Help me understand Li-ion balance charging

[wilheldp]

Guys, I'm trying to figure out a Li-ion battery charging circuit.  I have a battery pack that consists of 10, 18650 cells wired in series for a 36V nominal output.  From what I can tell, the charger is a dumb, switch mode, 42V power supply.  I can't find any circuitry to handle the CC/CV charging curve for a proper Lithium battery pack charger.  The battery pack has a two-conductor cable coming out of it which attaches to each end of the series string of cells and to a small PCB.

The PCB has four 3-pin devices (similar size to TO-220, labeled "BOJ70N62") attached to the input voltage.  The center pin of the devices is cut near the package.  The outputs of those four devices are routed through 8 transistors to 10 identical parallel channels.  Each of those channels has the application circuit out of the DL01 battery protection IC, and they are attached to an 11-pin balance cable. 

I understand how the DL01 circuit works to protect the battery, but I'm not sure how the rest of the circuit works.  I cannot find a data sheet on that 3 pin device, and I can't really figure out what it is doing.  My best guess is a voltage regulator to knock the 42V input voltage down to Vcc for the DL01 circuit...but why would the center pin be cut off?

[wilheldp]

I think I figured out what I was missing (literally), but I'm still confused.  There are 3 connections to the charge protection PCB other than the 11-pin balancing connector.  The solder connections are labeled C-, B-, and P-.  From looking at other BMS listings on AliExpress, I see that those connections are for Charging negative (C-), Battery negative (B-), and Discharging negative (P-).  But in the battery pack I'm looking at, the C- terminal is attached to the black wire from the charging cable (correct).  The B- terminal is attached to a red conductor coming from the battery cells (can't tell what it is attached to; may be correct, but probably shouldn't be a red conductor).  The P- terminal is not attached at all. 

The link below is to a very similar, but not identical, PCB to the one I'm looking at. 

http://www.aliexpress.com/item/PCB-PCM-BMS-Protection-Circuit-Module-for-10S-36V-16A-Li-ion-Li-polymer-LifePO4-battery/32348029983.html

[Audioguru]

A protection circuit is not a charger circuit.
A balanced charger measures the voltage of each cell to make sure no cell is charged to a voltage that is higher than 4.20V to prevent a fire.
The cells will each have a different capacity so some cells will be fully charged before other cells and the balancing circuit prevents them from charging to a voltage that is too high.

When a charging cell's voltage reaches 4.20V it is only 70% fully charged then the charger circuit limits its voltage to 4.20V but continues charging and measuring the charging current. When the charging current drops to 1/40th the mAh rating of the cell then the charging of that cell is stopped because then it is fully charged. The balancing charger continues charging the remaining cells and monitoring them.

[Siwastaja]

In addition to what Audioguru said, you must have a properly current-limited source, i.e., a standard lab CC-CV supply. A proper current setting is a must. Li-ion is easily destroyed with excessive charging currents, even when thermal properties would allow that.

If you are unsure, 0.5C is the maximum charging current, i.e., 5A for a 10Ah battery pack and so on.

If you can't find current-controlled supply and only have a voltage supply, you can use beefy series resistors. Calculate the values (ohms and power handling) like you do with LEDs. Charging gets very slow towards the end, so you may need to swap the resistors during the cycle.

It's hard to analyze the balancing circuit from your description only. We would need a full schematic. Terrible hacks exist in this field, so never trust a system before you have analyzed it.

[wilheldp]

To clarify, I am not designing a circuit to charge this battery...I am trying to figure out how an existing circuit works.  It is a hoverboard battery and I am trying to figure out why they keep catching fire.  I have narrowed it down to lack of/or improperly designed charging or balancing circuitry.  I was actually surprised to find battery protection ICs in the pack, but now I think that the whole circuit may be wired wrong (i.e., the discharging negative is not attached to the protection board).  But I can't make any bold claims like that without understanding how the circuitry works.  I understand that the ICs measure the voltage of the individual cells and try to disconnect them from the circuit if they are under or over voltage.  But I don't know how that can happen during charging if the 42V switchmode supply is just attached across the series string of batteries.  Pictures of the actual board I am trying to figure out are attached.

