This blog is the story of taking a 10 year old Twike that has been setting idle in a backyard in Portland, Oregon and restoring it to is former glory and getting it back on the road. It is Twike number 434.
Friday, March 6, 2009
Latest update on Twike restoration
I have not posted since November, 2008. Sorry about that. Here is a summary of what has happened on this project since then. I purchased the rest of the A123 M1 cells I needed for $5.50/cell. I have 450+ cells now. I have been looking at BMS designs so I can protect these cells (they need to stay between 2.5 and 3.7 volts for best life and safety). Apparently, this is an area that is lacking in good off the shelf solutions. I will either need to pay lots of $$ for a solution (i.e. more than the cells themselves cost) or build my own so I can customize it to my specifics. As part of this effort, I have a demo board of the Linear Technologies LTC6802 chip. It seems to function with my testing so far but I will have to add up the cost of the parts to build a functioning BMS (shunt based design) to see the true cost. In the meantime, I started to do a serious attack on the TWIKE battery BMS setup. At this time, I am not pursuing the software based BMS any further as it became too difficult to see what was going on to try to emulate it correctly without any working real BMS boards. I turned my focus to the hardware itself and will the help of a friend at work, we got 5 working BMS boards to talk to the service program. I also located a disassembler for the BMS software to get a better understanding of how it works. That was fun as I looked at 3800 lines of assembly code! Over the last 30 days, I have been doing charge/discharge cycles on one of the best looking old NiCd packs. I was able to restore about 60% capacity by cycling the old cells. Last night at Synkromotive I carefully soldered back in a working BMS board (14 temp sensors, 7 voltage monitor wires, 14 cell to cell connections, 1 power connection, 1 RS485 4 wire communication connection). As I wired up the last connection, the board red led started blinking! After connecting it to the car and putting the car in charge mode, the car downloaded BMS software to it and began to read out the battery pack voltage! This proves the car charge/controller system is working properly and the other battery pack I was trying to charge has a bad BMS. I also discovered how to get into service mode on the controller (TWIKE Access Level 2). This lets you set all kinds of good stuff on the controller including the number of batteries. I tried setting it to 1 battery (which I did not know was even supported) and it seems to be working happily with 1 battery. When I tried to charge the battery, it complained that the battery was too hot (77 deg C). Clearly, I messed up one of the temp sensor solder joints. I'll have to run the service program to find out which one and fix it before trying to charge it for real. Next steps are to put another working BMS into another candidate battery after I do the battery testing to make sure the cells are good. With two old tired batteries, I should get 10 to 20 miles of range (for testing) while I work on my Lithium solution. I'm thinking 112s4p would be a good starting setup. Even with LION cells and my own BMS for Low Voltage Cutoff and Shunt Based overcharge protection, I may integrate 1 or 2 old BMS boards to make it trivial to keep the onboard controller happy (Both on charge and discharge). A hybrid solution may be the path of least resistance. We'll see. Stay tuned for more.