Fire History

[alcatraz]

Yeah the heat is only one concern. If the cells were separated by more than just their heatshrink, then even with the heatshrink softening they wouldn't shortcircuit. They would be more resiliant to vibrations which in conjunction with tightly squeezed together cells could cause shortcircuits. (in shipping for examples)

Seeing as gotway wheels with the old panasonic ncr18650ga cells rarely caught fire, and they were built similarly, then it leaves me to suspect either bms or cells, or both.

The packs have become heavier by ca 50% (21700) so the cells on the outside of the pack are experiencing stronger impacts. 

If the packs received impacts before installing them, it could also explain some shipping fires. If we inspect the packs for dents etc, and add padding, it should reduce the risk of spontaneous fires.

Edited December 30, 2021 by alcatraz

[supercurio]

11 minutes ago, alcatraz said:

Yeah sustaining 60A is unusual. But it's under ideal conditions. Say that some cells are hotter than others and/or the rider is 120kg etc and/or joy riding in hot summers etc. All it takes is one cell to soften its heatshrink and a short can occur.

By the way did you mean 60A sustained per cell?
60A sustained for all packs combined is possible, but per cell is not. You'd have at 1/4 of that.

60A per cell would require high current cells to be shorted with a very low resistance. Example:
 

 

This cannot happen while riding. Any mainboard will burn up way before that.

11 minutes ago, alcatraz said:

All I have data on is that the M50T are the lowest peak current cells of all major cells today, and it just so happens that those packs are mostly affected. Even the Tesla 21700 cells appear to be better at peak currents. They stopped being used because of other issues.

Suggested read, @RagingGrandpa's tests.

I agree there's an issue with these, or more specifically the cells Begode has been using in their pack. For unknown reasons some of the cells are not performing as spec and it leads to diverse adverse results.

I think you got your figures mixed up tho on current per cell and overheating: the limits are elsewhere.

 

[alcatraz]

M50T can do 14.6A sustained for 5min before they overheat. That's their peak current. To deplete them completely without overheating they can do 6.x A. That's their continuous current rating. These are independent tests.

1800Wh wheels are 4P which means total (5min) peak current for both packs is ~60A. (Of course you can exceed this for short bursts.) Compare this to 4 year old Tesla packs that can do 52A for 5min. Compare the peak current draws that those two wheels can do and you're likely going to find the RS/Nikola/MSP are going to top the chart in "current draw vs rated peak pack current". I'm just pointing out that the "hot" wheels in such a chart correspond to some of the fires. It could be coincidental but it could be correlated. It's up to either one to decide.

By "overheat" I mean that the cells reach a certain temperature. The numbers are comparable, but there is no reason to focus on the "5min". All cells were tested the same way and the M50T have lower currents under the same conditions. That's what's useful.

https://www.batemo.de/products/batemo-cell-library/inr21700-m50t/

This german testing page is very interesting. Had I seen it earlier I would have written here earlier.

There's testing data for 48X, 50E, Tesla 21700 and M50T, and a lot of other cells but I take it these four are the most relevant.

Edited December 30, 2021 by alcatraz

[supercurio]

@alcatraz it would be a common-sense precaution to have at least 2 temp meters in each pack with the cells, (one where it would be the warmest, one where it would be the coldest)
And probably a 3rd strategically placed on the BMS.

In order to provide:

• over-heating protection for the cells and BMS
• under-temperature protection for charging

And the board would require the rider to slow down/stop and avoid charge unless safe conditions are matched.
I mean, nothing special here.

[alcatraz]

57 minutes ago, mike_bike_kite said:

I understand Gotway/Begode Amp figures are reported differently than other makes (they show 3 phase figures which are higher apparently) so it can be difficult to make comparisons. I seem to remember that Marty said he had his current alarm set to 90Amps. It's also odd that none of Marty's wheels have caught fire (AFAIK) when he runs them up mountains in the desert. It also doesn't explain why wheels are catching fire in shipping containers or while stored at night. There might also be more than one reason for the issues.

Yeah, the numbers shouldn't be used as absolutes. Simply to compare apples/apples.

Like this:

M50T 1800Wh packs have 17% less peak current than 3y old (msx84) 1600Wh packs.

An RS displaying 130A peak amp draw pushes its packs 30% more than an MSX displaying 100A. (just as an example, I don't know the figures)

[supercurio]

23 minutes ago, alcatraz said:

Yeah, the numbers shouldn't be used as absolutes. Simply to compare apples/apples.

Like this:

M50T 1800Wh packs have 17% less peak current than 3y old (msx84) 1600Wh packs.

