12-cell CAN-based BMS modules

How do you store and manage your electricity?
Post Reply
zeva
Senior Member
Posts: 425
Joined: Sat, 15 Dec 2007, 05:09
Real Name: Ian Hooper
Location: Australia
Contact:

12-cell CAN-based BMS modules

Post by zeva »

Hi all,

Time to reveal the new BMS project I've been working on! Developed in part to fill the requirements of one of my clients, but intended for general EV use as well. (I'll be building a set for my RX7 in the near future.)

In brief it's based around the LTC6802 BMS chip for cell monitoring and shunt management, galvanically isolated over SPI to a CAN-enabled microcontroller. Supporting up to 192 cells on the same bus, approx 6ms polling time per module of 12 cells. PCB size is approximately 80x50mm, and under 10mm total height.

The LTC6802 is a very expensive chip in an annoyingly fine-pitch package (0.5mm) and was somewhat difficult/complicated to interface with, but I've been impressed with its accuracy and a lot of thought must have gone into its design/development. The sigma-delta ADC it uses seems to be much more accurate and noise immune than the usual successive-approximation type. Just a pity they're almost $20 per chip..

Image
Ian Hooper
--
"Never doubt that the work of a small group of thoughtful, committed citizens can change the world." - Margaret Mead
http://www.zeva.com.au
User avatar
Johny
Senior Member
Posts: 3733
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

12-cell CAN-based BMS modules

Post by Johny »

Looking good Ian. I don't see any bypass resistors though. Is there any provision?
When you say maximum of 192 cells - I have 16 cell sub-packs and 192 cells pair all up. Could I use 2 of these per pack - each handling 8 cells and still get 192 cell handling total? Not too bad if not - just run 2 buses.
zeva
Senior Member
Posts: 425
Joined: Sat, 15 Dec 2007, 05:09
Real Name: Ian Hooper
Location: Australia
Contact:

12-cell CAN-based BMS modules

Post by zeva »

The bypass resistors are the little SMD 100ohm ones near the quick-connect terminals along the bottom. I decided an interesting way to go would be to use very weak shunt resistors - only ~35mA shunt current - but leave them on for a relatively long time. So the microcontroller maintains shunt timers which kick off in proportion to how far over the shunt threshold the cell goes. Say if it hits 3.65V the shunt might stay on for 10 minutes, or if it hits 3.8V (the high cutoff) the shunt stays on for an hour, even if the charger has stopped. This way the board doesn't need high power resistors and doesn't generate much heat (no chance of fire etc), but still balances a fair few joules per charge.

16 cells per sub-pack.. yeah unfortunately that'd need two modules doing 8 cells each. The cell limit is partly to do with isolation voltage limits, and partly "number of positions on that rotary switch" limited. It would actually be no problem to program modules to support CAN packet IDs beyond those selectable by the 16-position switch (perhaps on some modules the selector range is ID 17-32 instead), and probably easier than running dual CAN buses. I don't expect there'd be many EVs out there needing more than 16 modules though - yours is certainly quite high voltage!
Ian Hooper
--
"Never doubt that the work of a small group of thoughtful, committed citizens can change the world." - Margaret Mead
http://www.zeva.com.au
User avatar
Johny
Senior Member
Posts: 3733
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

12-cell CAN-based BMS modules

Post by Johny »

Thanks Ian. It's one of those things I did to get "on the road" that I need to go back to - better cell monitoring. The "Headway" BMSs (not really Headway) actually are doing a great job but I would like better control and monitoring - I also can not easily replace them.

I'm really glad that you have used the LTC6802 - I remember reading about it when it came out.
A year ago I would have though that 100Ω bypass was a bit shy but I have experienced the same thing as many people - that once the pack is balanced and in regular use the cell balance stays very good and only needs the slighteset "tickle" to maintain.
It's long term storage that appears to unbalance cells the most.

