Last weekend we had another nice development session, and we’ve finally got all the modules assembled on the SoapBox mark II. It is now sitting indoors, patiently awaiting the surgeons’ skills and attention to bring it to life.
If you’ve been paying attention, you can see we’ve changed the battery pack. The 10Ah lithium-ion pack recycled from an electric bicycle died an awful death during recharging, all bloated and leaky and smelly. It went from 3 bad cells to 10 bad cells in a single charging. I don’t understand the problem of those cells or that charger, but it’s all over now, we won’t be using them again.
So the new pack, kindly loaned by NJay for our tests, is composed of 4 lead-acid 7Ah 12V batteries, disposed as 2 series in parallel for an output of 24V at a capacity of 14Ah. Not bad, we actually upgraded in capacity – and of course weight. They’re strapped down to the chassis with double-sided velcro strips (cable ties) which absorb vibration and tie the load quite well. It doesn’t need to pass a crash test, it just needs to keep the batteries there.
Another item that died a mysterious death multiple times was the DC-DC converter NJay had fabricated to supply 5V power to the GTA02 Freerunner board and the USB joystick, and every time it was my fault. Damn, why doesn’t positive electricity have a different colour from negative electricity? So Manso has skilfully rebuilt the circuit from scratch on a perf-board (this time with diode protection), and the central brain is powered-up once more. It’s quite nice to be part of such a skilled team – who would have thought a mechanical engineer was so good with electronics? And Manuel did the mechanics work of attaching the converter and the heat sink on the power chip, carefully avoiding the destruction of all the delicate handwork.
Something else you can see on the right side of this picture is that I had to resolder the I²C bus wires on the GTA02 board because the golden testpad simply broke away from the board (I hate it when that happens). It was then soldered to the debug board connector pins, but they broke again and again, so I ripped it out and soldered directly to the board – on those tiny 0.1mm pitch contacts (when the wire is twice as thick as the soldering pad, and the iron tip is 10 times as thick as the wire, you know you’re in trouble). That single soldering took me several cups of herbal tea and several days of mental preparation, but it worked out just fine.
So we put everything on the trike concept, and we fired it up. It doesn’t work, to say the very least. There’s all sorts of glitches in the communication between the 3 controllers: the GTA02 and the 2 micro-controllers on the power bridges. So we’ve spent a few hours poking and prodding trying to debug this connection, with the oscilloscope showing more of an art exhibition than serial comms square waves.
We got to see a bit of everything as we went along experimenting with resistors on the bus: shark-fin waves, pure noise, 3-level waves, you name it. Its like NJay usually says: “This may be a digital bus, but the universe is analog.” So finding the right tuning of the bus is essential to guarantee that it performs its duty of delivering digital words back and forth. The only time we actually saw well-shaped square waves, it didn’t work – because the “high” and “low” voltage levels are different between the master controller (3.3V) and the slaves (5V).
This hadn’t been a problem until now because we were using a very short cable and only 1 slave for testing, but with a much longer and Y-shaped cable things are quite different. So we’re now busy cleaning up this mess – which was created by badly-defined design specifications, of course (I forgot to tell NJay the bus was 3.3V). But no worries, da masta is in da house. He already has a few ideas and this should be sorted out in a session or two.
And then we’ll finally have it do its thing: rock and roll on the road!!!
So, happy hacking everyone!