This took longer than it should have (I’ve been busy doing things for solarcar, I promise!) but I’m ready to ship the next revision of the MPPT daughterboard.

Here’s the changelist:

- Changed some hole sizes to better reflect real parts
- Increased the size of the input and output capacitor
- Changed the footprints for the inductors to represent the actual inductors I am going to use
- Reduced the size of the input connector from four pins to two pins
- Removed the offset nulling circuit for the current amplifier
- Removed the output current sense resistor
- Implemented both a synchronous and a standard rectifier
- Removed one of the output protection diodes because it was redundant
- Beefed up the TVS diode on the output

And here’s a screenshot:

There are plans to have a ceremony where an oversize check will be presented to the team. Several Maxim executives will likely come, and there will be several short speeches. We’re trying to find an appropriate Stanford faculty member to also take the podium and say a few words.

This is one of our larger donations and will be very important for our success in the coming cycle. Our coffers started a little low this year since we’ve had to take care of residual expenses from the Australia run, this will go a long way towards bringing us towards a comfortable funding level for building a car.

Look forward to an announcement of the date and time of the ceremony if you’re interested in attending.

Also look forward to further announcements of large sponsorships. There are several in the works.

Edit: Maxim has commented (see below) that they are not, in fact, “Maxim Semiconductor” but rather “Maxim Integrated Products”. I have corrected the title of this post to reflect the correct name. Sorry for any confusion.

Dan got around to assembling and soldering together the last of the battery bricks that will go into the new battery pack. That means we have 32 completed modules. Sooner or later we’ll get up the will power to put together the two spare bricks. Tomorrow we’ll be testing the bricks to make sure that their onboard electronics works properly. Afterwards we will be giving them their first charge.

Several of us got together on Thursday and drafted the protocol for the new telemetry system, as well as assigned the first level of its implementation. It’s a less verbose format than last year when we simply built a transparent CAN bridge between the car and the van. Instead of simply sending all messages across the network, this year we will have the telemetry board parse some car “vitals” and pass them along in a dense format to the van. Part of the reasoning for doing this is to prevent bandwidth saturation between the two cars. Hopefully our new CAN bus will run at 500kbps or even 1mbps - far outstripping our telemetry modem’s bandwidth.

The higher CAN bandwidth will let us interact with our new motor controller without sacrificing certain types of data.

CFDesign makes easier to use software for modeling flow for both aerodynamic and thermal applications. While the design for Apogee’s body is largely complete, we still have some opportunities to play with it. Having a fast way to iterate through various tweaks could be a good way to ensure that we have the optimal design that meets all of our constraints.

Furthermore, I would like to model airflow and heating in the battery box to get a better understanding of the temperature distribution in the cells. That will help me figure out the rate at which the battery pack will unbalance itself due to temperature induced changes in internal resistance.

Nathan, Andres, and I spent some time this evening figuring out what was wrong with the CAN bus. Up until tonight we had been unable to get messages to exit the transmit buffer, let alone actually make it across to a receiving node.

As it turns out, the problem was not too severe. It just hadn’t gotten the diligent attention it deserved. The two nodes were not operating with the same timing values, and the difference was sufficiently subtle to avoid immediate detection. The other problem was a failed magnetic isolator in one of the CAN modules. We popped out that module and popped in a new one and fixed the software. Now we can send the number “42″ succesfully between two nodes. Hurray!

I’m reasonably pleased with the modular architecture. It makes it easy to pull and replace hardware that might be suspect. That’s a really good thing both for benchtop debugging and field repair.

Saturday and Sunday were massively productive. We moved our new flammables cabinets into position, and cut up and got rid of the table in the middle of the composites room. We also got the oven on wheels!

It was a serious undertaking. We would use jacks and blocks of stuff to prop up the oven while we attached the casters underneath. Here are some pictures:

Oven on wheels

Wheels under the oven

We also tossed out the old SAE freezer (it didn’t work… otherwise we would have kept it) and the composites rolls rack. To replace the rack, we built a new, taller, sturdier one with beefier steel pipes. This frees up a lot of wall space. Pictures:

Letitia and I did most of the woodwork

Rack, still waiting to be fully populated

Dan Robinson and I have been busy putting batteries together. We’re half done! Here’s a picture of what half of our battery pack looks like:

Lithium polymer batteries for Apogee

I’ve also started crimping together the interconnects between cells. We should have all the cells done before the end of the week. If all goes well we’ll also start assembling the battery box soon. I’m working on the CAD model for it now.

