|TuxScreen on SourceForge||TuxScreen CVS||search the Wiki|
|Browsing -> Wiki -> Users -> [SteveMerrifield]|
|edit, info, topics, orphans, hubs and nodes, or recent changes in the Wiki||create a new user or login|
25 Apr 2002
Added a pre-compiled filesystem image and kernel to the download/microwin-touchscreen directory so those that can't/don't want to compile can try it out.
17 Apr 2002
Put together a working microwindows tree with touchscreen support and merged it into the latest buildroot. Placed it in http://tuxscreen.net/download/microwin-touchscreen/ Follow the README, and you shouldn't have any problems getting it to work.
07 Apr 2002
Finally finished off my non-US power supply. I decided against the PCB, and instead, just hacked it up using vero-board. Check out http://www.labyrinth.net.au/~steve/TuxScreen/index.html for the schematics and some photos showing the construction.
27 Mar 2002
Updated the kernel to 2.4.18-rmk2-tux1 for no reason other than to see if it worked. It does :) Microwindows with touchscreen support works just fine.
17 Mar 2002
Hmm... More than a month since the last update. Not much to add though, I've been playing with reading/writing to the GPIO pins. Also laying out a PCB for my power supply. Posted a screendump from my Fluke Scopemeter on my web page showing the 555-generated PFI signal.
16 Feb 2002
12 Feb 2002
Downloaded and incorporated Alex Holden's microwin-aph-11feb2002.tar.gz snapshot into buildroot-tux. Built cleanly - no problems. Re-flashed the ramdisk and fired up - hehe... touchscreen now works :) Not only does it work, but it's also pressure sensitive! I had to make some minor changes though - I copied all microwin/src/lib/*.so files to root/lib in order to get nano-X to run, and I created /etc/ucb1x00-tsdata with my own config data to fix the slight offset with the calibration. I used the old tuxcal program with kernel 2.4.7-blfix to work out what my calibration data should be.
07 Feb 2002
Downloaded and applied patches patch-2.4.17 and patch-2.4.17-rmk5, and incorporated this into buildroot. Applied Alex Holden's patch-2.4.17-rmk5-aph and rebuilt. No problems, flashed the kernel - boots nicely. Hacked the microwindows source to change all /dev/ucb1200 entries to /dev/ucb1x00. Rebuilt and fired up microwindows - appears to work, but the touchscreen does not function. Brightness/Contrast can be adjusted okay though. TinyTOE causes a floating point exception, so I manually created /etc/tsdata using entries obtained using kernel 2.4.7-blfix. This allows TinyTOE to run and exit without problems, but still no touchscreen.
02 Feb 2002
Hmm.... Looked at the MicroWindows source - it was trying to open /dev/ucb1200-ts. That's strange, I only have /dev/ucb1x00-ts... Oh well, added the appropriate entries to devices.txt and rebuilt. Problem fixed, nano-X now starts up and everything works as expected. I had to go back to kernel 2.4.7 though to get the touch screen to work.
After all this, I can finally run CosmicPenguin's screengrab program....
01 Feb 2002
Grabed the latest CVS buildroot and downloaded everything. Did a standard make which worked first time - flashed the kernel and fs image - works perfectly. Thanks to all involved for getting this fixed!
Rebuilt the filesystem, but this time with MicroWindows enabled. Didn't work. Needed to add a line BUILD_DIR=$(BASE_DIR)/staging_dir in makefile.microwindows to fix the problem. This has beed added to the CVS entry and it now builds cleanly.
Still having problems with the touchscreen driver though :(
31 Jan 2002
Solved the nfs mount problem. It turned out to be the way I had built busybox. Further investigation showed problems in several applets - other people have had similar problems as reported on the uClinux-dev mailing list (email@example.com) - see the "busybox for ARM" thread.
29 Jan 2002
My ethernet adapter arrived today. Plugged it in, turned it on, and it works. Setup the network parameters, and a few seconds later I'm pinging the outside world. Very cool. However, I have a problem trying to nfs mount my development machine - it doesn't work. At first, I tried mount -t nfs 192.168.0.12:/home/sjm/tux /mnt which gave me
NFS: mount program didn't pass remote address! mount failed: invalid argument
Incidently, this command works from my uCsimm/uClinux board, so I know the server is okay....
Then I had a look at /linuxrc, which uses mount -o nolock a.b.c.d:/dir /mnt, so I tried that, but this time it gave me: No such file or directory.
Yes, /mnt does exist! (755 permissions) I can ping/telnet to the server, and I can mount the exported filesystem on another machine. I have tried with 2.4.7 and 2.4.16. If anyone has any ideas I'd like to hear them please :)
AkulaSJ adds: I think it is the -o nolock part that will do the trick. Try mount -o nolock 192.168.0.12:/home/sjm/tux /mnt
Err.... No, I've already tried that :) But I managed to fix it. See above...
