Kingsong S20/S22 (Confirmed)

[supercurio]

31 minutes ago, RagingGrandpa said:

+1 

I'm very impressed that they performed a destructive test that replicated the fire, and published the result. Kudos, Kingsong.

I wish they would change their fuse rating from 60A to 30A... but if the BMS is working properly, it's fine as-is.

Did they really perform a destructive test which replicated the fire tho? Or it is their assumption that it would result in a fire, and they were not really up for destroying a perfectly good pack?
This picture highlights how their testing environment is entirely inadequate for a test which will result in the pack catching fire.
The video does not show the pack catching fire.

So maybe they replicated the fire another time away from a camera, and this re-enactment is only for picture taking purposes. The lack of transparency here is unfortunate.

When you made your own destructive battery test @RagingGrandpa everything was recorded and you highlighted what got destroyed in the video.
A shame they did not do that IMO.

Edited April 22, 2022 by supercurio

[Robse]

13 minutes ago, supercurio said:

Did they really perform a destructive test which replicated the fire tho? Or it is their assumption that it would result in a fire, and they were not really up for destroying a perfectly good pack?
This picture highlights how their testing environment is entirely inadequate for a test which will result in the pack catching fire.
The video does not show the pack catching fire.

So maybe they replicated the fire another time away from a camera, and this re-enactment is only for picture taking purposes. The lack of transparency here is unfortunate.

When you made your own destructive battery test @RagingGrandpa everything was recorded and you highlighted what got destroyed in the video.
A shame they did not do that IMO.

When i first saw this video, i was worried:  What have the planned to do, if the pack explodes and catch fire in the face of the technician ...   what surroundings are they in?  hmmnn

[chanman]

I think we can fairly safely assume that the red hot nickel strips would have lead to a proper battery fire if the test was continued. I suppose I did find it interesting that there was videos for the remaining tests but just photos for the failure replication.

[RagingGrandpa]

57 minutes ago, supercurio said:

Did they really perform a destructive test which replicated the fire?
The video does not show the pack catching fire.

I'm impressed they showed anything! 
Imagine what the youtubers would do if KS provided a video of a pack burning down?!  It would be sensationalized out-of-context for clickbait, as fast as people can click.

They showed flames. It's smart, and sufficient. 
(Hopefully they drowned it in a bucket as soon as the fireworks started.)

1 hour ago, supercurio said:

if there's a complete short, the nickel strips would take a while to get hot because of this thermal mass while the fuse has negligible thermal mass therefore blows instantly

Evidently, yes.

Luckily, we don't have to theorize here- it's happened twice now  

1 hour ago, supercurio said:

any sustained or pulsating current for that matter which gets the nickel strips red hot but doesn't blow the tiny fuse highlights that either the nickel strips are not thick enough and/or the fuse is rated for a too high current

Agreed.

This is surely a "fault current" - such high draw is not possible during riding. So, a lower-rating fuse would be the appealing choice.

But if the BMS rapid-retry issue is fixed, it shouldn't be possible to glow the strips, so the fuse won't matter except for during multi-point failures (catastrophic physical damage, etc).

1 hour ago, supercurio said:

What would you think of an automotive fuse in a holder
I think it's similar to what LiTech packs are using.

I don't like those, since their blade-style connection is not very reliable during high-vibration use. 
Better would be to use a bolted fuse holder like Leaperkim does. Takes up more space, though... if I owned an S22 I'd leave it alone.

1 hour ago, supercurio said:

what if you drain 70A from the pack continuously

70A from only 2 cells in parallel (e.g. one S22 pack) is damaging to the cells. We need to prevent it; and there is no need to upsize the connections to support it.

 

Edited April 22, 2022 by RagingGrandpa

[Cress]

@Tawpie and @RagingGrandpa - I agree, and you are correct, King Song found and fixed their firmware/fuse issue.  @roghaj, I understood, was saying that he trusts fuses.  If a firmware/circuit-breaker saves my soldered-in fuse I'm O.K. with that.    I want to use the advantages of  Li-ion battery chemistry without starting a fire. 

Participants in this forum do a good job of diagnosing EUC design issues, not different from pre-production engineering reviews.  The reviews work best if they are done before production.  If everybody in this forum reviewed a design we'd have a list of modifications and suggestions without 100% consensus, management would make choices and in any case after 10,000 production units something would fail or need changing.  The process of multiple people reviewing a design is not happening with EUCs.  EUCs have beta-testing, corrections post-production, and some points of failure stay hidden until they fail in the future.

