Information on this blog is raw and sometimes unverified reporting straight from the road by teams. The event will issue a media release for any events requiring an official notification.

Note that links in blog entries are not maintained, so while a link may be verified to work on the day of publishing, this is not guaranteed beyond that day.

Friday, September 23, 2005

Leeming Sungroper: Friday 23rd: Tritium

Tritium have loaned us a controller, but have suggested that we should use it only as a spare. In the morning, we phone Dave at Tritium to run past him the concept of us swapping in the Tritium controller now. He gives us a few tips. There are two sets of wires going to the motor: power and data. The data cable has an 8-pin circular DIN-esque connector, to fit the Lillington controller. The Tritium has an 8-pin DIN to go to the motor's data cables, but Dave lets me know tha the pin-outs will be different.

I hatch a plan to build a short adapter cable. That'll involve me getting one connector for the Tritium end (and Tritium have thoughtfully included two in the case), and one for the Lillington end. First, out to the pits to check exactly what the Lillington end is: looks like a DIN-8. Then out to Jaycar to buy the bits, then back to Alatai to receive the email from Dave containing the pinout of the Lillington connector, then back to the track to make up the adaptor cable.

And the connector doesn't fit. The Tritium end is fine (of course), but the Lillington end doesn't fit.

Back to Jaycar, this time with the controller. There are two other teams there when I arrive. The connector looks like a "MIC" connector, which I'm not really familiar with, but's it's fairly academic anyway, as Jaycar top out at 5-pin MIC connectors.

Back to the track, and back to the drawing board. Intead, I build a double-ended cable: the Tritium plug hanging off the side next to the Lilligton plug. Ick. We jack the back wheel off the ground and plug the plug back into the Lillington controller to make sure it still works after I've bowdlerised it; it does.

Unplug the Lillington controller from the car, which involves unbolting the 120V input cables from the controller box. AUAGH!

Of course, the controller has dirty great capacitors in it, and although I've disconnected the motor from power, it's still got some stored up in it; more than enough to bite me. I would probably have figured this out a bit quicker, but when I work hard for many days running, I get a bit fatigued, and I can feel my intelligence and presence of mind slowly dribbling out my ears.

I remove the remaining power from the controller with a wrench and a big fat spark. Probably not good for the controller, but right now I want a bit of revenge on it.

Now, set up the Tritium controller. Plug it into power. See the display light up and the LEDs flash. Plug it into the data coming back from the motor and run it magic PhasorSense configuration, which probably should be renamed "You Just Sit Back And Relax While I Configure Absolutely Everything; All You Have To Do Is Spin The Drive Wheel A Bit". It flashes its lights to indicate that it now understands our motor. Plug the power lines in, including the external Hall sensor module. Hall sensors are nifty transistor-like devices that sense a magnetic field. We've got three Hall sensors buried in the motor (that's what the data coming back from the motor's data connector is: signals from the Hall sensors), and Tritium give us another two to stick outside, around two of the power wires. Now, turn the speed knob on the Tritium's driver control.

It works. It does everything it's supposed to do. It goes forwards, it goes backwards, it does regenerative braking. The creaking groaning noise our motor used to make at low revs is gone. It goes faster. You have better control over the speed. And it stops really nicely on the regen brake.

Right now, the Tritium controller is just sitting loose on top of the Lillington. We remove the Lillington, and unscrew everything from the Tritium in preparation for installing it in its final location.

Of course, before I touch the screws for the 120V, I check it with a multimeter to see if it's safe to touch. Well, of course it is. The Right Thing for the controller to do when powered down is to remove any excess voltage stored in its capacitors, down to say 3V or so. So that's what the Tritium does.

Its final location is upside-down underneath a pair of metal rails -- its a little longer than the Lillington, so it won't fit in the old spot. We reconnect various bits to it, put John in the driver's seat, and push him out into the pit lane.

We've made two serious mistakes, but we havenb't realised it yet.

John turns the speed knob.

Nothing happens.

Dave let us know that since this is a prototype controller, it's possible for the controller code to fail to boot from a cold start, and that the cure is to flick the reset switch. It sort of reminds me of Reason, in Snow Crash.

So we pop the array off and flick the reset switch.

Still nothing.

We roll back into the pit. Take the top off, look at it properly. Find one of our mistakes: we've failed to reconnect the external Hall sensors. The supplement to the manaul adresses this. Quite clearly. And repeatedly. Never power up the controller without the external Hall sensors attached, it says. Damage to the controller will result, it says. Assign someone to always check that they are connected before powering up, it says.

Well, we didn't. We put ourselves in a pressure situation, then it all seemed like it was going well, and we reconnected everything except this one piece.

We put the controller through its paces.


Three different teams, on hearing that we've had a Tritium controller coming, have offered help with installing it. Many teams are at scrutineering today, but one of them is on site: Aurora. I walk down to their pit, and we ask for help. Tom comes and helps. He plugs a laptop into the controller, and shows us how the PC interface works. I've seen it in the documentation but haven't tried to play with it. It's got parameters out the wazoo.

Using the laptop to talk to the Tritium proves that the Tritium's brain is happy and undamaged. But it doesn't directly solve our problem.

We phone Dave at Tritium. We 'fess up to having powered the controller up without the Hall sensors.

The conversation goes on. Initially Dave thinks that we must indeed have blown up the controller, but as we talk more, he becomes less and less convinced -- the nature of the fault we are seeing does not match his expectation of the sort of thing we should be seeing.

We finish the conversation without a conclusion. We continue troubleshooting by ourselves back in the pit. Perhaps there's a problem with my ugly double-ended cable. We check it out, and indeed a wire has come loose: when we flipped the controller upside-down and fixed it into its final location, we must have tugged a little on this cable, and one of my solder joints must have been weak.

While we're doig this, Dave calls back. He's thinking that it's something to do with the Hall sensors in the motor, which are connected to the very connector we're messing with right now. Oh yes, I say. Dave explains how to check the Hall sensors in the motor from the connector end, by hand-turning the motor slowly, and using a multimeter.

We do so, and find that one Hall line is behaving differently from the others -- it's shorted to ground. But only when it's plugged into the controller.

What's changed?

Well, now the controller is upside-down.


We flip it the right way up. The Hall line isn't shorted any more. We power it up. It works.

So we made two mistakes, and there was another built in to the controller: we failed to connect the external Hall sensors, which would normally destroy the controller, but we had a fault in our data connector, and the controller had another fault on that same connector. And those faults prevented the controller from spinning the wheel, which prevented it from destroying itself. So we dodged a bullet.

We track test, but it's too windy to get any idea of the efficiency of the new controller.

In the evening, we go to Parliament House. Speech by Claire Martin, the Chief Minister; and Chris Selwood, the WSC event director. We track down Tom and thank him. He tells us that in his opinion, Tritium controllers are the best in the world, and I'm far from disagreeing with him. My opinion of the controller has only gone up. We haven't had a chance to compare its efficiency to the Lilligton controller, and now we won't until we get back to Perth; but on any given issue it does the Right Thing -- you can figure out how its going to behave by figuring out what you want it to do, and pretty much always that's what it does. The only exception is the external Hall sensors -- they're an accident waiting to happen. Nevertheless, it's an accident we've dodged, and we're very happy.

Tomorrow: track testing. We have to run a qualifying lap to get our spot on the starting grid, and have to pass brake and steering tests administered by NT Roads.

-- Doug Burbidge