[suicidaleggroll]

I understand that the ICs measure the voltage of the individual cells and try to disconnect them from the circuit if they are under or over voltage.  But I don't know how that can happen during charging if the 42V switchmode supply is just attached across the series string of batteries.

One way is to stick a transistor in parallel with the cell.  When the transistor is off, the current is forced through the cell and it charges normally.  When the transistor is on, current is allowed to bypass the cell and pass through the transistor instead.  So instead of the current going through cell A, B, C, and then back to the battery, if cell B is full then the current goes through cell A, through the balancing cable, through a transistor, back through the balancing cable, through cell C, and then back to the battery.

[Siwastaja]

To clarify, I am not designing a circuit to charge this battery...I am trying to figure out how an existing circuit works.

The circuit doesn't seem like a charger circuit at all! It's either a stand-alone balancer, or a cell-level BMS, but anyway, charger is separate. You need to find and analyze it as well.

It is a hoverboard battery and I am trying to figure out why they keep catching fire.  I have narrowed it down to lack of/or improperly designed charging or balancing circuitry.

How did you narrow it down like that? I'm 99.99% sure it's about cell quality. Proper cells very seldom catch fire even if severely abused by improper charger or balancer.

Of course, improper charge/balance circuit can be an additional factor, but the cells are the root cause, I'm almost sure about that.

OTOH, even high quality cells go into flames if thermally abused, so a broken balancing circuit which produces too much heat and directly heats up the cells to the point of thermal runaway, at about 150 deg C, is a possibility.

There is a possibility that a broken charger causes the fires even with proper cells, but it needs to be a rather extreme case then.

But I don't know how that can happen during charging if the 42V switchmode supply is just attached across the series string of batteries.  Pictures of the actual board I am trying to figure out are attached.

In a properly designed system, the charger (the switchmode supply) has both voltage and current setpoints, so that even if the cell-level voltage monitoring fails, the charger still does pack-level voltage monitoring, so unless there are huge imbalances, no dangerous levels of overcharging happens.

So in any case, you must analyze the whole system, which includes the balancer you posted, and the charger.

With low quality cells, too high a charging current and/or voltage are a lot more dangerous than on high quality cells.

One more thing: the battery management must check that each and every cell is over about 2.0V before allowing the charger to start at all. This is also more important with low quality cells.


Edit: the cells in the picture look a bit like Samsung, does it say so on them? If so, consider the possibility they are counterfeit.

[wilheldp]

I have taken the stock charger apart.  Like I said, it is just a standard switch mode supply.  It outputs 42V at 1.5 amps, and has no feedback from the battery pack.  There is no battery charger IC in the power supply housing and the balance cable does not exit the battery pack.  There is just 2 wires from the charger, through the main control board (with another switch mode converter for low-power rails), to the battery pack. 

I charged it for 1.5 hours last night, and the hoverboard drew 65 watts the entire time.  The power supply heat sink and mains voltage input cap both heated so much that they were uncomfortable to touch (I have my FLIR now, so I plan on doing actual measurements tonight on both the battery and the charger).

The reason I settled on the balancing/BMS board is the fact that it seems to be mis-wired.  I knew the power supply/charger wasn't appropriate, but that shouldn't cause any damage to the cells if the BMS board is wired properly, no? 

I can't tell if the cells are Samsung or not.  I cut through the shrink wrap, but didn't take all of the wrappings off of it because I didn't want it to lose its shape.  Attached is the best pic I could get of the markings on the cells.

