How to realize monitoring Electric Vehicle's temperature

lingyueqing

I'm looking to implement a system to monitor the cell temperatures of about 80 cells in a 240-cell battery pack for an electric vehicle (yes, at least 80 cells must be monitored). I'm looking for a method that will minimize the complexity and cost of implementing this system, particularly with respect to wiring complexity.

Using analog sensors sounds like a wiring nightmare (unless there's a trick I haven't thought of). My initial thought was to make a little PCB for each cell with an I2C or SPI sensor and two ribbon cable ports, so I could just daisy-chain a ton of these sensors and have them all read by a single master. I'd likely have to put in I2C buffers every 8 cells or so to manage capacitance, etc. However, I'm worried about EMI, as these protocols aren't (AFAIK) made for long off-board (a couple meters probably) runs in an unfriendly EMI environment. Would shielding be enough to offset this problem?

Even better would be to use a more robust protocol like CAN, but I can't find any temp sensors that communicate over CAN, and dropping a micro on each board just to do this conversion would make the board larger and more expensive.

Is there a better method I haven't thought of? If not, how likely is it that the I2C method could work if care was taken with shielding? EMI will be mostly produced by the large fields in the kilohertz frequency range by the AC motor and motor controller (we'll use shielded cables there too, but still).

EDIT: I see devices like the MCP2515 CAN Controller with SPI interface. I don't imagine I could directly hook this up to a SPI temp sensor and expect it to work, could I? Still not ideal as it's another chip on each board, but perhaps better..

bp_me

Could you go wireless?

https://www.enocean.com/en/enocean_modules/stm-331/

cros13

enocean stuff is enormously expensive.

Maybe a couple of i2c multiplexers/switches?

http://www.ti.com/product/tca9548a