E+ Battery died on me while riding

[Chriull]

2 hours ago, Keith said:

With the relatively small capacity of Ninebot's, I really cannot believe that Ninebot have no balancing.

It would make Ninebot, the crappeast EUC out there and loads of people would be getting very poor battery life. I'm pretty sure Ninebots must have balancing within the BMS, This is very easy to prove, if the BMS has one wire/connection  going to every single series cell - it will have balancing.

Every BMS has (should have/needs to have/...) wire/connections to each single cell to check for over/undervoltage. But:

I'm not 100% sure if i'd still follow strictly my conclusion stated in this post.

If this value for Rvcx is true (which i have no reason to doubt) the only way to balance the cells without overcharging some cells would be to charge with a constant current of about 4 mA. Which would lead to a balancing time of ~3000mAh/4mA=750 hours....

So the best advice should still be imho to open up the pack and charge the cells individually if some have undervoltage and are still alive.

2 hours ago, Keith said:

@HunkaHunkaBurningLove, is spot on, 90% of the charge happens very quickly but it is that last 10% that takes several hours to complete and includes the balancing as most BMS do not actively balance during charge they simply clamp the first cells to reach full charge whilst the others catch up. The pack absolutely needs an, at least occasional, full charge plus an hour or two after the charge LED goes green.

Yes. But with such a BMS without external balancing circuit there is just not enough balancing current possible so the "stronger" cells get overcharged and the BMS cuts off charging in because of cell overvoltage. But at least whilst this time the "weaker" cells get a bit more charge.

[Keith]

@Chriull,

18 minutes ago, Chriull said:

If this value for Rvcx is true (which i have no reason to doubt) the only way to balance the cells without overcharging some cells would be to charge with a constant current of about 4 mA. Which would lead to a balancing time of ~3000mAh/4mA=750 hours....

That is a very poor extrapolation. The charger will top out at a nominal 4.2V per cell so 63V for a Ninebot. When the battery also reaches 63V as well  no current will flow and, more to the point, current progressively, lowers over time once the charger hits 63V and the battery total voltage gets nearer to that figure. so it is absolute nonsense to suggest the entire charge would need to be at 4mA. Even assuming that the design is so poor at it can only support a 4mA shunt Let us say that at the point the charger is at 63V one cell is right down at 4V, then all the others are merely  at 4.21V way to low for the BMS to cut charge, even if it is programmed to do so. Even at 4mA, that lower cell will top up slowly and no other cell will get any higher than 4.21V. That is why it is important to leave the charger on after it goes green. In practice few BMS, even with components external to the balance chip can handle more than 50mA or so, they still balance fine as long as they are given enough time to do so. A Charge Doctor is very useful in monitoring that. What is poor in this design is that there would be REAL ISSUES if (say) half the cells were badly out of balance as either over voltage or BMS protection would do harm to the pack. However, that would mean the pack is landfill anyway.

[Chriull]

21 hours ago, Keith said:

@Chriull,

That is a very poor extrapolation. The charger will top out at a nominal 4.2V per cell so 63V for a Ninebot. When the battery also reaches 63V as well  no current will flow and, more to the point, current progressively, lowers over time once the charger hits 63V and the battery total voltage gets nearer to that figure. so it is absolute nonsense to suggest the entire charge would need to be at 4mA. Even assuming that the design is so poor at it can only support a 4mA shunt Let us say that at the point the charger is at 63V one cell is right down at 4V, then all the others are merely  at 4.21V way to low for the BMS to cut charge, even if it is programmed to do so. Even at 4mA, that lower cell will top up slowly and no other cell will get any higher than 4.21V. That is why it is important to leave the charger on after it goes green. In practice few BMS, even with components external to the balance chip can handle more than 50mA or so, they still balance fine as long as they are given enough time to do so. A Charge Doctor is very useful in monitoring that. What is poor in this design is that there would be REAL ISSUES if (say) half the cells were badly out of balance as either over voltage or BMS protection would do harm to the pack. However, that would mean the pack is landfill anyway.

Imho the current of the chargers after the charging cycle stays somewhere in the 100mA region? With the "shunting capacity of" ~4mA for balancing the "balancing capacity" of the BMS is 4mAh per hour. If the charging current is more than this "shunting capacity" of the BMS, the other already full cells will get charged furtheron. So the weak cell gets 104mAh per hour and the stronger cells 96mAh.

An external balancing circuit which can shunt currents in the region of the chargers "balancing current" would make sense and be a profound design decission - one could balance a pack "forever" until the cells really reach the same capacity without "any danger".

