DIY Electric Car Forums banner

1 - 10 of 10 Posts

·
Registered
Joined
·
1,626 Posts
Discussion Starter #1
Hello

Most of you knowing the cell-Log. It's an awesome little device to monitor up to 8 cells.
Now, I want to use it for my BMS. The problem, I want to monitor my 48S battery pack with one 8S device.

The trick:
I'm interested to monitor each 6 cells group (22,5v nominal).
If the differential voltage from one group of cells is too important: ALARM (problem, balance your cells)
Low voltage: ALARM (stop to drive)
High voltage: ALARM (stop charger)

For me this advantageous set-up mean: only 9 wires, super simple, super cheap, super useful (compared to top balance or botom balance only).

But to arrive at that and because the input per cells on the cell-log is 1,3v to 4,9v, I want to divide the voltage from each group of 6 cells by 6.
To be clear, I want to read each group of 6 cells as one cell(3,75 nominal, chevy Volt cells).
Ok.. so how to do this simple and reliable?

I'm mecanical designer, so keep this electronic solution simple as possible please ;)

I had this idea in my head since long time, but it appears to be possible few days ago when I saw the Acciona team use one cell log on his Dakar racing electric truck.
 

Attachments

·
Registered
Joined
·
383 Posts
this has been done many times before, have a quick internet search

unfortunately the cell log draws more power from one end of the pack than the other, this is not so significant when your cells are very large but make sure you have means to balance the pack if you are going to leave the logger on for long periods.

also that's quite a high voltage to be putting into a device, it may become more complicated because of the need for isolation, that however will solve the balance problem.
 

·
Registered
Joined
·
290 Posts
I did solder together a point to point bleeder circuit with my limited electronics skills, and it was fun and enlightening. But I tired out when considering learning about EMI and Isolation questions.

Also I was quite surprised to learn that even 50ma quiescent drain can haunt you.

you could utilize the miniBMS head board but build your own cell modules? This would reduce some of the work, and the headboard is only $45. I've converted to miniBMS from Ewert and just found the ultra simple design to be a good match for my simple car.

Josh
 

·
Registered
Joined
·
324 Posts
FYI my notes on a lead acid interface circuit I did based on some of the ideas on endlesssphere start with this post:
https://endless-sphere.com/forums/v...1de827c8a44e5b01a91c296184&start=200#p1025492



It worked well, though I never got around to adding the regulator to make it completely balanced with regard to current draw from each cell as I ended up ditching the lead acid.

You are on the right track with a series of resistor dividers, but it really needs to be buffered with op amps to be accurate. Otherwise the current draw of the celllog inputs will throw off the dividers.

Rob
 

·
Registered
Joined
·
1,626 Posts
Discussion Starter #7
Ok, thanks for the links and the comments everyone.
I've also find this about using the the cell-log with higher voltage: http://www.diyelectriccar.com/forum...digital-battery-monitoring-system-134346.html
And about the mods to stop the parasitic drain: http://www.diyelectriccar.com/forums/showthread.php/bms-using-celllog8-68159.html

Well, since it doesn't sound super easy* for me I will learn and try to understand this electronic stuff.

*I thought about few resistor in serie with each 22.5v modules :rolleyes:
 

·
Registered
Joined
·
1,626 Posts
Discussion Starter #8
Ok, I'm not familiar enough with electronic to fully understand the schematic below, but a friend of mine yes.
So, with his help, I will build a small pcb board to transform each 22,5v group (6S) into 3,75v. The cell-log will monitor 8 groups of 6 cells as 8 cells.

Because my top balance chevy Volt stay pretty balanced since more than a month / +1000 km I'm confident about this cell-log solution as cheap BMS.
 

Attachments

·
Registered
Joined
·
17 Posts
I drew that one a while ago (schematic).

The biggest problem is getting your hands on one of them - they sell out really fast from hobbyking, and take ages to get out of backorder.
 

·
Registered
Joined
·
324 Posts
Mine was based on the same basic schematic, with the following mods:

1) I scaled each resistor divider with the increasing tap voltage. Since V=IR, if you scale R with V the current draw from each cell will be the same. So instead of using 133k total resistance for each tap, I used 40k total for the first tap, 80k for the 2nd, 120k for the 3rd, etc. You won't find exact values of course, but it will get you a lot closer to current matching the load on each set of cells.

2) Instead of using two resistors of different values to create the divider, I used packaged resistor arrays. Using two values you will have some error by nature of often not being able to find the perfect values, but a bigger error due to the 10% tolerance on each resistor. For example, if you wanted to use 100k and 20k to get your 6:1 divider, right off the bat you'd be stuck choosing between 18k and 22k, as 20k isn't a standard value. Lets say you go with 22k for 5.55:1. The problem with 10% resistors is they are never less than 5% off, because all the closer values were weeded out and sold as pricier 5% and 1% resistors. But they can be off 10% high or low. So each 22k will likely either be around 19.8k or around 24.2k, and your 100k will either be around 90k or around 110k. So now each divider can be around 6.55:1, around 5.55:1, or around 4.72:1, and even if you buy all the resistors together you can end up with a mix. So when a channel on the celllog reads 3.75V, the module could be anywhere from 17.7V to 24.6V, or the average cell voltage could be anywhere from 2.95V to 4.1V. That's not going to do you much good.

The alternative is to buy higher precision resistors, which will still have both a fixed and random error, or buy a bunch to test and come up with the best matches you can. You may not be able to get rid of the fixed error (from having to use standard sizes) but you can reduce the random error. Without matching, even using 1% resistors could still give you about 100mV uncertainty on average cell voltage.

The packaged resistor arrays/networks have the same kind of tolerances on value, but all the resistors in the package tend to be nearly identical. When I used 4 segment Bourns isolated SIP arrays, every divider came out basically perfect 4:1. They also make 8 segment DIP packaged isolated arrays that are about 74c each on Newark or Mouser that would give you the 6 resistors you need to make a perfect 6:1. You just stack all 6 in series and tap the output at the top of the bottom most resistor giving you 1/6 the total voltage. Just like a battery string, really.

3) I omitted the zener diodes to simplify and reduce cost. They may be a good idea for protecting the opamps and cellog, but as long as you connect the batteries sequentially from the bottom up they shouldn't be necessary. Once wired, I also didn't have any issues unplugging the celllog from the powered down board. I routinely powered the board down when not in use by interrupting the connection to pack ground, powering down the opamps and the dividers without issue. After I had run my sims I also seem to recall having an opinion that at least one set of the zeners likely wouldn't have worked as intended from a protection standpoint, but unfortunately I don't remember why.

I ordered my celllog and a JST-XH 8S pigtail cable from progressiveRC. I went with the 8S celllog for datalogging/usb, but if you don't need it the 8M is even cheaper.

Good luck,
Rob
 
1 - 10 of 10 Posts
Top