My Seiko based BMS

[reagle]

I finally got around to throw it together- basically an 8 channel Seiko S-8209 based BMS with OV/UV and balancing.
The chips are working pretty well, and are very easy to string together. Getting them in the first place is a different story.
I used a separate charge input, with an NMOS FET, and a load output with its own protected high side switch (BTS555 from Infineon). The part is not cheap, but is built like a tank, and with 2mOhm Rdson, it barely flinches running my 20A+ lawnmower motor
The balancing shunts are 6x100Ohms 1206 resistors in parallel- I was too lazy to order 2512 size ones I was planning to use originally, so in go a bunch of smaller ones. They spread the heat well though.

 

[fechter]

Cool.

Here's the datasheet for the chip: http://datasheet.sii-ic.com/en/battery_protection/S8209B_E.pdf

Where DID you score those? Usually you have to buy like 1000 of them.

I've been searching a long time for something better than a LM431 (which sort of sucks), but it is readily available.

 

[reagle]

Oddly enough, Avnet was able to sample me some, and then proceeded to tell me they can't sell any
Seiko does sell them directly in partial reels from what I heard.

 

[fechter]

What kind of batteries are you using (like what voltage set points do you have)?

What are the little white things in series with the connector?

Something like that could really simplify some the things I'm working on.

 

[reagle]

Running an 8S1P pack of Headways, using Seiko S-8209AAB-T8T1S (The datasheet is different, as these are A part, not B http://datasheet.sii-ic.com/en/battery_protection/S8209A_E.pdf), so 3.8V OV, 3.65V balance and 2V UV.
It's a compromise,but seems to be OK in my case. In series with each cell sense line is a small 2A fuse- this way a short in the BMS does not kill anything.I also added a Zener across each cell after the fuse, to handle cell string connection in random order

 

[fechter]

Nice. I really like that chip. Stuff like that for Li-Co chemistries have been around a while, but not too many LiFePO4 varieties. I wish there was more availability.

Just out of curiosity, what did they quote for a price?

Nice on the fuses and zeners. I've been looking at TVS diodes (basically the same thing). Since I'm cheap, I was considering making some "fuse traces" on the PCB itself. Just sections of the trace where it's thinner than the rest so it can blow like a fuse (and hopefully not turn into an arc welder), or some kind of fuse wire soldered on.

 

[reagle]

Indeed, these were very convenient in my application- a last minute need for a BMS , since my SLAs died in the mower but grass has not stopped growing. Surely I had a year to make it, but nothing gets me going as well as an actual need. So instead of using all kinds of fancy chips I've been planning to, I picked a quick and dirty approach
The price I had quoted was around $0.63 each at 800 pcs qty

 

[grindz145]

What an awesome IC. Wish I could get my hands on some... Nice layout, you must have some experience....

 

[reagle]

Some more picture goodness- mounted in a Hammond box, with lightpipes and wiring, next to the cell pack and the mower. The plug is for charging, and the Lightpipes are over 7 out of 8 LEDs on balancing shunts. The last one did not fit Still need to tweak some wire length and routing, but its getting there
Headway charger does CC/CV and then starts pulsing at the end of the cycle, making LEDs on balancing shunts dance a bit- see the video

 

[grindz145]

Looks like some kind of balancing dance party to me

 

[fechter]

I don't think the blinking will cause any real problem. Each cell should still do what it is supposed to.

Have you tried measuring all the cell voltages after the charge cycle? Just curious to know what kind of variation there is between cells. It should be pretty tight.

 

[reagle]

Yeah, it just shows that some cells hit the threshold earlier than others. At the end they all light up, so they all hit 3.65V/ The blinking dance is mostly due to the chargers' pulsing at the end of the cycle
Also tested UV lockout under load- had to go around second time, mowing until UV hit. Things just shut off and stayed that way till cells recovered a bit. Then they ran a few seconds and shutoff again. Looks pretty good there,

The cells are pretty close- I ran the first few cycles with LTC6802 demo board in parallel as a monitor, watching them.
Right now, after being fully charged and sitting idle with just quiescent drain of the BMS for about a day, I see:
3.337
3.343
3.346
3.339
3.337
3.354
3.339
3.338

As measured with wavetek Meterman 35XP

 

[fechter]

Those numbers might be more of an indication of cell condition than BMS performance.

How do the numbers look about 15 minutes after end of charge?

 

[reagle]

Sorry it took so long to get to that:
15 minutes after charge ( and many more cycles since last post)

3.528
3.573
3.574
3.534
3.595
3.592
3.573
3.595

Those numbers might be more of an indication of cell condition than BMS performance.

How do the numbers look about 15 minutes after end of charge?