18650 Hotswap

[martin_sch]

Hi,
I am currently designing  a project that uses 6 18650 to power itself. I would like to be able to hotswap 2 or 3 of them whilst its running.
The Power Draw is about 1.5A in normal Operation.
My current circuit uses schottky diodes to combine three packs of two batteries and three chargers to charge each of them.

Will this work like this?
Is there a better solution than using diodes?
Can I use it while its charging?
Is there a jellybean chip or solution to charge 3 18650 in series? (to minimize the drop across the diode)

Greetings Martin

[jbb]

Something of a grab bag here to get things started.


Generally, the jellybean solutions are for single cells only. And certainly a multiple-battery swap system is beyond the usual scope of generic designs.

Combining the battery outputs with Schottky diodes is a classic move. You can have a look for Ideal Diode or ORing or Hot Swap controller chips from several companies to do a better job (e.g. less voltage drop, more control).


Your circuit doesn’t seem to have any over / undervoltage protection on the cells. This is important for safety!


I see you have a ‘12V’ input. If we assume a 10% drop for tolerances and cable resistance, you only have 10.8V, which isn’t enough for 3x LiIon cells (might be OK for 3x LiFePO4).

FYI, if you’d like to ship a product with Li batteries in, there are limits to the size of the cells and total weight of cells.

Further, if you want to take the product on an aircraft in your hand luggage, the maximum allowable battery pack size is 100 Wh.


The way you tap the system load off the chargers can be a little problematic, because it means your chargers don’t have a measurement of the true battery current. What are your charging and system load currents?

[thm_w]

If you search for 3S BMS on ebay/alix you can find a lot of boards. One is using the BM3451 IC to manage balancing and protection. You can buy the chip alone as well.

I see you have a ‘12V’ input. If we assume a 10% drop for tolerances and cable resistance, you only have 10.8V, which isn’t enough for 3x LiIon cells (might be OK for 3x LiFePO4).

His current design is two cells in series, so 12V is fine for MCP73213. But it does mean low efficiency, maybe 60% efficient, due to the voltage dropped across the linear charger.
I'm surprised they don't mention anything in the datasheet about matching the two series cells. Do they think balancing is not required here at all?

[Siwastaja]

I'm surprised they don't mention anything in the datasheet about matching the two series cells. Do they think balancing is not required here at all?

No need to be surprised; if you look around more, you'll see it's always like this. It is fairly rare to use balancing or center tap monitoring in 2s packs, especially physically small and therefore thermally coupled. There really is no strong reason to do it. Cell matching happens automatically when buying proper cells from the proper manufacturers. This is all by design, and very well proven by the industry over a long time. You are of course free to implement balancing if you think it helps, but then you should be evaluating the resulting reliability and safety after adding the extra components and extra complexity. The baseline reliability of a small, thermally coupled 2s without center tap connection is very hard to beat.