[nctnico]

To clarify, I am not designing a circuit to charge this battery...I am trying to figure out how an existing circuit works.  It is a hoverboard battery and I am trying to figure out why they keep catching fire.  I have narrowed it down to lack of/or improperly designed charging or balancing circuitry.  I was actually surprised to find battery protection ICs in the pack, but now I think that the whole circuit may be wired wrong (i.e., the discharging negative is not attached to the protection board).  But I can't make any bold claims like that without understanding how the circuitry works.  I understand that the ICs measure the voltage of the individual cells and try to disconnect them from the circuit if they are under or over voltage.  But I don't know how that can happen during charging if the 42V switchmode supply is just attached across the series string of batteries.  Pictures of the actual board I am trying to figure out are attached.

This circuit looks like a typical BMS to me (I work on a lot of those for one of my customers). A BMS has two functions: protect the cells against abuse which is: deep discharge, overcharge and over current and to balance the cells which is done by discharging cells with a higher voltage than the rest. On some BMSs there is a seperate charge and discharge connection and the discharge connection may allow unlimited charging which potentially could set the pack on fire. Looking at this particular BMS I think the MOSFETs (BOJ70N62) are in a back-to-back configuration but I think the current limit is very crude and only intended to avoid short circuit problems. The cells are probably exposed to a lot of abuse. Also the way the wires are soldered into the board leaves much to be desired.

Regarding charging: In order to charge a battery pack you need a charger made for charging Li-ion and is tailored to the battery pack. A charge cycle consists of charging with constant current and when a certain voltage is reached constant voltage charging. For clarity: the battery charger is supposed to do the constant current/constant voltage cycle!

[wilheldp]

This circuit looks like a typical BMS to me (I work on a lot of those for one of my customers). A BMS has two functions: protect the cells against abuse which is: deep discharge, overcharge and over current and to balance the cells which is done by discharging cells with a higher voltage than the rest. On some BMSs there is a seperate charge and discharge connection and the discharge connection may allow unlimited charging which potentially could set the pack on fire. Looking at this particular BMS I think the MOSFETs (BOJ70N62) are in a back-to-back configuration but I think the current limit is very crude and only intended to avoid short circuit problems. The cells are probably exposed to a lot of abuse. Also the way the wires are soldered into the board leaves much to be desired.

Regarding charging: In order to charge a battery pack you need a charger made for charging Li-ion and is tailored to the battery pack. A charge cycle consists of charging with constant current and when a certain voltage is reached constant voltage charging. For clarity: the battery charger is supposed to do the constant current/constant voltage cycle!

Yeah, I noticed the crappy solder job too, but the wires seem to be pretty well connected.  There's a particularly bad solder job on the switch mode rectifier in the power supply.  One of the legs isn't even attached to the PCB, but the leg is folded over and soldered to a nearby component lead (kinda like point-to-point wiring...only crappier). 

As for this line..."On some BMSs there is a seperate charge and discharge connection and the discharge connection may allow unlimited charging which potentially could set the pack on fire."  Did you mean to say "not connecting" the discharge connection may allow unlimited charging?  Could you explain what failure to connect P- would do to the cells?

I know the cells used in these battery packs aren't the greatest, but I really think there is more to the story than that.  They went to the trouble of putting battery protection circuitry in the pack which should prevent the cells from being overly stressed.  But if the BMS is being bypassed, or if the charger is just overloading the whole battery pack (as opposed to individual cells), the failure rate would make a lot more sense.

[Siwastaja]

I have taken the stock charger apart.  Like I said, it is just a standard switch mode supply.

The difference between "just a standard switch mode supply" and lithium-ion charger is small, but very meaningful: a charger has proper (fairly accurate) constant-current mode.

It outputs 42V at 1.5 amps, and has no feedback from the battery pack.

Have you measured the current that goes into the empty battery? What's the battery configuration, how many in series, how many in parallel, cell capacity?

in the power supply housing and the balance cable does not exit the battery pack.  There is just 2 wires from the charger, through the main control board (with another switch mode converter for low-power rails), to the battery pack.

