Li-ion battery pack with PCM

[yann4905]

Hi everybody,

I am a diver, and I'd like to custom my dive lamp with LED XML. I have electrical/electronical skill due to my profession.
As my application is for a dive lamp, it means that the battery pack will be in a case.

It means that it is really important to control the battery pack (li-ion), to prevent damage.

The aim is to obtain 14,8V and 6600 mAh.

I would like to control charge and discharge with PCM.

My question is here: what is right and best choice between both schematics?

Thanks

[poorchava]

I think the safest solution is to use cells with individual PCMs and then solution 2. This way every cell is protected and each cell can be monitored separately by general PCM (or rather battery management module). Solution 1 is safe too, because of the integraded protection circuits, but PCM cannot control cells one by one, which may or may not be a factor n your application.

Another thing is how deep you dive and in what setup. When diving to significant depths (as in: requiring decompression chambers and all that) you need to be 1000% sure that your setup will not catch on fire or release any sort of gas. This is because breathing gas composition and pressure may be very different from atmospheric air and gasses which are normally rather harmless or mildly toxic can get absolutely lethal (eg. single sneeze of WD40 in decompression chamber in certain conditions = you're dead).

I can't tell you more (despite the fact I'd like to), but to be really safe you need a really tough and gas-tight enclosure for the battery pack. How does one make that? That's the part which is not exactly straightforward and I can't tell you.

[yann4905]

Thanks poorchava for the answer.

The dive lamp is only for recreational dive.

When you say "I think the safest solution is to use cells with individual PCMs", you mean cell protected like the model in attachement for exemple? If yes, the solution 1 is not to applied because the "onbord" pcm will not know which cell is defect, due to the parallel connection. Right?

Also which solution will be better to balance the cells? I thought solution 1 will be the best. What do you think?

Best regards

[discopope]

I would use a real BMS for this, it would be pretty easy to design with this low cell count, the common ground and
about 17V max. voltage.
You can build it with some analog muxes, a differential amplifier a good voltage reference
and a µc with a good ADC.
For balancing the upper 3 cells you can use some PMosfets with level shifters, the lower ones can be done with
a directly from the µc driven NMosFet. But carefully chose the parts, especially those that are directly wired
to the cells, with small cells the standby/leakage currents are important. And don't leave the lamp uncharged
for month on the shelve.
It can be done with standard parts that are easy to get and with a small µc you have most flexibility in balancing,
SOC, SOH and charger control algorithms.
If you double that circuit, without the balancing part, and redundant charge/discharge switch you have a
second protection and a whole redundant system.
But there are enough second protection chips for 4 cells in series available, as you don't have the need
to change the limits for over- and under voltage, that approach would be simpler and cheaper.
I'm currently modifying some similar, but scalable circuit for use in a system that goes down to 3000m.
It looks somewhat like your solution 2, but it is a matrix with more cells per pack and more packs in parallel
and series. Each pack has it own switch.
If it is important for you to have light even in case one of the cells fails, I would use solution 2
and would keep one line of cells in spare to switch over to this line (with a warning LED flashing) when
line 1 and 2 are empty or fail in another way.

In any case, I never would use larger Lithium cells without independent second protection.

[yann4905]

Thanks for the answers,

A BMS sound pretty hard to make to do for me... My aim is to get the simpliest solution.

It's not important to me to get redundancy.

So I have to find a solution with BMS. Any idea where to get more info?

Best regards

[discopope]

Ok, you want a relative simple to use solution without a lot of design work.
Then I would take a look into the RC model hobbyist market, there should be some ready to integrate solutions
for your application. Since I'm doing my own BMSs for somewhat bigger stuff like AUVs, cars and aircraft
(no, not the Dreamliner  ) I'm not able to recommend a special product, but there is a large
experienced RC-community they should be able to help.

[yann4905]

I didn't think about RC, that's a good idea.

Thanks again.

[amyk]

Do you really need 14.8V at 6600mAh (which is close to 100Wh) or would 3.7V at 6600mAh be enough, and using a boost converter to get a higher voltage?

The reason is that cells in parallel avoid most of the balancing issues, and you can get 6600mAh with 3 x 2200 18650s, which are common.

You could look into repurposing laptop batteries, they already have the protection/monitoring built in and all you need to do is design the case around them.

[yann4905]

Hi,

I really need 14,8V and 6600 mAh. I have 7 CREE XML-2 LED (3,3V each and 3A) with a boost driver.

As you wrote, I can look into laptop batteries. Good idea.

