Tear down of a Hoverboard battery

Mitsubishi i-MiEV Forum

Help Support Mitsubishi i-MiEV Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

PV1

Moderator
Staff member
Joined
Apr 27, 2012
Messages
3,242
Location
Pittsburgh, PA
First off, I know they don't hover, but you feel like you are hovering when you are riding one.

By now, I'm sure many of you are familiar with the Hoverboard fires that are going on. As with the Tesla fires, these are made out to be much more common than they are. Out of an unknown millions of boards sold, there have been up to 52 reported fires. This failure rate is still considerably lower than the percentage of cars that catch on fire, not due to a collision.

Debating the safety and/or concept of hoverboards isn't the main point of this topic, but instead I will be tearing down and inspecting a Hoverboard battery. While there are many different brands out there, this is a battery from the Smart Wheelz Smart Drifting Scooter.

A bit of background. My friend started selling this brand along with his eBikes and solar installations. His demo board started having a sporadic issue where the board would just shut off without warning. Our speculation is that the cells weren't perfectly balanced and under load, the BMS was cutting off the output. The only way we could get the board to come back to life after shutting down was to plug it into the charger for a few seconds. This, according to our hypothesis, was resetting the BMS and allowing the battery to function. It would only cut out during a sudden momentum change (rapid stop in my case, quick turn in my friend's case). Being that the board was exposed to cold weather, we think that came into play as well.

He took the battery out of his demo board and gave it to me to investigate. This thread will detail my findings. I will be opening the pack, measuring the voltage of each pair of cells and their capacity.
 
Right off the bat, I find that these batteries do, in fact, include a balancing BMS (the piece that many of the burnout boards do not have). The battery is made up of 20 18650 cells, wired 2P10S.

Measuring the cell voltages should be pretty easy since there is a connector with leads to each cell plugged into the BMS.

Code:
Cell Voltages
Cell pair 1 - 4.16
Cell pair 2 - 4.13
Cell pair 3 - 4.16
Cell pair 4 - 4.14
Cell pair 5 - 0.17
Cell pair 6 - 4.16
Cell pair 7 - 4.15
Cell pair 8 - 4.15
Cell pair 9 - 4.17
Cell pair 10 - 4.13

:shock:

Well, I think I see why the board was shutting down, but then again, a proper BMS shouldn't allow this pack to perform at all. Cell pair 5 has such a low voltage that the cells must have been reverse charged or close to it. Pack voltage is 37 volts. I actually thought that lead on the connector was the thermistor at first.

Can someone help with figuring out the BMS board?

https://www.dropbox.com/sh/uqswx10gcxv125f/AAC0zGi0WqKp4B_OyJiKpt6ma?dl=0
 
PV1 said:
By now, I'm sure many of you are familiar with the Hoverboard fires that are going on. As with the Tesla fires, these are made out to be much more common than they are. Out of an unknown millions of boards sold, there have been up to 52 reported fires. This failure rate is still considerably lower than the percentage of cars that catch on fire, not due to a collision.
But some of those fires have also burned down the houses the boards were in!

I'm amazed that fires in EV's are such a rare occurrence - If a production EV burned down a house every now and then like the hoverboards are doing, it would set back the EV industry in a big way. Hoverboards are now (thankfully) banned from airplanes and many other places because of their reputation

I'm sure it's probably only a few manufacturers whose quality control lapses are giving all of them a bad name, though it seems several different brands have spontaneously self combusted. At any rate, there needs to be some standards put in place . . . . before my next door neighbors house goes up in flames taking mine with it!! We have agencies like the Underwriters Laboratories who put their stamp of approval on many electrical appliances (which could also burn down your house) and IMO, the hoverboard industry needed something like this before these fires gave them all a bad name

Don
 
It's my understanding that quality hoverboards are indeed UL-listed but that the problems are occurring with the Chinese knockoffs. A properly-designed BMS should provide failsafe protection no matter what the issue with the batteries and charging circuits. This does seem to be the case with all modern EVs (except with the early days of the latest RAV4-EV and its input connector melting).

The coolest feature I've seen is Tesla monitoring ac input current and cutting back the power draw if any fluctuation occurs, which would indicate a high resistance in the source (house) power wiring, thus avoiding an overheating hazard. Hmmm, wonder if they put an error code on the screen alerting the owner?
 
Tesla owners, at least they ones I've met, are keen to watching the input voltage and current.

