After returning from Europe 10 days ago, I’ve been thrashing around for a sense of direction. When I posed the question, on the server-devel list, of what sort of integration of my changed code to shoot for as I continued XS-ARM debugging, I got somewhat demoralized. Daniel Drake pointed out, I think correctly, that supporting more than one distribution of XS was a substantial commitment, not to be taken lightly.
I had also been wondering how to relate to all the different ARM devices. Most of them use Uboot, in some form or other. But what might work for a pogo plug might not work for Tonido plug, etc. in terms of creating installation media.
I build up a current probe, using a hall effect sensor, and wrote the code to accumulate charge and discharge information and display it on an LCD display. Since the LCD shield I was using had 5 push buttons, I wrote code to calibrate and zero the current probe, and store the new calibration values in the eeprom of the Arduino.
Then I came across Scott Ananian’s blog about a $5 arduino board he had designed using Eagle computer aided design tools. A really low cost arduino sounds exciting to me. Maybe I could do an Eagle CAD project for the battery monitor, and create a shield to go along with his main board.
There’s no real substitute for experience. The current sensor, from honewell, turned out to have a very fragile circuit board attached to the electrical terminals. Fortunately, I had purchased 2 sensors, and was able to improve the design, to account for this fact, when I used the second one.
There’s a potted current sensor, ($18), that is mechanically much more robust. But at this point I’m focusing my attention on an op amp solution, which I built up 2 months ago. All of the parts for this approach may cost two dollars or so.
ARM project may be back on front burner
I got email from Andrew and Sridhar at OLPC-Australia. It turns out that they are interested in using the XO-1.75 as a School Server. I’m very excited about collaborating with them, and can imagine some form of consolidation and integration path with XS Boston in the long run. In the mean time it helps to have someone on the ground to focus my efforts, and collaborate with.
An an aside, I noticed that pre-release XO code 12.1.0 uses systemd for start up. This suggests to me that in the longer time frame, the Trimslice start up problems are strategic and need to be solved.