An RS displaying 130A peak amp draw pushes its packs 30% more than an MSX displaying 100A. (just as an example, I don't know the figures)

Let's not compare apple and oranges by mixing up phase current with battery current in the first place.
Mainboards act as step-down converter to convert battery voltage into motor voltage, the later varies with motor rotation to push current against back-EMF voltage generated with the rotation of the motor.

So what the mainboard does typically is converting high battery voltage, low current into low voltage high current.
And the faster the motor turns, the closer motor voltage & current get to the battery one.

Then, peak is not a reliable measure for two reasons:

• infrequent data sample rate over Bluetooth, leading to miss most of the actual "peaks" the controller experiences
• during peaks energy doesn't only come from the battery but also the capacitors

One last factor is that it might compare different motors, like a MSX 84V C30 with an RS C38, which current vs speed can't be compared either.

[alcatraz]

Is there any way of comparing current draw between different Begode wheel models?

[supercurio]

12 minutes ago, alcatraz said:

Is there any way of comparing current draw between different Begode wheel models?

@Seba has implemented in EUC World a method to estimate battery current using phase current, battery voltage, speed and free-spin speed as inputs.

He described the method for initial calibration here: https://euc.world/blog/battery-current-vs-motor-phase-current/

You can expect inaccuracies adding up from measurements, variation between models, variation between each unit, variation between firmware versions etc, but it still give a reasonable starting point for Begode and Shermans.
And it's useful for setting current alarms on your own wheel, especially when matching them with beeps.
For instance after having my 1st Sherman board fail and noticing that the 55A estimated continuous was close to the beeps at higher speeds, I set a software alarm as such, which is also in the ballpark of 2x30A fuses.
It served me well so far, avoiding cutout-out or burning a second board despite an aggressive riding style and love for off-road.

TLDR, not calibrated for all wheels but still usable.

[RagingGrandpa]

3 hours ago, alcatraz said:

All it takes is one cell to soften its heatshrink and a short can occur.

Why would that be?

(Gotway packs since 2018 use cell frames...)

[Planemo]

6 hours ago, alcatraz said:

If the cells were separated by more than just their heatshrink, then even with the heatshrink softening they wouldn't shortcircuit.

Personally I would stop focusing on the heatshrink. If heatshrink is getting hot enough to melt and cause shorts (which in itself is difficult given the cell separators and the fact that all the likely contact points would be negative only) you have far bigger problems within the cells themselves.

[Planemo]

6 hours ago, supercurio said:

I agree there's an issue with these, or more specifically the cells Begode has been using in their pack. For unknown reasons some of the cells are not performing as spec and it leads to diverse adverse results.

This.

Its got nothing to do with heatshrink, cell separators or even the design of BMS. We know for a fact that all this was the same and fine, used for years prior to the 900wh packs being a problem.

It was interesting to see someone else pick up on the fact that the interconnects appeared to be sized for 18650 but were used on 21700's. I would like to see this bottomed out, because thats a shorting possibility right there.

[RagingGrandpa]

47 minutes ago, Planemo said:

the interconnects appeared to be sized for 18650 but were used on 21700's.
thats a shorting possibility right there.

Why does an offset matter, given that the cell rim is guarded by a thick chunk of plastic?

On 12/29/2021 at 12:34 PM, RagingGrandpa said:

I think it's prevented by the plastic frames, which cover the rim of the cell top at each positive end.

 

[goatman]

1 hour ago, Planemo said:

Personally I would stop focusing on the heatshrink. If heatshrink is getting hot enough to melt and cause shorts (which in itself is difficult given the cell separators and the fact that all the likely contact points would be negative only) you have far bigger problems within the cells themselves.

i agree about the cell wrappers not being an issue from heat but once you connect cells in series the can is no longer negative,

 

[alcatraz]

9 hours ago, RagingGrandpa said:

Why would that be?

(Gotway packs since 2018 use cell frames...)

You're right. The short is being caused by something else. What's the chance of these frames cracking? Hmm.

This is unrelated to your point but I just looked at some pictures and what struck me as a hot spot are the tiny nickel strips between the cells and the bms. For each 5A of current there needs to be enough crossectional area of nickel to keep temperatures in check. On my DIY V8 battery I put about 5 times as much as I can make out from the pictures. I'm surprised they don't burn holes in the heatshrink, or melt the solder. Are there perhaps nickel strips going between the cells, and not everything is routed through the board? If that's the case then shouldn't the output wires go to the cells? I'm pretty sure they go to the board. 

I'd love to see how a 48X updated pack looks compared to an M50T one. Here's an M50T pack:

 

Edited December 31, 2021 by alcatraz

[alcatraz]

12 hours ago, supercurio said:

@alcatraz it would be a common-sense precaution to have at least 2 temp meters in each pack with the cells, (one where it would be the warmest, one where it would be the coldest)
And probably a 3rd strategically placed on the BMS.

