LiCoO 2 Fuel Gauge, SOG, Protection and more (newbee)

Discussion in 'Power Electronics' started by lemon6949, May 21, 2019.

  1. lemon6949

    Thread Starter New Member

    May 21, 2019

    Im completely new to the forum, and worst yet completely new to electronic circuits so I have a massive learning curve, So I apologies in advance and will say don't be gentle with me.

    I wanted to understand how we could better manage large... really large packs of lithium Ion batteries.
    Im not building an ebike, or a diy powerwall but have taken an interest in what those guys are doing, and have to say it scares me, and I wanted to know if there was a better way to do it.

    Background.. wrong, right or the ugly.
    Most "DIY/Makers" playing with lithium ion batteries and building packs (they old 7s100p) are relying on the nature of LiOn to "Self" balance when in series, and fuses (fusing each cell) as there safe guards, with BMS providing either/or or both balancing, over and under charge protection.

    I can list a who bunch of reasons why this is scary.
    - This doesn't take into account the small difference in chemistry between batteries
    - defects within the batteries
    - batteries in series can still become unbalanced, and can result in over charging or under charging
    - only offers protection to over current, circuit shorts
    - provides no isolation at a cell level
    - provides no thermal protection (for example ensuring the right charging conditions, or isolation from thermal runway)
    - fuses only don't prevent thermal runaway (side note its funning most reference tesla style packs and the same protection "fuses" but completely overlook the radiator in the pack which would be dissipate heat in an attempt to prevent venting via flames in some failure scenarios but also provide heat in low temperature charging scenarios)

    So it got me interested in my childhood interest in electronic circuits, some 30 years ago, and I started to wonder .

    So I actually want to build a battery pack that was 14s10p (42v-58v 52v nominal) @ 20AH how would I go about doing it (factoring in all the safe guidelines for LiOn".
    as I looked into it, found that most that there really isn't a lot for managing large packs, with most IC designed around 4~ish cells, 1s..2s..3s..4s, there are two options for 12 and 14 cells but they are really expensive options when building large packs, I also found that most protection IC charge to 4.2 (instead of 80% 4.15) and discharge to 2.7 (instead of the 3-3.5v that some popular cell brands support/recommend)

    I was thinking that it needs to include
    - SOC
    - Temperature
    - Passive balancing
    - Voltage/Heat per cell (or at least every few cells)
    - the ability to isolate per cell
    - hold timers for charge and discharge (the idea of charging and discharging a LiOn battery at the same time doesn't seem like a sensible one to me)
    - minimal power drain (maybe sleep between sampling)

    so what is a cost effective way that people would manage this?

    something like MAX17015 or bq2405xon the pack to monitor/fuel gauge the overall SoC
    fast/large/accurate ADC to monitor each individual cell voltage and temperature (instead of more expensive MC33771)
    resistance protectors ( like Cyntec protectors)

    what components would people recommend I look at.
  2. Zaishens

    New Member

    May 20, 2019
    First off all, you say "relying on the nature of LiOn to "Self" balance when in series, and fuses (fusing each cell) as there safe guards, with BMS providing either/or or both balancing, over and under charge protection"

    This is contradictory, if the BMS does the balancing, you are not replying on "Self" balance.

    The BMS is more of a safeguard than an actual active component, your controller should know what current and till what voltage the pack should be discharged at. The charger should know the current and voltage the pack should be charged at. The BMS takes care of fault conditions like short circuit, unbalance and if the controller or charger fails, that's why the delta V on the BMS is so high (4.2-2.7).

    there are BMS/monitoring systems that also have thermal protection, SoC etc, although they are somewhat expensive, for example google "batrium".

    "provides no isolation at a cell level" I dont know exactly what you mean by this or why you would need it. The tubing and air seperation is mostly condidered enough isolation.

    Cooling can be done best by tab cooling at the terminals of the cell, because over the outer shell doesnt penetrate well into the battery due to the rolled contruction (high thermal resistance).

    The MAX17015 doesnt seem able to handle more than 4 cells, the bq2405xon i cannot find.