Anyway - great job.
zeva
Senior Member
Posts: 425
Joined: Sat, 15 Dec 2007, 05:09
Real Name: Ian Hooper
Location: Australia
Contact:

12-cell CAN-based BMS modules

Post by zeva »

Thanks. Likewise, over the years my thinking has migrated away from the need for high energy shunts, after calculating/estimating how little balancing is required to maintain pack balance. I believe it's in the order of 1% per year drift, e.g two similar 100Ah cells might end up 1Ah apart after a year due to differences in self-discharge. That equates to less than 5 minutes of balancing from a tiny 35mA shunt each day to maintain pack balance!

Of course there is always that one instance when you first assemble a pack where it would be nice for the BMS to balance it completely in one hit, which is (IMHO) the main benefit of high power balancers.

I figure in this case the BMS master could have an "initial balance" mode which changes all the modules' shunt threshold voltage down to say 3.35V, and the little 35mA shunts could stay on (perhaps for days) until the cell voltages all match. The LTC6802 is accurate down to the millivolt, so despite the relatively flat voltage/charge curve of LiFePO4s around that 3.35V threshold, it should be able to get them all to a pretty close SoC this way.
Ian Hooper
--
"Never doubt that the work of a small group of thoughtful, committed citizens can change the world." - Margaret Mead
http://www.zeva.com.au
a4x4kiwi
Senior Member
Posts: 772
Joined: Thu, 03 Jan 2008, 19:04
Real Name: Malcolm Faed
Location: Australia
Contact:

12-cell CAN-based BMS modules

Post by a4x4kiwi »

Need a beta tester? I have an 8S pack to test in an ebike.
Silicon is just sand with attitude.

Blog: http://malfunction.faed.name
derBastler
Noobie
Posts: 7
Joined: Sat, 02 Nov 2019, 21:33

Re: 12-cell CAN-based BMS modules

Post by derBastler »

zeva wrote: Fri, 25 Oct 2013, 02:16 Hi all,

Time to reveal the new BMS project I've been working on! Developed in part to fill the requirements of one of my clients, but intended for general EV use as well. (I'll be building a set for my RX7 in the near future.)

In brief it's based around the LTC6802 BMS chip for cell monitoring and shunt management, galvanically isolated over SPI to a CAN-enabled microcontroller. Supporting up to 192 cells on the same bus, approx 6ms polling time per module of 12 cells. PCB size is approximately 80x50mm, and under 10mm total height.

The LTC6802 is a very expensive chip in an annoyingly fine-pitch package (0.5mm) and was somewhat difficult/complicated to interface with, but I've been impressed with its accuracy and a lot of thought must have gone into its design/development. The sigma-delta ADC it uses seems to be much more accurate and noise immune than the usual successive-approximation type. Just a pity they're almost $20 per chip..

Image
I would like to have your schematic for this pcb. I am trying to develope a board with the LTV6802-2, too but have big problems with the serial communication.
TyTower
Groupie
Posts: 42
Joined: Thu, 17 Dec 2020, 05:25

Re: 12-cell CAN-based BMS modules

Post by TyTower »

I'm assuming these are for LFP ?
If so the 3.65v is a max and no further no matter what anyone says . You are clogging the SEI layer just by being up at 3.65v .
They should stay down at 3.4v because the majority of the capacity is within 3.0 volts and 3.4volts .
Can they be adjusted down?
How would I get one to test please?
Attachments
EVPower!00AhDischarge.jpg
EVPower!00AhDischarge.jpg (247.12 KiB) Viewed 195 times
zeva
Senior Member
Posts: 425
Joined: Sat, 15 Dec 2007, 05:09
Real Name: Ian Hooper
Location: Australia
Contact:

Re: 12-cell CAN-based BMS modules

Post by zeva »

The design has evolved a fair bit since 2013 but still uses the same fundamentals, which have worked well. Here's the current version: https://www.zeva.com.au/index.php?product=133

The BMS modules report voltages (over CAN bus) to a master controller when asked, and perform shunt balancing when instructed, but don't know/care what voltage thresholds the master controller is using for under/over-voltage protection. The BMS modules can read anywhere from 0-5000mV per cell input so can be used with any lithium chemistry.
Ian Hooper
--
"Never doubt that the work of a small group of thoughtful, committed citizens can change the world." - Margaret Mead
http://www.zeva.com.au
T1 Terry
Senior Member
Posts: 1215
Joined: Thu, 30 Sep 2010, 20:11
Real Name: Terry Covill
Location: Mannum SA