My Ninebot charged the cells to a max of 4,156V, thought they made a reserve in case one starts downhill (regenerative breaking). Could also be a reserve for balancing, so the other cells get no real overcharge while balancing. Could also be, that i just got a charger with the max voltage adjusted a bit low... Also i have no idea how accurate the internal voltage measurement is.

Anyhow there should be no danger of some cells getting overcharged, since the BMS checks all single cell voltages and cuts the charger off...

Ps.: so i stroke out my "conclusion" in the old linked post since the also quite low balancing could/should help and there is no overcharging danger...

[noisycarlos]

Well. I got a Multimeter. 

I'll preface this post saying that it's my first time using a multimeter. I know very basic stuff about electricity and batteries, so I was very careful and followed the advice on this post:

However, I generally have no idea what I'm doing here. I took a picture of how I had the multimeter set, in case I did something wrong.

Now to the results:

The reading of the main connectors out of the pack was a whooping 5.8. Not 58, 5.8. So something is definitely wrong somewhere in the pack (not sure where). I took the reading three times... same thing.

Additionally, I measured the big contacts in the middle of the pack, in case it means something to any of you. I put the black probe on the far left, and took a reading against the other contacts (see picture, marked on black and red). I also tried readings on the little contacts, but got 0.0 (marked in white).

Not sure what it all means. But such a low voltage from the overall pack surprised me. Especially, since when it was plugged to the NB, the it read a full charge. I borrow the multimeter. So let me know if I did something wrong, or if there's somewhere else I could take a reading to get some insight. Thanks guys!

 

[Hunka Hunka Burning Love]

What's the voltage reading at the power plug (the yellow XT60)?  Did you try charging it up for a few extra hours inside the bot?

[Chriull]

1 hour ago, noisycarlos said:

The reading of the main connectors out of the pack was a whooping 5.8. Not 58, 5.8. So something is definitely wrong somewhere in the pack (not sure where). I took the reading three times... same thing.

Additionally, I measured the big contacts in the middle of the pack, in case it means something to any of you. I put the black probe on the far left, and took a reading against the other contacts (see picture, marked on black and red). I also tried readings on the little contacts, but got 0.0 (marked in white).

The little contacts are for the (not mounted) components for "external" balancing.

Your measurements are some of the battery cell conectors (V3,V7,V11,V13). To get the state of each cell pair you need to measure all of this points (you left out the top and bottom ones). You should have B-, V0 or V1 (the point you had your black probe), depends on numbering scheme and the following V1-15 or V2-V16.

The measurements you made: V13 (32,5V) and V11(24,2V) with a difference of 8,3V for 2 cell pairs could be 8,3V/2=4,15V per cell pair. Which seem ok.

V11 24,2V and V7 16,6V. Here is a difference of 7.6V for 4 cell pairs - so in average 1.9V per cell pair. (or 2 cell pairs almost full and 2 cell pairs absolutely empty...) Here you have some dead/bad/empty cells.

V7 16,6V and V3 8,2V. Here you have 2,1V per cell pair in average. Again too low.

Quote

Not sure what it all means. But such a low voltage from the overall pack surprised me. Especially, since when it was plugged to the NB, the it read a full charge. I borrow the multimeter. So let me know if I did something wrong, or if there's somewhere else I could take a reading to get some insight. Thanks guys!

I'd assume with the low voltages of some of the cell pairs the controller detected the dangerous undervoltage and cut the output off. So the internal resistance of your multimeter and the switched off Mosfets form a voltage divider - the 5.8V are the result...

So till now you could have about 8 cell pairs with undervoltage. If you measure all of these points and one gets all individual cell voltages one could tell if some of these pairs are just dead or maybe could be revived if individually charged (i don't know the voltage limits - some here should know or one could look at batteryuniversity.com).

Edit: Just found the relevant site: 

http://batteryuniversity.com/learn/article/low_voltage_cut_off

" Depending on the manufacturer, the protection circuit of a Li-ion cuts off between 2.2 and 2.9V/cell. "

" Some battery chargers and analyzers (including Cadex), feature a wake-up feature or “boost” to reactivate and recharge batteries that have fallen asleep. "

" Do not boost lithium-based batteries back to life that have dwelled below 1.5V/cell for a week or longer.  Copper shunts may have formed inside the cells that can lead to a partial or total electrical short. "

"A sleeping Li-ion does not reveal the voltage, and boosting must be done with awareness. "

So it could be that you have already some "sleeping cells" where the protection already triggered and so you can't measure a meaningful full output voltage. If you can measure at all the points (B-/V0,V1-V15 or B-/V1-,V2-V16 depending on the numbering scheme) valid voltages and the individual cell pair voltages are above the mentioned 1.5V/cell you could have chances to revive the pack.