    As for as components go, i would put a simple yet sturdy BMS like: link
    For the charger, any 58V (14*4.15) SMPS will do that you can limit in current and cut-off at +/- 10% current.
    For the monitoring, you can simply measure voltage and determine a 100 and 0% voltage, or you have to measure the ingoing and outgoing energy (voltage*current*time)
    Thermal protection is relatively easy to do diy, take a few thermocouples or NTC and use that signal to enable/disable the charger and controller.

    btw, you cannot charge and discharge at the same time, current can only flow in or out the battery, it is possible that this is done in (repadily) successing instances, which should be avoided, this can be done with starting to charge batteries that are lower than X volts and making sure that charge current is higher than the load current, so effectivly the load is powered by the charger and not the batteries during charging. Placing some capasitors in parralell with the whole pack is another good measure te reduce peak and ripple currents on batteries which prevents losses and thermal build up in batteries in some occaisions.

    My 2 cents, any other questions?
  3. lemon6949

    Thread Starter New Member

    May 21, 2019
    Hey Zaishens,

    BMS like batrium (in most deployments) are preforming pack level balancing, SoC and protection not cell level.
    As a fuel gauge it unclear to me if a batrium only does voltage+temperature SoC "as it control logic is configured around voltage" and battery condition or uses other measures and algorithms

    It is possible due to a number of reasons for a single 18650 to become unbalanced, especially in the DIY powerwalls as the builders are using dissimilar chemistry cells in a pack, differences in chemistry or internal failures/age/heat of a cell can result in a weaker cell in a pack (by pack Im referencing cells in series). That weak cell may not fail immediately but end up in an under voltage or overcharge situation possible ending in venting with flames.

    Im talking about options to design a battery pack where every cell is monitor and has the option to be isolate should a cell develop a fault.
    Isolation meaning electronically disconnection from the pack, as opposed to thermally decoupling the cell from other cells in the pack

    bq2405xon is a typo the model is bq2405x
    and MAX17015 was a typo too for MAX17215
    both MAX17015 and bq2405x chips track state of charge but typically are targeted to 4 cells, so its a question what other options are there, or tricking them to thinking they are managing large capacity cells on the FE and other protection on the cell at the BE (by way of voltage/thermal monitoring and resistance protectors)

    By "provides no isolation at a cell level" I mean if a cell was found to have faulted, be in a serve under voltage state, or over temperature state it would be isolated from the pack (disconnected) either indefinitely or for a hold time,

    The reason for an accurate SoC is because LiOn batteries can not accurately be measured based on voltage alone in most operating states.

    By charging and discharging at the same time, as you pointed out I meant rapid succession between charge and discharge states, as far as I can tell most BMS including batrium don't including any hold time between charge and discharge cycles, and a number of these DIY powerwalls are "stressing" the LiOn batteries

    Im was looking at a way to build a smart battery pack that focus on safer approach to managing such large (multiple cell) packs
  4. lemon6949

    Thread Starter New Member

    May 21, 2019
    with regards to surface cooling vs tab cooling.

    Given the design of an 18650 cell I wonder if thermal management/cooling on the negative tab only would be as affective as cooling on both tabs.

    The positive tab would be cooling any gas within the pressure compartment where the negative tab would radiate through the cell
    most images of a thermal runaway venting with flames gas the heat initially building up towards the centre of the cell

    certainly designing a battery pack with thermal management on the negative table would be easy to fabricate, and flame/thermal resin to encase the battery to restrict the risk catastrophic failure of the pack in the event of thermal runaway and flames
  5. Zaishens

    New Member

    May 20, 2019
    Any BMS does in fact do balance at cell level, that's why you have 15 connections with a 14SXXP pack (14 voltages+GND). Each level gets individually balanced (gets a bypass current when it's voltage reaches a certain overvoltage treshhold). So that cells in series will be topped off to the same voltage. The cells in pararell act as one "cell" level and will always be at same voltage (not current).

    Now i understand that you mean disconnect instead of electrical isolation (it's similar but confusing terminology). This is normally not done because it increases the cost and losses of a pack significantly to include for example a MOSFET to each individual cell. The fusewire that is sometimes used does provide this function, but only if the current is large enough (not neccecarily only in short circuits).