So it really seems that the module you posted is only a stand-alone balancer which shunts part of the decaying charging current in CV mode (they practically never can shunt the whole current), through a balancing transistor and resistor, around the cells that are nearer to the top than the others.

These kind of stand-alone balancers are difficult to get right and wrongly used, they may cause overcharging. If there is no communication to the charger, the charger CV voltage must be accurately set.

I charged it for 1.5 hours last night, and the hoverboard drew 65 watts the entire time.

I'd suggest you the following test setup:

- Plot current: Measure current flowing into the battery! Plot it as a function of time.
- Plot voltage: If you don't have dozens of multimeters, just probe around one cell at a time during the end of the charge. The voltage doesn't rise suddenly, you have time to probe around with just one multimeter manually. Write down all the cell voltages at the point of time when one of them first hits 4.20V. After that, continue probing to see if any cell is getting over 4.25V, which would indicate a design problem, and over 4.30V would indicate a severe problem.

The power supply heat sink and mains voltage input cap both heated so much that they were uncomfortable to touch

Input cap heating that much sounds like it's not going to live long.

The reason I settled on the balancing/BMS board is the fact that it seems to be mis-wired.  I knew the power supply/charger wasn't appropriate, but that shouldn't cause any damage to the cells if the BMS board is wired properly, no? 

If the "BMS" board is only a balancer, i.e., it cannot stop the charger (by communicating with charger) or isolate the battery pack using relays or semiconductor switches, then the charger can damage the cells.

Even if the BMS board can shut down the charging, charger still can damage the cells if it outputs excessive current; the BMS probably isn't monitoring current.

I can't tell if the cells are Samsung or not.  I cut through the shrink wrap, but didn't take all of the wrappings off of it because I didn't want it to lose its shape.  Attached is the best pic I could get of the markings on the cells.

Doesn't look like Samsung now that you posted a better view. Sorry.

[Siwastaja]

I know the cells used in these battery packs aren't the greatest, but I really think there is more to the story than that.  They went to the trouble of putting battery protection circuitry in the pack which should prevent the cells from being overly stressed.  But if the BMS is being bypassed, or if the charger is just overloading the whole battery pack (as opposed to individual cells), the failure rate would make a lot more sense.

(Note: I have designed a digital, distributed BMS myself)
My experience shows that people are always overconfident in their BMS design skills, but in reality, BMS's often fail by design, i.e., the designer didn't have a clue really what a BMS should do and how. Even when the BMS is OK, it is often miswired so that it can't do its job.

Hence, cell safety is really paramount.

Poor quality cells indeed blow up easily when abused. Abuse increases fire rates on poor cells much quicker than it does on good cells. But that's a bad excuse; those poor quality cells randomly blow up even when not abused! This is why only high-quality, safe li-ion cells should be used. Li-ion chemistries are very tricky to get right, there's a lot of R&D on safety on good brand cells.

It happens, people (EEs designing and installing BMS) make mistakes. That's why there are safety mechanisms - on the good cells.

I think that the combination of good cells and proper BMS should be used, but the safety mechanisms in the cell can do so much more than just a BMS, which is kind of limited by nature.

This design probably has both BMS and cell quality problem. And I say that the cell quality problem is more underlying and more acute. Fixing the BMS probably wouldn't solve the problem, just decrease the rate of fires considerably. Using proper cells would probably solve the problem, even with the broken BMS/charging system.

Proper safety mechanisms are not only voltage and current limits; they are PTC devices incorporated in cells, proper shutdown vent structures (CID), shutdown separators, and everything that is done on the chemistry level using billions and billions on R&D during many years (even decades) at Sony, Panasonic, Samsung etc.

Cell safety mechanisms also help when the BMS can't do anything - for example, in case of short circuit inside the pack, or an object penetrating the cell casing, causing internal short.

If they try to charge the pack with a basic voltage source, then that BMS can't do a thing - they are fucked. That's why I'm suggesting that you measure the charging current. It's the most basic measurement you should really start from.

