Had a lot of trouble over the weekend getting my master and slave boards to communicate.
Seems the main trouble is that the built in pull-up resistor on the Avago fibre receiver is too strong (1K) for the wimpy parallel opto to overcome so I've had to do a little stanley knife PCB adjustment and bung in a couple of wire links. It's a wee bit annoying having to modify my brand new PCBs but not the end of the world.
I wish I had made the thermal breaks round the pads on the ground plane a little more generous too as it is really hard to solder/desolder components that attach to the ground plane. I suppose it's all a learning curve
I still like Gregs surface mount idea, that's a definite for the future.Got most of the display routine functional now. I decided to draw most of the screen at the start of the program and just use the main display routine to change numbers on the screen, this means less serial comms and more speed.
Derek
I have bypassed the transistor on the masterbus output on both boards as I couldn't get it to work properly. The fall time of the square wave looked a little slow but I just don't know enough to fix it, I must have been driving it incorrectly but these are the little annoyances that crop up when you are trying to teach yourself electronics as you go. The theory behind the transistor was that it would allow me to drive the Avago transmitter hard without overloading the opto outputs. I should have read the datasheet properly though as the optos are capable enough of driving the Avago 
boo hiss. The workaround is not elegant but it works,
It is always battery 6 comms error, which is at least consistent, unlike the old BMS. The fibre cable has eliminated problems with the data busses so I am confident that the problem is EMI related. The way I have the batteries distributed and cabled means there is a large loop of high current cable with the rear BMS board sitting slap bang in the middle 
I am going to try shielding the enclosure and grounding the enclosure to the traction pack ground whilst keeping it insulated from the chassis ground. Hopefully this will work, but I'll be looking to reroute the traction cables to keep them close together instead of forming the 1metre diameter antenna that it is just now. I had originally routed the cable this way to keep the cable runs as short as possible but an extra metre or so of cable wouldn't be too high a price to pay for less interference. 
Any EMI experts out there?
Moving the wires further away helped, but I had also modified the master board when I was trying out the current loop idea, basically it was still connected to the master with Peter's isolated DC-DC trick as well. The current loop was getting it's supply from the same 12 volt supply as the charger relay circuit, so the noise must have been feeding back through this. I have now removed the current loop circuit from the master and it charges fine. Will give the car a run later, as I had to leave it at work last night, and check it's all o.k.
I am investigating a different idea to improve serial comms between the slaves, its called the 
, how retro can you get, but with the added benefit of 20 I/O pins from the pic being available. Yes the interpreted basic won't run as fast as a complied program, but this is a 32 bit 80mhz processor. And being able to program your master software in a free version of basic is nice, but with this you don't even need a computer to do the editing, just plug in a keyboard and away you go. At the moment though it has one glaring omission, no serial comms. 