Roy Frostig and Dan are also working on getting a telemetry program running. Since we haven’t defined the communication protocol for the telemetry link yet, they’re working on the backend server and the GUI now, putting off the communication translation layer until we have a better idea of what it’s going to look like. The thinking right now is to clone the Prius driver user interface for “at a glance” ease of figuring out the status of the car. There will be more in depth information available as well, to aid in diagnostics and strategy.

I’ve spent the last few days in Texas getting familiar with how NASC works. When I took the team lead position there was some question as to whether or not my lack of race experience would be a problem for the team.

While I still have not been on a race, I have seen how scrutineering and qualifying work and what sorts of problems have come up over and over again. Hopefully that will translate into a smoother path to raycing for us as we prepare our next vehicle, Apogee.

Also while on the track I got to meet and make friends with many of the members of other teams. They’re extremely knowledgeable and interesting to talk to. Most everyone on the track was friendly and interested in being a good sport.

I’ve brought back a great deal of information that will hopefully guide our design and help us prevent many design pitfalls. Our team has been innovative on many fronts, but on others we have stagnated. Copying someone else is the next best thing.

While at the race I spent most of my work time helping the University of Kentucky team to get ready for the race, bringing a little bit of knowledge passed along by my own team back from Australia. I also helped them write a very simple telemetry gatherer so they didn’t have to stare at lines in Hyperterminal stream by to try and see what’s going on. Most of my chillout time was spent with Waterloo and Minnesota either just chatting or talking about building solar cars.

Hopefully some of the pictures I took will turn out well and I can post them up here a bit later.

The race officially started around 8:00AM central time this morning, out of Plano. Michigan started with pole position but lost it to Minnesota early in the day. As far as I can tell they’ve reclaimed first place. The University of Kentucky started off in fifth, made it up to third or second, but fell behind again after the minimium speed limit took a toll on their battery charge and left them limping.

There’s still a long way to go to Calgary, Canada and it’s still anyone’s game.

Today I assembled the first production battery module for the next car. The circuit board on top is of my design, and mostly assembled by Jordan LeNoach. The batteries were selected from our boxes of cells by a program that my roommate, Justin Lebar, wrote for me for this purpose. Their selection was based on arrangements of five cells into modules that will produce a combination of low resistance, high capacity, and low RMS error on both.

The battery pack will have a capacity of 6.19 kWh. If the data sheets are correct, the lithium batteries will weigh 29kg. The parallel cells in each module will have an average of 0.8 milliohms of internal resistance. Based on very early tests, it seems that once packaged and including the resistance of the interconnects, the module resistance goes up to about 1 milliohm. That’s much lower than our previous pack, where module resistance was a theoretical minimum of 4.7 milliohms without including interconnects.

This large difference in resistance will save us about 50 watts continuous in cruising losses.

In the middle of the circuit board you can see part of the battery balancing circuitry. It’s a pair of back to back n-channel MOSFETs that act as a solid state relay. There’s a gate drive isolation transformer, a rectifier, and a parallel RC and zener diode to act as a gate driver.

Hopefully this will allow us to avoid the problems we had with Equinox’s battery pack getting unbalanced during the race.

I switched over to a boost topology converter rather than a SEPIC because of the lower components count and lesser stress on the power elements. The converters are now lighter and more efficient. The down side is that it is now impossible to have an output voltage lower than the input voltage, but upon reflection I realized that this was a very unlikely use case.

IXYS has also come through and helped me find a really nice FET to use for the main switch. They’ve also helped cook up a synchronous rectifier for the output which will buy back the ability to shut the converter off and raise efficiency.

Right now with the Schottky diode system efficiency is about 97.5%. With synchronous rectification I expect that to go up to 97.9 or maybe even 98%. At that point I’m going to call the power stage done. Right now it weighs about 112 grams. The synchronous rectifier and the potting compound will bring that up. I expect the final weight to be around 150 grams per channel.

I’m very pleased with progress!