Yes, some boards have burnt down homes, but how many cars burn up in the garage? I'm not trying to make the hoverboard issue sound minor. It is a major problem that needs fixed, the problem being how flawed merchandise can be sold without undergoing some kind of testing. True, there is the UL listing, but it doesn't catch everything (as evidenced by the knock-off hoverboards).

On that note, it's even surprising that more smartphones aren't going thermal. My Droid Turbo will take its lithium ion battery to 4.4 volts during turbo charging. Standard charging takes it to 4.3+ volts. By comparison, our i-MiEV charges to 4.105 volts per cell. A fully charged lithium ion battery's voltage is 4.2 volts.

After seeing .17 volts on a pair of the cells in this battery, I'm now tempted to take the battery out of my board and check its balance. I feel confident that it is balanced, being that I have torture tested it twice (very hard riding for 4 hours straight both times) and it hasn't given me any signs of a problem yet.

On the next day that we have decent weather and I have time, I'm going to try to remove these two cells from the rest of the pack and try to recover them. .1 amp charge and far away from the house, just to be safe.

Down the road, I want to test the BMS for balancing, low and high-voltage cutoff, and see if there is over-current protection.

Battery Space sells a replacement hoverboard battery with UN 38.3 certification pending. It contains their BMS and 20 Panasonic cells, and actually has almost twice the capacity of the battery I'm working on.
 
i would be happy to help with the BMS board, but would need hi-res photos to identify the components. Better yet would be an actual board in my hands to trace the schematic.
 
KiEV, I added a picture of the board to the link above. I didn't see anything on the bottom side of the board, but I didn't try removing yet it since it is glued down. If you need pictures of the bottom side, let me know.

It looks like it controls the negative output for LVC.
 
I purchased an electric bicycle with a dead battery. The original cells were shot, but in replacing the cells with smaller and newer Samsung INR 18650s I also ended up ditching the BMS altogether. The reason was that I found it was putting out a LOT of heat when top balancing the cells. Stacked right against the cells with this extreme heat it's no wonder there have been issues with packs burning up houses.

Instead I just balanced the cells by hand and wired a third wire into the charging connector, which had an unused pin. The third wire I connected mid-pack, so I can at any time compare the pack halves. If all cells are good the voltages should always be equal. On the other hand if a cell goes bad it's very likely to drop in voltage and it would then be immediately noticeable on the third wire.

So far I've registered at most 0.01 voltage between the pack halves. The pack is 3p10s, so when full it's 42 volts for all and 21 volts per half. The charger tapers by itself and stops charging when it can't keep the voltage at 4.2 volts per cell, so charging is not a problem. You need to make sure that you don't drain the cells completely, but that's easy to do by watching the 5 bar meter on the bicycle. Just stop if it's down to zero bars. Never reached more that 2 bars though.

Now that there is absolutely nothing connected to the cells when it's off the bike and off the charger it's as safe as a 18650 can be. Nothing is draining any current and the cells are completely disconnected from everything, which also means it will retain it's charge practically forever. No need to up the charge even if you leave it for weeks and the pack will never go bad by itself, even if I leave for months.
 
Manufacturer: Shenzhen Some Electronics Co., those labels are funny

Howdy Tim,

The second photo has good lighting and shows that there are 10 identical circuits (probably for balancing). These are controlled by the U1,U2,...U91, Ua1 ic's and each one has two transistors Qn1, Qn2 with 4 bleed resistors, maybe you can check if they are "901" or "106" in value, 900 or 1M Ohm.

Since cell pair 5 was out of family, let's trace that circuit thru from the harness to the U5 ic and the Q51 and Q52 pair, and check R51-R55. i would guess that either the ic is toast or one of the Q transistors is shorted and drained that cell pair.

There is another circuit that is numbered with the 100's series, e.g. Q100-103, R100-123, that likely drives the base of the 4 big transistors Q?,Q2,Q4,Q5 with the white glue. This circuit probably all acts together like a relay to open or close the negative or return side of the pack between B- and C-

Would you be able to peel or scrape off the white glue on those 4 big transistor devices and read the part numbers? Also maybe try to read the part numbers of the control ic's U11-U91,Ua1--they are all likely the same. Once we get the pn and a datasheet it will be easy to figure out the logic.
 