Thanks

[mark03]

I would *not* look into RC batteries.  Made that mistake when choosing Li-ion cells for my Cree XM-L bike light.  Reasons:

1) They are optimized for high current, much higher than you need => heavier for a given AH capacity.
2) They are optimized to deliver power no matter what, up to the point of self-destruction.  After all, if your helicopter crashes, it doesn't really matter whether the battery went up in smoke, does it?  You'll have to get the protection modules and hook those up yourself.  This can be a scary operation when you consider what will happen if you accidentally cross the wrong wires, even momentarily.

[yann4905]

You are right. I looked in RC parts. Allmost all batteries are LiPo, which are the most dangerous. In fact, the best bateries for my case would be LiMn, which are safer, and have a higher rate of charge and discharge cycle. The negative point is that the mAh/cell is not "high" (around 2200 mAh max).

[discopope]

In case of a crash with an internal or an external short, before the fuse, no protection device in the world would prevent the destruction of the cell(s), whether RC or other BMS. The same applies to a developing internal short caused by quality problems of the manufacturer or wrong mounting etc.

There are a few cell types that are more rugged against mistreatment or less harmful in case of failure. But in the end you always have to deal with the stored energy.
I have seen test results from cells with built in over current protection and less aggressive chemistry. If you mistreat them in the 'right' way you get them all to blow off in fire, albeit less spectacular than LiPoly.

A BMS can protect against over charge and -discharge but can provide only limited protection against over temperature and over current in either direction. A fuse with the right max. voltage and current breaking capability and mounted close to the terminals is the only measure for real short circuit and severe charging over current protection.


For this application there are two major scenarios that are dangerous:
1. Unattended charging with the possibility of burning down your house.
2. Explosion of the watertight and somewhat pressure resistant case.

Provisions against No. 1: Don't charge the lamp unattended, even if you have the best protection electronic.
Charge with a window or door nearby where you can safely throw the stuff out. A pair of leather gloves or a metal trash bin can be useful too ;-)

Additionally I would store the lamp away from other flammable stuff and not in my bedroom.

Provision against No.2 can be a bit difficult, you need a blow-out disc that can stand more pressure from the outside than from the inside. Maybe in this case with the small and few cells a simple shuttle valve would be enough. It should be mounted in a way that it blows away from the user.
If the design of the case and gaskets provides venting of inside pressure before the case breaks you don't need a special valve at all.

[peter.mitchell]

here is a multi cell low voltage cut off board i designed a while ago;
https://dl.dropboxusercontent.com/u/677635/battery_monitor_reva.pdf
it's the first rev and i can't remember what changes i made to it, i think i changed the power on section.
its pretty ugly and needs some work, i designed it this way so there was no need for a micro, so you could use different numbers of cells and so the voltages were measured on a per-cell basis
basically, it takes power from the first 2 cells, boosts it up to 38v to run the show
a reference voltage is made simply from a tl431 and some resistors, this is the undervoltage threshold
the opamps are a difference amplifier with an output of the cells individual voltage
the reference and the individual cells voltage are fed into comparators
the output of the comparators are fed into a nand/and gate
the gate drives a transistor that turns the main discharge fet on or off.
the output of the gate also drives a transistor that turns off the boost converter.

[electrodacus]

I will recommend the BMS that I'm working on at the moment but is probably an overkill for your application and not just that but it will only be ready probably at the end of summer if the Kickstarter campaign will be successful. 
As for battery types it will depend what you want. If mobility is most important I will say Lithium-ion, if safety then LiFePO4, if the best price/energy stored during the lifetime of the battery is important then Li-ion charged at 3.92V will last 8x longer.
   

[yann4905]

@electrodacus
Why will your BMS be an overkill?

[poorchava]

In theory if you have good quality cells with integrated PCM such as this one:
[url]http://bto.pl/B2CProdukt.aspx?id_artykulu=43537[\url]
You are relatively safe no matter what. LiPo cells that are meant for RC typically do not have integrated PCMs because of extreme currents that are being drawn.

If you want external PCM, then for highest reliability I would connect cells in series and connect multi-cell capable battery management system to that chain. This enables you to do following stuff:
-measure voltage across every battery
-avoid imbalance currents between parallel-connected cells (because there are none)
-balance the cells should you need to
-detect if a single cell falls below certain voltage

[yann4905]

Ok guys,

I looked into a laptop battery. There are made so: for a 14,8V 6600 mAh, 3 cells of 2200 mAh are mounted in parallel, then 4 "packs" of these cells are in series. then they are linked to a SBS (Smart Battery System) with SMBus. My aim is to not spend a lot of time in battery design for my application. What do you think if I use this Smart Battery System for my application?

[yann4905]

In fact, would it be possible to use the electronic circuit of a laptop battery like PCM, without use of the SMbus?

[Jibby]

The second solution is the correct one.