file to contain a hacked (Atom-compatible) kernel for 10.6.2, 10.6.3 and a special 64-bit version of 10.6.3, but here's a quick how to on the upgrade:

sudo su cd / mv mach_kernel mach_kernel_old mv ~/Desktop/mach_kernel / (wherever your downloaded mach_kernel is) chmod 644 /mach_kernel chown 0:0 /mach_kernel

The version of SleepEnabler.kext included in the Misc folder in the ZIP file is compatible with 10.6.X, so there isn't really any need to update that. Though don't take my word for it, might be a good idea to just copy it over anyway and rebuild the kext cache, especially when doing a kernel update.

If you're having trouble booting, you can always boot up the Snow Leopard installer, fire up the terminal and restore the kernel, though it might be worth checking out Extra/com.apple.Boot.plist under your EFI partition (mount it to a temp dir, as you did previously).

At the time of writing this, I'm running 10.6.3 kernel, but Hyperthreading does NOT work, even though it did on 10.6.2. I'll have another go at it later tonight and post an update if I get it going.

[dids]

What? Boxee Apps?!?

Yes, what you're seeing is true, we've got our own Boxee Application Repository.

At the time of writing this, there's nothing there yet, so don't bother trying it out.

In the next few days, we'll be working rigorously on an application on called Boxify.

Boxify is for all you couch potatoes out there (including ourselves) who'd simply LOVE to access Spotify directly from Boxee.

Keep an eye out for posts regarding Boxify, who knows, it might be out soon.

[dids]

F.R.A.N.K

Woo! It's time to unravel the mysteries of an unknown yet technologically advanced robot, but you can just call it Frank.

Yeah yeah, the image is upside down, I know, but I blame my iPhone.

Dry run before the actual test.

First (semi-succesfull) test!

[dids]

Christmas came early this year!

Just when I thought things couldn't get any better, I noticed I got something in the mail: Nixie tube kit AND a ton of stuff from Sparkfun, woohoo!

The Nixie tube kit (ArduiNIX from RobotPirate):

The nixie tube kit components, brr, that's a lot of soldering:

Built a small prototype using some of the stuff from Sparkfun (gearshaft motor kit + wheels + ball bearing "wheel"):

Bottom shot, closer look at the gears:

The ball bearing "wheel":