There are several links in the chain of events that caused the S20 fire.  If the fire were in an airplane or automobile, etc. the manufacturer and the government would review each link in the chain.  King Song's response was good, better compared to Begode/Gotway, but Li-ion batteries powering hub motors come with lots of possible points of failure and without a history that we can refer to.

I thank King Song for their response, I hope EUC manufacturers trend toward that kind of response.  The chain-of-events for EUC issues has a lot of links that have to be sorted through and we'll all respond until we run out of links. 

[Tawpie]

3 hours ago, supercurio said:

What is your understanding regarding:

• nickel strips glowing red hot under pulsating current & fuse not blowing
• fuse blowing immediately under non-pulsating current & nothing happening with the nickel strips

Think about your toaster or electric heater. The wires in there that get red are mostly likely nickel chromium, a material chosen because it's relatively resistive, and is pretty good about staying hot once it gets hot. It's powered by AC mains, which means that 50 or 60 times a second power is applied and then removed (sine wave)... it doesn't receive continuous power from the mains. But the little doses of heat build up in the wire and it eventually gets hot enough to make toast. The thermal conductivity of NiCr is low. This is how we get red nickel strips without blowing the teeny tiny fuse, nickel just can't conduct heat very well so it tends to build up where it's generated. The redness is, BTW, welded directly to the battery terminals so it has a pretty good thermal path to the individual cells (the cells are going to cool the strips! This is not ideal, by any stretch of the imagination).

When KS disabled the BMS's MOSFET circuit breaker and shorted things together, there was a sustained over current situation through the fuse that continued until the metal in the fuse melted. The short circuit didn't exist long enough for the nickel strips to heat up—they don't cool fast so they also don't heat particularly quickly. Your toaster takes, maybe, 5 seconds before the wires are red?

Grab a glass automotive fuse and look at the element. It's kind of hour glass shaped... the two ends are wider and it's narrowed down to a thinner strip in the center of the fuse.   The thin section is the fuseable link, it's the part that will melt. Under AC (pulsed) conditions, during the 'off' time the heat generated in the thin section gets transferred to the wider sections on each end and from there to the metal caps and on to the fuse holder. Under continuous over current conditions, the heat generated in the thin center can't get to the edges quickly enough and builds up in the thin section where it will melt fairly quickly. If you have better cooling of the fuseable link or the fuseable link has higher thermal inertia (it has more mass), it's called a "slow blow" fuse.

What's really wonky about this bug is that if the BMS overcurrent shutoff was delayed a little bit longer, the PCB fuse would have blown and the nickel strips (probably) wouldn't get red hot. If they hadn't bothered to protect the board with the resettable circuit breaker MOSFET and just let the fuse blow, the batteries might not have gone into thermal runaway. (I say probably and might not have because we only presume that there wasn't an unrelated problem that caused a cell to go thermal, we don't know and won't ever know)

It's ironic that a 'feature' (non-destructive, fast acting, resettable, and programmable over current protection) plus a software mistake (didn't latch off when tripped) made a 'smart' BMS into a dangerous smart BMS.

3 hours ago, supercurio said:

Or is the idea that if there's a complete short, the nickel strips would take a while to get hot because of this thermal mass while the fuse has negligible thermal mass therefore blows instantly?

This is more correct

Edited April 22, 2022 by Tawpie

[Tawpie]

3 hours ago, supercurio said:

What would you think of an automotive fuse in a holder like that?

Did you see the LiTech guys jump when the in-line fuse blew during their 80A load test video? It'll work, but it's still fun. And probably the fuse holder needs replacing when the fuse does blow.

[Tawpie]

3 hours ago, supercurio said:

In short: what if you drain 70A from the pack continuously:
Wouldn't the strip get red hot already, while the fuse slowly gets to the temperature where it'll blow?

A 60A fuse handling 70A should go pretty quickly but might take several seconds. The strip will get hot for sure, but I don't know how hot. At a continuous 60A draw, the 60A fuse might take a very long time to blow (ref LiTech 80A load test). I have no idea how hot the nickel strips would get but if they're anything like the LiTech packs it doesn't seem like they get that hot.

Edited April 22, 2022 by Tawpie

[supercurio]

2 hours ago, RagingGrandpa said:

I'm impressed they showed anything! 
Imagine what the youtubers would do if KS provided a video of a pack burning down?!  It would be sensationalized out-of-context for clickbait, as fast as people can click.

Fair enough, it's still better than nothing.
I guess I feel like there's something missing to be transparent and thorough as they showed us a solution but not how they reproduced the issue in the first place.

