VERY interesting! Thank you, thingstodo.
This is exactly what I was looking for - a robust industrial solution that might extend to automotive use. It's good to hear the system works around the 575VAC, because this data line may have to connect to battery cells in the front and back. This means the data line will be in a "tray" going from the rear of the car to the front along with the main DC power bus (and all the noise that gets injected onto the power bus from the motor controllers.)
The BMS system I'm working on is actually pretty similar in some ways. It's called isoSPI, and works at 1Mbps. (Sorry, I was wrong about the 1Gbps statement earlier
)
LTC6804-1/LTC6804-2 - Multicell Battery Monitors - Linear Technology
Anyway, it uses a driver IC that translates standard SPI to a differential pair isoSPI. The system can be configured as either a daisy chain set-up or a parallel setup with multiple nodes on one bus. The voltage is much lower than 24V, but it does use termination resistors and each node is isolated from the bus with a transformer like a 1:1 ethernet transformer.
IF I understand the shielding correctly (and PLEASE correct me if I'm wrong) the shield provides a path for the noise to go to "ground." In my case, I may be able to use short (about 10" long) shields connecting to each battery module's low side "ground" This would limit the voltage potential between the data wires and the shield to about 50V. Or would this not work because that low side "ground" could be at 600V, perhaps higher than the noise source?
I would like the communication bus - whether this one or the CAN bus to the motor controller to be extremely - read - "automotive" robust. However, I do NOT plan on putting any control critical information such as ABS or throttle position on the bus. I do think it would be really cool to put information such as motor temp, motor controller temp, etc. on the CAN bus using standard automotive CAN protocol.
- E*clipse
Quote:
Originally Posted by thingstodo
We run Devicenet (a variant of CANbus) in electrical rooms, run in the same cable tray with 575VAC, within a couple of feet of 4160VAC. It is *VERY* robust. But the data rate is comparatively low. There is a conductor described as ground, but the cables have no overall shielding, and no shielding on the differential pair. There is a terminating resistor at each 'end' of the bus .. but that's it.
The way that Devicenet powers communication is .. interesting ..
A separate 24VDC power supply, that is rated for the Devicenet spec, is used to supply the power for the isolated side of each communication board, in each device on the network. That's ALL that is powered up by that power supply. It disables on short circuit, ground fault, and a host of other problems but by doing that it protects the communication boards for each device.
Would that method give you good isolation between the DC bus and the communications? I'm not sure if that is paranoid enough for the communications between controllers in a car. But it's something that is used in industry and I thought it might apply.
|