23 Jan 2002
The last couple of days have been spent playing with nano-X and MicroWindows - building my own GUI and writing simple apps. I've also spent some time playing with TinyWidgets - it's pretty cool, and offers some nice features. I had some initial problems getting it to link though - the linker was picking up the libc6 version of the jpeg library, instead of the uClibc version. But it's all sorted out now :)
18 Jan 2002
Downloaded and flashed the kernel provided by KenRestivo which includes the touchscreen fix. Works perfectly! Brightness/contrast/blanking all work well. My advice for newbies, who don't want to compile anything, would be to install tuxscreen-kernel-blfix and microwin-image.jffs2 from the v0.4 directory. It's not bleeding edge, but it works.
I built my own filesystem image with a few new apps (Hello World :) and flashed it just to make sure my toolchain was setup correctly. Everything looks good.
17 Jan 2002
Wired up a null-modem cable and tried to run inferno. Binding to /dev worked fine, but the rdp command would only give me garbage. Tried all baud rates, but still no joy. After a couple of hours of frustration - making a new cable, trying different versions of emu, I eventually realised gpm was still running. Doh! As soon as I killed that off, everything worked as expected.
I tried connecting at 115200, but it would hang half-way through. Running at 9600 was no problem. I then decided to copy out the original Inferno image. Hmmm... 4MB at 9600 = 1 hour. No thanks, lets see how fast I can crank this thing. It seems as though the maximum I could reliably connect at was 38400, so I stuck with that. Removing the protection on the flash worked liked a charm. It took three attempts to reset all bits.
Flashed blob, and rebooted. Cool. Installed the pre-compiled, v0.4 kernel and fs image. Rebooted and had a play. Very nice :) Then I rebooted into blob and reflashed the fs image with the pre-built microwindows image. Rebooted and it fired up nano-X, but got stuck at the calibration. Logging in on tty2 and killing the calibration process caused the window manager to appear, but no touch capability. That's tomorrow's task :)
15 Jan 2002
Thanks to SteveDavies for pointing out that the PFI line needs to be connected before the phone can do anything useful. Without it, I was able to dial out, but the connection would only last a few seconds, and it would only do pulse, not DTMF. So, following his advice, and after looking at the waveform of the PFI signal posted by KenMcGuire, I wired up a simple 555 circuit (see schematic at http://www.labyrinth.net.au/~steve/TuxScreen/index.html ) to generate a 120Hz waveform with the appropriate duty cycle. I connected this to my Tux, and I was impressed!
The acoustic feedback when touching the screen now works, I can dial DTMF digits, and the line doesn't hang up suddenly. This thing is pretty cool.
Now it's time to surf with it :)
I setup all the ISP stuff and loaded the web browser. It only connected at 33600 :( (JohnLaur adds: Well that makes sense since it is not a 56k modem! If you are sure that it connected at 33600, then that answers the question of is it 28.8kbps or 33.6kbps capable (ucb1200 chip) -- it also verifies that the ucb1200 chip is the modem and the telecom codec can be made to work properly. I wouldnt be frowning if I were you. Thanks for discovering this!) With my USR I connect at minimum, 44000, usually 45300. Oh well. It'll do for now... Went to a few sites, and found it's verrryyy slow rendering the page. Loading is not toooo bad, but a page with a few images takes a lonnnggg time to redraw.
Okay, I've had a play with it. It's too slow for me, time to install Linux :)
12 Jan 2002
Started playing with Inferno and sboot. For some reason, I cannot do the Pi trick to get a console. :( Ctrl-P works fine however. I also managed to get stuck in the red-screen/download software loop, but a quick look at SoftwareDownload showed Esc-2 should fix the problem. It did :)
11 Jan 2002
Cut off the US power supply and wired up the Tux cable to my own DC power supply. Fired it up. Ooohhh.... Nice :) Had a play with the built-in software. I particularly like the web browser and calendar apps. The note drawing app is quite nice too, but it slows down a lot with a big image. I also like the pop-up keyboard facility so you don't need the external one!
Out of curiosity, I measured the current flowing into my DC-DC supply. At power on, when the image is decompressing, it runs at about 400mA. After it's booted, and during normal use, it's about 270mA. When the screen blanks, it drops down to about 110mA. Not particularly impressive if you want to run this thing from a battery... Of course, this is the entire box, stripping out all the unneccessary hardware should make a difference.
10 Jan 2002
My TuxScreen arrived today! It's been a long wait, but the past weeks have allowed me to download all the sources and setup a cross-platform development system. I've also built a 3.3V/5V power supply to power up the Tux. Since I'm in Australia, the normal power supply is not suitable. Looking forward to getting home and firing this thing up!
|Content of these pages are owned and copyrighted by the poster.|