In order to provide:

• over-heating protection for the cells and BMS
• under-temperature protection for charging

And the board would require the rider to slow down/stop and avoid charge unless safe conditions are matched.
I mean, nothing special here.

I like the idea of attaching a device to the packs. Like a logger. Temp gauge, balance wires, maybe a current sensing loop around the output wire. Then go for a hard ride, heavy braking, fast charge, whatever. If everything checks out then it's all good. If such a device is expensive then just get one and move it around, from wheel to wheel.

If we had the ability to monitor what's going on we would have solved this issue a loooong time ago. 

If the device proposition seems far fetched then lets just hope for smart bms systems. Like this (bluetooth).

Edited December 31, 2021 by alcatraz

[BIGboiFASTboi]

I didn't see this KingSong 14D fire on the list... So here it is.


Couldn't post link here, but here's the original URL.

youtube com/shorts/EN3ZRIthsdQ?feature=share

I didn't see this KingSong 14D fire on the list... So here it is.


Couldn't post link here, but here's the original URL.

youtube com/shorts/EN3ZRIthsdQ?feature=share

[RagingGrandpa]

40 minutes ago, alcatraz said:

Are there perhaps nickel strips going between the cells, and not everything is routed through the board?

Right. The high-current path does not use the board. The pads on the PCB are just a convenient location to solder cell strips to the output wires. The cell strips are folded onto the board and onto eachother at B-, Cell 12 (your pic shows this well), and B+. And as my recent test showed, the cell strips melt during a short-circuit condition, leaving the board unscathed.

  

40 minutes ago, alcatraz said:

I'd love to see how a 48X updated pack looks compared to an M50T one.

They're both 21700-size cylinder cells... aside from BMS PCB revision, I presume the construction is the same (but would welcome a teardown).

 

Edited December 31, 2021 by RagingGrandpa

[Paul A]

https://www.allaboutcircuits.com/news/just-how-much-breakthrough-teslas-tabless-battery-cell/

 

Tab manufacturing is challenging and can affect battery reliability and performance. For example, welding burrs can penetrate the separator layer between the positive and negative electrodes and cause an internal short circuit.

The CT scan of a battery with a welding burr is shown below. 

Such manufacturing defects can lead to battery fires and explosions.

___________

*Couldn't upload image of CT scan.  The link to the article has the picture.

Edited December 31, 2021 by RagingGrandpa
(image URL bugfix)

[Caveira FPV]

i'm scared, my msp caught fire in the morning without being connected to the power.  5 months of use

 

https://youtu.be/_QW108ospc8

 

https://youtu.be/AtwbeD7VE1k

[Caveira FPV]

On 23/12/2021 at 13:10, RagingGrandpa said:

Perdeu mais um MSuper 100V no Brasil ... duvido que conseguiremos muito mais informações por meio do boato internacional, mas talvez.

i am the owner, i am new here.  I am at your disposal for any clarification

[Tawpie]

That's very very scary. I don't have anything I can tell you, but am glad there don't appear to have been any people injured. At the very least, you should contact the reseller of your wheel and report this to them—I doubt they'll be able to do anything, but they should be made aware. Yours is not the first wheel to catch fire, and sadly won't be the last. You were very lucky.

Edited January 3, 2022 by Tawpie

[Paul A]

The owner states that the MSP Super was bought new, about 5 months old, about 2k km, never wet, wasn't being charged at the time, still under warranty, and the staff of the dealer are being very helpful.

He also had a KS16x stored in the building that was also destroyed by the fire.

He states that if he had of been taking a bath in the building at the time, he would have been trapped and killed as there was no way to go through the hallway.

He states he is not going to purchase a Gotway again.

He states that the scare was very traumatizing.

 

[RagingGrandpa]

9 hours ago, Caveira FPV said:

my msp caught fire in the morning without being connected to the power.  5 months of use
i am the owner, i am new here.  I am at your disposal for any clarification

Bom dia

I want to know: When was the wheel manufactured?

You can tell it from the label on the motor, or the label on the circuit board inside.
Maybe you have a picture of those labels, from before the fire?

Thx

[Caveira FPV]

4 minutes ago, RagingGrandpa said:

Bom Dia

Eu quero saber: quando a roda foi fabricada?

Você pode perceber isso pela etiqueta no motor ou pela etiqueta na placa de circuito interna.
Talvez você tenha uma foto dessas etiquetas, de antes do incêndio?

THX

unfortunately I don't have this information, as the wheel was just purchased at the store and was never opened.  it was still under warranty.

[mike_bike_kite]

@Caveira FPVJust curious but did you have power pads fitted? Did you ever have any major impacts on the wheel? Did you use it for jumps? How often did you leave the charger battery plugged in at 100% to balance the batteries?