Battery Vehicle Society

This topic is for a slightly different approach to cell monitoring. Using a LTC6802-1 chip we can monitor up to 12 cells. While the chip is designed for stacking, I have decided not to go this route, but instead mount each one on a pcb with a 16F886 pic to control it and get the cell data out of it. This approach makes the system very modular because it uses the same topology as the present setup with a Master sending out commands on the slave bus and receiving data back on the master bus. Just daisy chain as many slaves as required. Aside from the use of the 6802 for cell measurements instead of generic pics, the other main difference is where isolation from the traction pack occurs. In this design the 6802's spi bus is isolated from the pic by U2-U5. The pic will control the data rate which I plan to run very slow, like 1200 baud which will be more than enough for measuring 12 cell voltages once per second. The pic can anticipate when will will ask it for a measurement and get all the readings ready for when it receives the request on the slave bus.

All the slaves will be powered from the master. While the buses aren't isolated there is nothing stopping us from powering it with a isolated 5 volt supply and using optos at the master. This trick works well for the existing master bus and would add an extra level of protection. For master-slave (-slave...) connections I have chosen to use off the shelf A-B usb cables. These are available any where in any length from 0.2 meters on up. While the signals won't be USB compatible, it shouldn't cause any damage trying to connect usb devices to it, they just wont work.

I've put 2 leds on the schematic. Command 7 is very useful for diagnosing problems, so I want to keep it. The other led can indicate that the slave has initialized the 6802 and all self tests are o.k. (the 6802 can carry out self test on itself and also detect open cell connections).

Each slave will monitor from 4 to 12 cells with a total stack voltage of 10 to 60 volts.

Previous schematics and initial discussion of this slave board can be found here.

Here's my latest schematic, main changes are as follows;
J1 is now a DB15 socket.
Added D3 for input polarity protection, just in case.
Added a prototyping area to help with fine tuning bus communications.

Any comments, questions or errors that I may have made?

I am going to work on footprints and then do a pcb layout. For the initial build I will only populate the right half of the board and test master and slave bus communications before adding on the 6802.

Good choice. We've built a few BMS's around these chips before with great success.

We have a stock of them left over if you still need to purchase any? Sometimes there is a very long lead time (hence why we overbought)..

PHEV wrote: We have a stock of them left over if you still need to purchase any?

I haven't bought any of the LTC6802 chips yet, so I would be interested. How many you got and how much do you want for them? PM me with the details. It will probably be at least a month before I need them. You got any hints, tips, tricks, advise on using the LTC6802-1?

Here's my plan. I am working on a pcb layout and hope to get it done and sent off before my holidays on Friday (and have it waiting for me when I get back). Then I want to build a few boards with just the 16F886 chip and test the bus communications with simulated cell voltages in software. My Matiz turns out to be a good test bed for this as the current system sometimes suffers from interference. If I can get reliable comms with a slave mounted behind the front bumper and one at the very back then I will know I've cracked that part of it. After that I will then solder on a 6802 and work on the cell monitoring.

Latest schematic.

I should have a pcb layout soon.

EDIT: uploaded wrong schematic.

Didn't get the pcb layout finished, so it will have to wait till I get back from my hols.

PHEV has sent me a link to a LTC6802 based bms project that he was involved in.
http://endless-sphere.com/forums/viewtopic.php?f=14&t=6602

Some good info in there, thanks.

Glad to help. Hopefully "Methods" will help more if he can.


Perhaps it's good you didn't get the board spun - it might well be worth looking into changing to LTC6802-2 before you go too far - we had a lot of messy problems with the v1 chips when in real life applications..

Have a good holiday

Steve

O.K., I'm changing to the LTC6802-2 chip. I've got some on the way, thanks to a ebay link from PHEV.

Quick update, latest schematic with changes to 6802-2 chip. Nearly ready to send gerbers off to china and get some boards made. Just have to figure out how to do polygon fills in kicad for the ground planes. Board measures 50x100 mm and will slide into one of the extruded alloy cases. One end has a DB15 connector for the cells and 2 rca jacks for 2 temp sensors. The other side has a USB A and USB B connector for power and bus connections. USB A-B cables are available off the shelf in lengths from 20cm to 5 meters, should make connections between slaves really simple.

Gerbers away, we'll see what comes back.

Gregg, good luck with the LTC6802s, its certainly something I looked at at the very beginning and they are definitely a good IC. I'll be interested to know what final accuracy you can achieve with them.

What company or contact do you have for making the PCBs? I'm wanting to send away a few PCBs for manufacturing and it would be great to have a trusty contact from China.

I've been using this company.

http://imall.iteadstudio.com/open-pcb/pcb-prototyping.html