For this you would need a charger for single li ion cells where you put the charging wires directly to the right points off the BMS board. ( I'd assume this should work without sideeffects, but i can't guarantee that for sure. Maybe @esaj, @hobby16, @Cranium or other forum members could give some more profound advice).

As stated at this batteryuniversity.com site it could be that a normal charger won't work and you'd need one with this "boosting" "wake-up" capability.

 

[Chriull]

15 minutes ago, HunkaHunkaBurningLove said:

What's the voltage reading at the power plug (the yellow XT60)? 

The charger plug (with the yellow wire) should show nothing - there is the protection diode preventing any meaningfull measuring.

15 minutes ago, HunkaHunkaBurningLove said:

Did you try charging it up for a few extra hours inside the bot?

I would not charge this pack anymore - there could be too many cells already outside of any specs and perform bad things when charged. And also in case the low voltage cells could be recharged i'd assume they have to be recharged individually - the cell voltage differences are way beyond the capabilities of the ninebot bms balancing circuitry.

Could also be, that already the charge side protection is activated and normal charging is not possible anymore...

[noisycarlos]

Thanks @Chriullfor the details.

@HunkaHunkaBurningLove, I did leave it charging for a couple of hours when it first died, hoping it would help. That's when the NB showed full charge (when turned on plugged to the charger), but alas, it still didn't turn the bot on with just the battery. It beeped like it was going to turn on, but nothing happened after that.

Just to take the lesson, would it be a good guess that using the timer to charge it most of the time unbalanced the cells this much? Or do you guys think some cells just went bad?

(Maybe @esaj would like to pitch in)

All repairs that include soldering cells back on the pack are far beyond my skill level, and while it would be fun to try, my yearning is not big enough to risk a fire.

I'd be happy to try/measure more stuff for the post-mortem though. Let me know if you have ideas.

Otherwise, I think I'll just salvage the good cells and use them for flashlights or something like that. But I'll take readings of the cells as I dismantle them and report back.

[esaj]

2 hours ago, noisycarlos said:

Now to the results:

The reading of the main connectors out of the pack was a whooping 5.8. Not 58, 5.8. So something is definitely wrong somewhere in the pack (not sure where). I took the reading three times... same thing.

When comparing that reading to what you got from the cell-tabs (the big contacts you measured straight from the board), it sounds like the undervoltage protection has triggered and shut-off the output (it won't "release" until it sees high enough voltage coming from the pack). With the large input impedance (well, resistance for DC) of a multimeter, you will see some "ghost" voltage due to very small leak currents. I saw similarly readings of a volt or two on my home-made computer ATX-based power supply, which used mosfets to turn outputs on/off. Placing a small load on the outputs (100k resistor), the voltage disappeared, ie. it was just some leak current. Not that you should try it here, but just an example.

 

Quote

Additionally, I measured the big contacts in the middle of the pack, in case it means something to any of you. I put the black probe on the far left, and took a reading against the other contacts (see picture, marked on black and red).

After staring at the BMS-picture long enough, I partially disagree with @Chriull 

1 hour ago, Chriull said:

V11 24,2V and V7 16,6V. Here is a difference of 7.6V for 4 cell pairs - so in average 1.9V per cell pair. (or 2 cell pairs almost full and 2 cell pairs absolutely empty...) Here you have some dead/bad/empty cells.

I think it's not V7, but V9. V7 seems to be left under that black insulator on the side. So there's two cell pairs between V9 and V11, giving 7.6V / 2 = 3.8V per cell on average.

 

Quote

V7 16,6V and V3 8,2V. Here you have 2,1V per cell pair in average. Again too low.

This however does look like there's more severe voltage difference. If the 16.6V -point is indeed V9 and not V7, the difference is even greater...  That would mean there are 6 cell pairs between those points, and then it would be only  16.6V-8.2V / 6 cell pairs = 1.4V per cell pair. Either there's more than meets the eye here, or there are several dead cells there. Or some cells are bypassed by the BMS due to short or damaged components?

 

Quote

I also tried readings on the little contacts, but got 0.0 (marked in white).

It would seem that these are some optional(?) resistors that were left out from the BMS. "Funnily" though, at a glance it would seem that those could be used for cell-measurement / balancing, like Chriull suggested? Maybe it's true that Ninebot BMS doesn't have balancing...