Re: 12-cell CAN-based BMS modules

Post by T1 Terry »

TyTower wrote: Sun, 21 Feb 2021, 06:03 I'm assuming these are for LFP ?
If so the 3.65v is a max and no further no matter what anyone says . You are clogging the SEI layer just by being up at 3.65v .
They should stay down at 3.4v because the majority of the capacity is within 3.0 volts and 3.4volts .
Can they be adjusted down?
How would I get one to test please?
What symptoms would you see if the SEI layer was clogging? We use LYP cells but the basics are the same from what I can see at a glance, just a wider temp operating range, particularly in below 0*C region, and they are now marked as 3.3v nom. where the LFP cells are marked 3.2v nom.

T1 Terry
Green but want to learn
TyTower
Groupie
Posts: 42
Joined: Thu, 17 Dec 2020, 05:25

Re: 12-cell CAN-based BMS modules

Post by TyTower »

Terry ions stick in it and are not able to move so they are taken out of the total available ions . Capacity loss is what you see.
T1 Terry
Senior Member
Posts: 1215
Joined: Thu, 30 Sep 2010, 20:11
Real Name: Terry Covill
Location: Mannum SA

Re: 12-cell CAN-based BMS modules

Post by T1 Terry »

TyTower wrote: Sun, 21 Feb 2021, 18:46 Terry ions stick in it and are not able to move so they are taken out of the total available ions . Capacity loss is what you see.
I haven't experienced that with our systems and they are capacity tested every 3 yrs or so. The longest operating 24/7 as a house power battery that we've tested was 8 1/2yrs and it still held better than 3v per cell under load on a 2C test after delivering the 400Ah advertised capacity. We build a lot of systems and as a result we get to test a lot of the systems over the yrs of the operating life ... yet to have one suffer loss of capacity unless they were seriously abused. Our system stops charging if a cell reaches 3.6v in any cell and disconnects the battery if any cell sees 2.8v, but there is an over ride switch where the owner becomes the BMS and this is where the abuse can occur if they forget to return the switch to the auto function position.
I think ou will find the recharging slow if you limit the voltage to below 3.4v per cell and actually start to experience capacity loss ..... but the cause of that is a lot harder to explain so I won't attempt it here :lol:

T1Terry
Green but want to learn
User avatar
jonescg
Senior Member
Posts: 3771
Joined: Thu, 21 Jan 2010, 23:05
Real Name: Chris Jones
Location: Perth, WA.

Re: 12-cell CAN-based BMS modules

Post by jonescg »

SEI is both necessary to protect the anode, and unfortunate as it depletes the lithium inventory. Crud has to come from somewhere, and the mobile Li+ ions are a good source. So gains in the SEI means losses in mobile ion stock. Net result is lower useable capacity and higher internal resistance. If it still delivers more than about 80% of the original tested capacity over the course of a decade, you're looking after them.

If you can charge your LFP cells to 3.55 volts they will have a longer life, but (back on topic here) the ZEVA battery management system allows you to set that limit easily.
AEVA National Secretary, WA branch chair.
TyTower
Groupie
Posts: 42
Joined: Thu, 17 Dec 2020, 05:25

Re: 12-cell CAN-based BMS modules

Post by TyTower »

Terry, I don't limit the charge to 3.4v The charger is always set at 3.65v . I stop charging when the plate voltage reaches 3.4v.

jonescg, I disagree, that's pushing t too close for me and I stick to 3.4v but I make sure its accurate. When you charge the plates to 3.55 v leave it for half an hour and usually it just drops to 3.4v anyway that I have seen. Just seems to absorb it .
The "crud" you refer to is thought to be impurities and chemical compounds coming out of the electrolyte into which the ions stick and become irretrievable reducing the available stock of ions and clogging the passage of other free ions but its all fluid atm so i'm watching for all papers on Lithium cells as I'm sure you are too..
Post Reply