Pruning Xubuntu

Today involved work to the operating system on the PC board.

Health Check

The most obvious thing to start with was to check the output from the picoPSU was as expected.  All of the voltages were pretty much bang on, and I found that the voltage reading on the bench supply isn't all that accurate - it reads about 11.8V when the output is 12V.

I noticed a 0.2V drop between the input and output of the picoPSU - this is a typical BJT saturation voltage, which is probably how the switch is implemented.

Graphics Settings

Because the system isn't very powerful, I made sure all of the 'fancy' options, like the compositor and shadows, were disabled.  As I'd done last week, I connected the system to my laptop using a wired network.  It was quite simple to share the wireless connection to the university network from my laptop to allow the robot's board to access the internet for updates and package management.

Pruning

Xubuntu Desktop comes with a lot of software to allow the average user to get on with what they want straight away.  I used aptitude (my favourite package manager) to remove just about all of these packages, as there is no need for them on the robot.  I meant to check how much space would be freed on the disk, but I forgot - it might be recorded in a log somewhere, but it doesn't really matter if it isn't.  Uninstalling took a while, and I followed it up by an update, checking that the kernel version to be installed wouldn't be incompatible with the CPU.

Wireless Adaptor

An 802.11b USB wireless adaptor was available for use on the robot.  It would be useful if the robot could be connected directly to the university wireless network.  When I plugged in the adaptor, I expected Xubuntu to react and notify me that wireless networks were detected.  lsusb showed the adaptor, so I turned to the internet for a solution.  The packages required were installed, and eventually it looked like I'd need to find the firmware for the card somewhere.  I looked around for quite a while, before checking the dmesg log.  This showed a URL where the firmware could be found.

After putting the firmware module in the appropriate place, I restarted the system.  The wireless card was operational, but I couldn't even see the university network.  It's possible that it's only visible to 802.11g cards, so I decided to use the wired connection for now, and use an ad-hoc network later on in the project when wireless communication is necessary.

GParted

I've used gparted in the past to modify disk partitions.  Since it's not a default package, I installed it, then had a look at the disk.  It looks like about half of the 4GB partition on the hard disk is currently is use.  That's pretty good!

Next Steps

I still need to figure out which microcontrollers to use in the project.  This is very important so I can start designing code and, eventually, PCBs.  Ideally, I'd like to produce the board quite early on in the project.  It would be useful if it's designed to be easily modifiable later, for example the ability to cut tracks at predefined points and provision of pads to connect fly-wires to.

Most ICs require 5V to operate - the obvious option for this is to use a 7805 regulator.  This is quite wasteful, and an alternative is to use the 5V line from the picoPSU to supply these devices with power.  For now, the 7805 is a good way to go, as it allows testing without powering the PC board.

I've got another meeting with my supervisor tomorrow, so I'll discuss suppliers and ordering parts.