 

1 hour ago, Chriull said:

So till now you could have about 8 cell pairs with undervoltage. If you measure all of these points and one gets all individual cell voltages one could tell if some of these pairs are just dead or maybe could be revived if individually charged (i don't know the voltage limits - some here should know or one could look at batteryuniversity.com).

Edit: Just found the relevant site: 

http://batteryuniversity.com/learn/article/low_voltage_cut_off

" Depending on the manufacturer, the protection circuit of a Li-ion cuts off between 2.2 and 2.9V/cell. "

" Some battery chargers and analyzers (including Cadex), feature a wake-up feature or “boost” to reactivate and recharge batteries that have fallen asleep. "

" Do not boost lithium-based batteries back to life that have dwelled below 1.5V/cell for a week or longer.  Copper shunts may have formed inside the cells that can lead to a partial or total electrical short. "

"A sleeping Li-ion does not reveal the voltage, and boosting must be done with awareness. "

So it could be that you have already some "sleeping cells" where the protection already triggered and so you can't measure a meaningful full output voltage. If you can measure at all the points (B-/V0,V1-V15 or B-/V1-,V2-V16 depending on the numbering scheme) valid voltages and the individual cell pair voltages are above the mentioned 1.5V/cell you could have chances to revive the pack.

For this you would need a charger for single li ion cells where you put the charging wires directly to the right points off the BMS board. ( I'd assume this should work without sideeffects, but i can't guarantee that for sure. Maybe @esaj, @hobby16, @Cranium or other forum members could give some more profound advice).

As stated at this batteryuniversity.com site it could be that a normal charger won't work and you'd need one with this "boosting" "wake-up" capability.

I think the "boosting" / "wake up" the article speaks off refers to protected cells (ie. the kind that have a "mini-BMS" -board inside them). If the voltage goes too low, the BMS switches off and needs to be "waken" with small current / high enough voltage to start working again:

Sorry, couldn't find a smaller picture with clear view right now. The cells used in the wheels aren't protected, ie. they don't have that PCB inside the cells.

Just last week, I took apart a laptop pack that had Panasonic cells in 3S3P-arrangament. Measuring the voltages of the parallel cells (1S3P), the values were something like 2.83V, 2.75V and 2.52V. I could charge them (starting with low current, 0.1A), but they had severely degraded in capacity (>20% capacity loss), but could be charged fairly fast without any heating (1.7A per cell) after bringing the voltages up as well as discharged with about 0.7A per cell (that's the best my cheap iMax B6 -clone could do ). So with a charger capable of charging single cells (or single paralleled cells) and adjustable charging current, it could be possibly to "nurture" the low-voltage cells back to life, but they will still likely drop their voltage much faster than the "healthy" cells, and if the Ninebot BMS indeed does no balancing, the problem could come back shortly...

 

[Chriull]

43 minutes ago, esaj said:

After staring at the BMS-picture long enough, I partially disagree with @Chriull 

I think it's not V7, but V9. V7 seems to be left under that black insulator on the side. So there's two cell pairs between V9 and V11, giving 7.6V / 2 = 3.8V per cell on average.

That seems right - V7 is the label of the solder joint below the tape...

Quote

 

This however does look like there's more severe voltage difference. If the 16.6V -point is indeed V9 and not V7, the difference is even greater...  That would mean there are 6 cell pairs between those points, and then it would be only  16.6V-8.2V / 6 cell pairs = 1.4V per cell pair. Either there's more than meets the eye here, or there are several dead cells there. Or some cells are bypassed by the BMS due to short or damaged components?

With the average of 1.4V at least one of the 6 cell pairs has to have a voltage <= 1.4V which is below the recommended limit from batteryuniversity.com:

" Do not boost lithium-based batteries back to life that have dwelled below 1.5V/cell for a week or longer. Copper shunts may have formed inside the cells that can lead to a partial or total electrical short. When recharging, such a cell might become unstable, causing excessive heat or show other anomalies. "

Quote

It would seem that these are some optional(?) resistors that were left out from the BMS. "Funnily" though, at a glance it would seem that those could be used for cell-measurement / balancing, like Chriull suggested? Maybe it's true that Ninebot BMS doesn't have balancing...

It should have internal balancing with the max balancing current limited by resistors. With the left out transistors and resistors this balancing current could be higher and so the balancing more effective.

Here is a post where most of the relevent posts from cranium regarding the ninebot BMS are gathered: http://forum.electricunicycle.org/topic/2247-decrease-charging-time-5a-high-current-charger-mod/?do=findComment&comment=39638.