Large transistors are:

K80E08K3

I started a recovery charge on cell pair 5. Voltage on that pair before starting was .18 volts. So far, after a .1 amp charge, I've gotten the cells to 2.5 volts. I've since switched over to a .2 amp charge, with the voltage currently at 2.7 and climbing. I haven't detected any temperature increase, but I don't have a thermal image camera to tell. Just going by feel, they are the same as the other cells.

I actually had to use the NiMH charge profile since my charger wouldn't recognize the battery under the lithium ion profile. I have a volt meter connected and haven't left it unattended.

One the cells get above 3 volts, I may try to read the internal resistance.
 
So far, I'm now at 3.45 volts. Interestingly enough, internal resistance matches the resistance of other cell pairs at 59 milliohms. Remembering that this is 2 cells in parallel, how does that compare to new lithium ion cells?

Could I be lucky enough to have an entirely good pack? At .5 amp charge, the voltage changes .03 volts between charge and idle.
 
Cells may also appear completely healthy when charged, but immediately fail and plunge in voltage when put under load.

Generally though, at least with LiFePO4-cells, if you bottom balance them and leave them be you can spot bad ones. They're the ones that don't recover in voltage, but instead keep dropping. A healthy Lithium cell doesn't go down in voltage after discharged and disconnected.
 
I understand that. Actually, as far as SoC goes, I'm somewhere below 20% (3.54 volts idle), so I have some ways to go yet. It just surprised me that internal resistance didn't change after being at such a low voltage. The cells are .02 volts below what they were last night a couple of hours after coming off charge, according to my Harbor Freight multimeter.

Tonight, I'll be finishing charging, then do a discharge test if I have time. After I finish the discharge test, a recharge, and a load test, I'll put this pack in my Hoverboard and test it if everything goes well. If it works, and my friend doesn't want it back, I have a new battery for my lawn mower. I estimate that with this pack and the stock battery running parallel, I should be able to mow for an hour and reduce the stress on the stock battery as well.
 
They may seem to recover from an over-discharge event but it is only an illusion--don't be fooled, replace that pair of cells if you want to use the rest of the pack.

Metal dendrites form when re-charging a depleted cell that can puncture the separator sheet causing a short circuit and lots of heat--that's how these things catch on fire...

The large transistors are N-channel MOSFETs rated to handle 80 Amps.

Any markings from the control ic's?
 
I got the cells up to 3.7 volts a couple of days ago. Unfortunately, I just checked them, and the voltage has dropped to 3.6 volts. Self-discharge is pretty significant, which may be why these cells were so low to begin with. Early i-MiEV packs had problems with random cells suddenly self-discharging, so do Hoverboard batteries. Mass production isn't perfect.

I think that this pack will be relegated to display use for now until I can remove this pair of cells from the rest of the pack. The last thing I want is these cells to go thermal and take the whole pack with them. Once they are separated, then I'll go for full charge, maybe down by the road away from everything else ;) .

So, what can I do with a 9s pack?

As for the BMS, the underside is marked with PEP-10s-20a-01. I get what the 10s and 20a stand for, but searching that string returned nothing. I'll try to get more of the markings once I disassemble the pack.
 
i'm thinking that the little 6-pin control chips (U1..A) are the Fortune DW01 Lithium battery protection ic which has built-in protection limits for overcharge detection @4.250+/-.05 and overdischarge @2.40+/-0.1, plus overcurrent detection. The current limits are cleverly determined by the ON state resistance of the large MOSFETs previously identified. Here is a link to the DW01 Datasheet.

Two of the large FETs are used in parallel to open the low-side in the event of overdischarge (designated P-) and the other two open the low-side for overcharge (C-). In normal operation all 4 are commanded ON which completes the return path for current to B-.
 
Would you check if C51 is shorted as this would cause the cell pair to discharge. You may need a wooden chop stick or an xacto blade to lightly scrape off the conformal coating on the silver ends of C51. Put your HF (hazard fraught) multimeter on the resistance setting and touch the leads to freshly exposed metal ends of C51--it should read open or very very high ohms, if it reads under 10k Ohms then you have found the culprit.
 
I got 51.7 on the 200k setting, which I'm assuming is 51.7 kOhms.

C41 gave me 54 kOhms for comparison.

Voltage is at 3.59 volts (almost no drop from the last check).
 
When you charge and disconnect a cell it will drop in voltage a little when the "surface charge" has dissipated and there's nothing wrong about that.

If you discharge the cell and it goes down in voltage afterwards, it's a bad cell. A good cell will probably bounce back a little.
 
Back
Top