As you might have read, we are running an installation here in amsterdam called radarFunk.
This installation is based on a Windows XP machine and uses a Nokia 6680 mobile for user interaction.
BlueTooth is used to establish communication between the 2.
I’m not going to explain every detail, I’ll just highlight the points necessary to tell the hassles we had on the way.
The machine has no screen, keyboard, mouse.
It’s a server that runs a Director application to host and play the sounds, and a BlueTooth server (our BlueSocket technology) to receive input from the mobile device and broadcast it to every TCP/BlueTooth client attached.
When preparing the machine for the installation, we thought it could just be running 24/7, as we thought about the Nokia 6680 too.
This world is everything but an ideal one, and windows machines just can’t run 24/7 (I turn my mac laptop off only when updates require a restart).
The Nokia is less of a problem, but restarting it every night just can’t do any harm.
Of course we figured this out when the installation was already running.
To have you understand what the real problems were, you have to first know how this works and then we’ll highlight the possible flaws/requirements.
radarFunk MINI is a J2ME application that requires a BlueTooth SPP (Serial Port Protocol) connection to a server to broadcast note signals.
In the first revision the device had to be configured on application startup to connect to a BlueSocket server.
Now in case the BlueSocket machine was failing (it happened first time because of overheating) the J2ME client looses connection, and since CPU cycles are precious, I didn’t want to implement a listener to check if the SPP connection is broken.
I figured out that to avoid all the necess fiddling with the configuration panel of the application, it was better to have it reconnect to the last known server on launch.
Well, this worked and still does, but then we had to face another problem:
the server was locked in a box with no access to power switch.
I could have used a KVM extender to enable the exhibition people to turn off the machine properly every evening and start it up the morning after, but I didn’t want them to connect monitor/keyboard/mouse every time. They already have to execute a sequence of operations in order to restart the Nokia device.
Let’s examine what we had to deal with:
- First obstacle faced (but an easy one) was the BIOS errors when no keyboard is found.
- Solved configuring the option “halt on” in the BIOS interface. I simply set it to “no error”
- Second (a tough one): skip the logon screen on windows in order to have the programs in the Startup folder to start without the need of a login
- Solved accessing a very hidden control panel in XP. It can be accessed running
control userpasswords2from a command prompt
- Third: ok to turn it on and avoid errors, but how do I turn off the machine with no user interface?
- Solution: in the control panel, you can find “Power Options”. In power options selec the “Advanced” tab, you have a custom action for the Power Switch. Just select “Shutdown” from the list and you’re ready to go.
Wow, now we have a machine that will start/stop with no complaints at all.
Wait a moment… I said the machine is locked in a box, how do I access the power button?
Do I leave it out so everybody can just go there, click the button and shut down our installation?
I really don’t think so.
How do exhibition managers shut down my machine then?
Here’s when I decided to build my remote power switch/LED remote USB key.
You can read about it and see some pictures here.
Just for the article here’s a picture of what it looks like.
I installed it this morning and it’s very easy to use.
Plug the modified USB pen into the port, press the button.
It’s nothing but an extender for the Power LED and Power switch, but it’s quite neatly done.
Now Ed and the others at the Arti’s building are happier.
In case you want more detailed information or schematics just drop a line.
I’ll be glad to help whoever wants to try this.
Popularity: 1% [?]