It could be that at the points i made a small yellow ellipse around one could measure the different cell voltages (against the measuring spot @noisycarlosalready used with the black probe).

If so the resistors r1,r2,r3,r81,r7,r8,... should be the balancing current limiters which are imho around 1k leading to ~4mA max balancing current. Maybe @noisycarlosyou could confirm the values of these resistors?

 

Quote

 

I think the "boosting" / "wake up" the article speaks off refers to protected cells (ie. the kind that have a "mini-BMS" -board inside them). If the voltage goes too low, the BMS switches off and needs to be "waken" with small current / high enough voltage to start working again:

...

Sorry, couldn't find a smaller picture with clear view right now. The cells used in the wheels aren't protected, ie. they don't have that PCB inside the cells.

Was imho mentioned around here quite some times - totally forgot about that ;(

Quote

...So with a charger capable of charging single cells (or single paralleled cells) and adjustable charging current, it could be possibly to "nurture" the low-voltage cells back to life, but they will still likely drop their voltage much faster than the "healthy" cells, and if the Ninebot BMS indeed does no balancing, the problem could come back shortly...

 

With this and maybe some cells got below 1.4V imho the "salvage the healthy cell for flashlights/etc" project should be the best solution....

[noisycarlos]

Well. I finally disassembled the pack and took measurements. It turns out there were 8 bad cells. 6 gave 0.0 volts, 2 gave 1.0 volt or so. All the others were exactly at 4.15v.

So there it is. I'd still like to know why they died though.

PS. Since I'm starting a YouTube channel, I figure I might as well record and post a time-lapse of the process of measuring and dismantling the pack. I don't expect this one to go viral, or anywhere outside the EUC comunity lol, but if you wanna check it out, here it is.

 

[esaj]

9 minutes ago, noisycarlos said:

Well. I finally disassembled the pack and took measurements. It turns out there were 8 bad cells. 6 gave 0.0 volts, 2 gave 1.0 volt or so. All the others were exactly at 4.15v.

Wow, that's a lot of dead cells, I don't recall hearing more than one or two usually failing... At least you now got a big bunch of working cells for other projects, maybe it's time to start learning electronics?

 

Quote

So there it is. I'd still like to know why they died though.

Start here and make your best guesses  

http://batteryuniversity.com/learn/article/what_causes_lithium_ion_to_die

http://batteryuniversity.com/learn/article/capacity_loss

http://batteryuniversity.com/learn/article/premature_voltage_cut_off

http://batteryuniversity.com/learn/article/bu_803b_shorted_cells

http://batteryuniversity.com/learn/article/bu_808b_what_causes_li_ion_to_die

...and probably other relevant articles at Battery University and elsewhere.

[noisycarlos]

Yeah, I was thinking of getting a headlight, I saw the ones that use 18650s are much much brighter than the ones I have based on AA.

My friend also wants one or two cells for vaping. But that still leaves me with plenty of spares. So I'll figure something.

Thanks for the links :-)

[Hunka Hunka Burning Love]

I'm surprised that the NB1 still showed 100% charge after charging the pack up for a couple of hours.   With 8 dead cells you would think it wouldn't charge up that high unless there was one cell distributed among different groupings.   It looks to have 6 groups of 4 cells and 2 groups of 3 cells.  Thanks for the post mortem as it's interesting to know how the NB1 behaves as the battery ages.  Maybe get a Charge Doctor to do the 90% charging for you, and ever 20 cycles do a full charge + a couple of hours extra to help balance the cells to see if you can get more life out of it. 

On a side note - what the heck are you doing Jetpacking with a Google Glass???  Aren't those things like thousands of dollars?  Are they waterproof?  

https://www.amazon.ca/Google-Explorer-Frames-Rocker-Package/dp/B00K306QH8

[noisycarlos]

On 8/31/2016 at 5:00 PM, HunkaHunkaBurningLove said:

On a side note - what the heck are you doing Jetpacking with a Google Glass???  Aren't those things like thousands of dollars?  Are they waterproof?  

https://www.amazon.ca/Google-Explorer-Frames-Rocker-Package/dp/B00K306QH8

Hehe. Yeah. I was worried at the beginning. They're not officially waterproof, but i saw videos of people in pools and stuff. Plus, the friend who let me use her Google Glass was part of the explorer program, and she said they replaced them for her a few times when they broke. So I wasn't that worried about splashing them after that.

I did tie a big floater to the legs and hid it behind my head, so they wouldn't sink if they fell. 

In a couple of weeks I'll post a fun video I shot yesterday, I went unicycling in the storm drain tunnels :-)