Charging current is really important. A cell that's fine for 1000 cycles at 0.5C may die in 20 cycles at 1.0C due to lithium plating. And poor quality cell may experience thermal runaway because of it.

[Kilrah]

It is a hoverboard battery and I am trying to figure out why they keep catching fire.

They don't really "keep catching fire"... a couple have and they've been overly mediatized, but it's just like cellphones and laptops before them (and still a couple of times each year).

The basic principle is the charger/PSU is a CC/CV supply and does the charging regulation, while the battery holds the balancing / protection circuit (BMS).

Those that fail are usually nothing but a case of badly copied designs or shitty quality components or assembly. Some guy took one apart and it had remaining solder blobs all over the place, excessively stripped wires just asking to cause a short etc.

These things are so trendy right now that an insane number of manufacturers of questionable quality have jumped on the bandwagon as soon as they could. Some offer them so cheap it would barely pay for the material... can't imagine much effort has been done towards quality / safety. The lowest bidders grab a "good" one from another manufacturers, look at it, try to copy it but taking the cheapest, more or less similarly rated components they can find that day etc without any idea of how / why the original design was done. Welcome to China.

[amyk]

It is a hoverboard battery and I am trying to figure out why they keep catching fire.

They don't really "keep catching fire"... a couple have and they've been overly mediatized, but it's just like cellphones and laptops before them (and still a couple of times each year).

The basic principle is the charger/PSU is a CC/CV supply and does the charging regulation, while the battery holds the balancing / protection circuit (BMS).

Those that fail are usually nothing but a case of badly copied designs or shitty quality components or assembly. Some guy took one apart and it had remaining solder blobs all over the place, excessively stripped wires just asking to cause a short etc.

These things are so trendy right now that an insane number of manufacturers of questionable quality have jumped on the bandwagon as soon as they could. Some offer them so cheap it would barely pay for the material... can't imagine much effort has been done towards quality / safety. The lowest bidders grab a "good" one from another manufacturers, look at it, try to copy it but taking the cheapest, more or less similarly rated components they can find that day etc without any idea of how / why the original design was done. Welcome to China.

AFAIK most of the ones that caught fire were using lipo pouch cells instead of 18650s. Those are far more easily damaged because they don't have a hard metal casing, and will produce the sort of epic flames lion fires are known for. The rest of the fires might not originate in the battery but be caused by something else shorting out and dumping all the energy of the battery into heat.

[d-smes]

I have taken the stock charger apart.  Like I said, it is just a standard switch mode supply.  It outputs 42V at 1.5 amps, and has no feedback from the battery pack.  There is no battery charger IC in the power supply housing and the balance cable does not exit the battery pack.  There is just 2 wires from the charger, through the main control board (with another switch mode converter for low-power rails), to the battery pack. 

Just to be clear, does the battery pack have a 11 wire harness connected to the BMS?  If not, the active balancing won't work and they're probably just using the overcharge & over-current protection functionality.  If so, then there's something else to look at-  For the balancing function to operate correctly for severely mismatched cells, the resistor that is switched in parallel with each cell must be able to divert a fairly large portion of the charging current.  For example, if the charger current limit is 500 mA and each BMS channel can only divert 20 mA, it won't prevent over-voltage / overcharge of a weak cell.  However, if the cells are reasonably well matched, the balancing function will still work.

[wilheldp]

There is an 11-pin balancing connector attached to the BMS board and the DW01 circuit seems to be configured correctly.  I just don't think the power supply is attached to the BMS correctly.

This exact model of hoverboard has caught fire at the battery pack *at least* twice.  I investigate fires in consumer products for a living, and two of these have been on my desk with clear burn patterns originating from the battery pack.  In fact, my company let me buy this thing on my company credit card to investigate it.  They didn't tell me I couldn't ride it around, though.