    Mixing weaker and stronger cells is not really an issue, the stronger ones will supply more current (because of the lower resistance), but the weaker ones will still participate with some current. Thermal runaway due to damaged cells and the resulting leakge current in parralell will remain a risk, unless you put themperature indication on each individual cell and disconnect the cell with a MOSFET. Some cells have a protection circuit attached to the top of the cell, recognisable by the negtive terminal running to the top over the side of the battery. If you want to use those, you have some built-in protection.

    Voltage is indeed not the best parameter for SoC, as I said, you have to measure energy in/out then instead. A microcontroller could be used for this, but probably there are dedicated IC's that i could find for it.

    Depending on application, i think a holding time is not as effectibe as a hysteresis for charging (start charging only when battery is below a certain voltage and stop charging when above 4.15 volt). This is easy to implement (schmitt-trigger). And the capacitor for intermittent discharging is always a good idea.

    I dont think its good or usefull to trick a chip into thinking it's charging only 4 cells... you introduce more room for faults and loose some resolution (14/4=3,5 times less accurate). The BMS i linked with a max 48V charging with current max and cut-off would be more fool prove.
  6. Zaishens

    New Member

    May 20, 2019
    Well, i think the positive tab is actually more effective to cool because this is the one connected to the inner layers and centre pin where more heat is concentrated than the negative tab, which is connected to the outer layers of the cell, see: link

    And this gas your talking about has a much worse thermal conductivity than metal, so more difficult to cool. But the aim is to prevent any gas forming actually, so this point is rather mute.
  7. Yaakov

    Well-Known Member

    Jan 27, 2019
    Welcome to AAC.

    Your project sounds very interesting and I am sympathetic to the goal of increased safety.

    Rather than address the technicalities of implementation, I wanted to step back and talk about design. I expect that you are already aware of a lot of this, so please don’t be offended if I am covering things you already know.

    When it comes to engineering a device, design comes first. Of course, you might have a rough idea and cobble together prototypes, but before that you’ve got an idea for something new or improved, or sometimes just to copy something you can’t get the details about. In any of these cases, though, you have to start out with a specification, and that would include broad goals and a cost-benefit analysis.

    In this case, you have a goal to increase safety. That’s great, so first you need to have some measure of the current hazard. That measure can tell you how much it is reasonable to spend to decrease it by some factor.

    Obviously there are many, many engineers and manufacturers designing packs and components for them, and safety is high on the list of requirements if for no other reason than liability. This means they have done some studies to decide what sort of protection mechanisms are required for a reasonable price and outcome.

    Sometimes, a risk has a high likelihood but its outcome is relatively minimal; sometimes the likelihood is very low but the outcome is catastrophic; and there are all sorts of combinations between these. Obviously, these balances determine something about how much it is worth spending to mitigate the risk.

    Since so many companies are making these things, you have to ask yourself why they aren’t selling something like what you are proposing. The probable answer is they couldn’t sell them at a profit.

    Now, of course, there are always niche markets. For example, something going into space or underwater might need much more reliability and safety and so the price wouldn’t deter them. But for the most part, those sorts of consumers have bespoke solutions.

    This brings us to you. Your desire for increased safety is a niche because it will cost substantially more to build the sort of thing you are proposing. This means you aren’t likely to find components that already manage batteries which will provide the functionality, and cobbling them together seems to run afoul of the safety goal.

    So, you are stuck with a bespoke solution. That being the case, I would back up and ignore the current strategies entirely for the first round of design. Figure out what your product would look like if there was a chip made to do it. Work that out, then look for components that fit as subsystems in that design. In other words, don’t design from the bottom up but the top down.

    Of course, all engineering is about optimization, so you will probably have to make compromises when you actually try to build it, but if you treat your design as a series of functional subsystems (e.g.: SoC, configuration switching (series/parallel/isolated), etc.), then you can look for components that will fit your design rather than fitting your design to the components. You may be surprised at where the parts come from, they may not be initially designed for battery management but something else.

    In the end, you are creating a niche product. It will never be as cheap as the conventional method, maybe not even by an order of magnitude. It may not actually offer substantially more safety if you look at rates and modes of failures. Yes, there are fires, etc., but how often compared to the enormous number of cells in use?

    This is not to discourage but to give you some of my experience as it relates to your project. I don’t know if any of this is new to you or informative, but I wish you luck.