I really appreciate the input, guys.  I plan on running more detailed testing of the charging and BMS circuits, like Siwastaja suggested.  But I leave today for a week-long vacation.  I was trying to cram as much work as I could in on it before I left.  I will be sure to update this thread when I get better test results.  I should have the EEVBlog Brymen meter by then, too, so I can put it to good use.

[Siwastaja]

BMS design checklist:

[  ] Charging prohibited below 0 deg C temperature
[  ] Charging stops when any cell hits HVC (typical: 4.2V)
[  ] Discharging stops when any cell hits LVC (typical: 2.8V)
[  ] Charging prohibited if any cell below 2.0V (typical), report "broken battery" instead.
[  ] Charging stops above 50 deg C (typical)
[  ] Discharging stops above 60 deg C (typical)

Battery system design checklist:
[  ] Charging current never exceeds the datasheet value (or 0.5C if unsure)
[  ] Discharging current never exceeds the datasheet value (or 1C if unsure)
[  ] BMS can stop charging, if not by communicating with charger, then by isolating the pack
[  ] BMS can stop discharging, if not by communicating with the load, then by isolating the pack
[  ] Physical construction that avoids accidental shorts
[  ] Pack is fused
[  ] Proper cells used (from reputable industry giants)

[Audioguru]

An ordinary Lithium battery cell is not fully charged when its voltage reaches 4.20V, it is only about 70% charged. Its charging voltage must be limited to 4.20V and its current must be monitored so when its charging current drops to 0.025 of its rated mAh then it is fully charged and its charging must be stopped.

[Siwastaja]

An ordinary Lithium battery cell is not fully charged when its voltage reaches 4.20V, it is only about 70% charged.

70% sounds quite extreme.

It depends on cell resistance (i.e., brand new power cells will be more full than some old sloppy energy cells) and charging current. If the charging current is, say, C/20 to begin with, then they'll be at 100% by definition without a CV phase.

My experimentation shows that, for example, new Samsung energy cells are at about 90-92% (IIRC) when 4.20V is reached at 0.5C charge, which is the recommended long-time maximum.

CV phase is there to compensate for voltage drop caused by internal DC resistance, without ever exceeding the cell terminal voltage limit.

[amyk]

There is an 11-pin balancing connector attached to the BMS board and the DW01 circuit seems to be configured correctly.  I just don't think the power supply is attached to the BMS correctly.

This exact model of hoverboard has caught fire at the battery pack *at least* twice.  I investigate fires in consumer products for a living, and two of these have been on my desk with clear burn patterns originating from the battery pack.  In fact, my company let me buy this thing on my company credit card to investigate it.  They didn't tell me I couldn't ride it around, though.

I really appreciate the input, guys.  I plan on running more detailed testing of the charging and BMS circuits, like Siwastaja suggested.  But I leave today for a week-long vacation.  I was trying to cram as much work as I could in on it before I left.  I will be sure to update this thread when I get better test results.  I should have the EEVBlog Brymen meter by then, too, so I can put it to good use.

If this report is to be believed, the INR18650PHH cells are made by "Shenzhen Konta" and the pack with the same or very similar protection circuitry passed some safety tests... how do you know if the battery is the origin of the fire, or was it ignited by something else close to it? Regardless of how much protection it has, the battery will be just as happy to deliver its rated current into a fire-starting resistive short as it is the usual load. If some loose detritus inside falls onto and shorts the cell terminals directly, the protection won't help either. Especially since hoverboards will probably be subject to a lot of physical abuse, I'd look at the other things first.

Reverse-engineering the board would also be a good idea, the single DW01 circuit is pretty common but I haven't seen the schematic for multiple DW01s. I don't think it's a balancer but just disconnects the whole pack if any of the DW01s detect a fault.


I think that 70% number comes from the bogus " battery university" site.

[nctnico]

Lets wait until wilheldp gets back from vacation and posts some pictures of a unit which caught fire.