2 hours ago, RagingGrandpa said:

They showed flames. It's smart, and sufficient. 
(Hopefully they drowned it in a bucket as soon as the fireworks started.)

3 hours ago, RagingGrandpa said:

Ah interesting you saw flames here, I only saw a spark, the red hot nickel strips but wondered about these vertical orange glowing areas. I can't tell. What do you guys think?

@roghaj in the podcast mentioned they could be reflections due to the translucent plastic layer on top of the BMS.
The description in report says, in the section titled Output short circuit test:

After several output short-circuit tests, the battery output short-circuit fails, and the nickel sheet burns red directly. If the short-circuit continues, the battery will catch fire

From what they shared my understanding was not that they were showing us the fire, nor that it was reproduced here (possibly due to the unsuitable environment). Only that they showed us the red hot nickel strips and stopped right here.

2 hours ago, RagingGrandpa said:

(on fuse vs nickel strip)
Evidently, yes.
Luckily, we don't have to theorize here- it's happened twice now  

Yes right, possibly however can we rule out with absolute certainty that it was red hot nickel strips which ignited the fire and not something else.

It happened once in Kingsong's testing, but the exact failure mode during the NYC fire is unproven IMHO until reproduced identically.

And that's still today the main gripe I have with all this. They showed us they fixed a bug they found. They didn't demonstrate they fixed whatever happened during the NYC fire. Instead they merely speculate on what happened (guessing a hall sensor failure) and mention no attempt to reproduce it.

I don't know, maybe I'm too pedantic here; or maybe it is indeed an obvious requirement in methodology.

2 hours ago, RagingGrandpa said:

This is surely a "fault current" - such high draw is not possible during riding. So, a lower-rating fuse would be the appealing choice.

But if the BMS rapid-retry issue is fixed, it shouldn't be possible to glow the strips, so the fuse won't matter except for during multi-point failures (catastrophic physical damage, etc).

I don't know about you guys but by now I lost trust in KingSong's software, remembering things like

• the buggy short protection in question
• it took them years to fix the vampire drain on 16X & S18 at the end of charge
• 18XL firmware update free-spin safety regression
• dysfunctional tiltback which led to overlean crashes on the S22 
• a broken battery % estimate which left almost all riders stranded in their range test, which was changed and made even worse in the last firmware. how hard can it be to estimate battery % from battery voltage.

It's just a crapshoot IMO. Aside from the main self-balancing loop which is clearly observable, I'm incline to consider all of their safety and algorithm which are not directly out here broken unless proven otherwise, and would not rely on any of them working when needed.
Lack or absence of QC  looks that bad.

2 hours ago, RagingGrandpa said:

I don't like those, since their blade-style connection is not very reliable during high-vibration use. 
Better would be to use a bolted fuse holder like Leaperkim does. Takes up more space, though... if I owned an S22 I'd leave it alone.

Ah yes good point, it would be counterproductive to introduce a point of failure with the fuse holder themselves. Something to look in depth into indeed.
I'm not sure there will be a good opportunity to attach a bolted fuse holder in that case, which is why I was thinking about this kind of automotive fuse holder, hoping the form factor could also help dampen the harshest shocks and vibrations as they're in made of hard rubber.
What would you think about that type held in place with some kind of roam?

I recall these blade connections are used on the 16X motherboard. But clearly they moved away from that.

2 hours ago, RagingGrandpa said:

70A from only 2 cells in parallel (e.g. one S22 pack) is damaging to the cells. We need to prevent it; and there is no need to upsize the connections to support it.

Yes I took 70A as something that would have a good chance to get the nickel strips red hot, stress or damage the cells but not blow the fuse immediately.

Edited April 22, 2022 by supercurio

[Tawpie]

6 minutes ago, supercurio said:

From what they shared my understanding was not that they were showing us the fire, nor that it was reproduced here (possibly due to the unsuitable environment). Only that they showed us the red hot nickel strips and stopped right here.

Correct. They showed that they could create a situation that has a high likelihood of leading to a fire, but did not show an actual fire.

[Tawpie]

19 minutes ago, supercurio said:

I don't know about you guys but by now I lost trust in KingSong's software, remembering things like

there is good and valid reason for concern... although I have a hunch that tiltback is not enabled in the demo firmware (you know, speed speed speed)

Software is really really difficult. Once you get it mostly working, you usually ship. BG has their unfixed strangeness. The V12 does all sorts of odd things. Shermans have negative mileage. We've been pretty tolerant of "user interface" bugs because we don't want to wait...

No excuses though. Safety issues have to get rigorous scrutiny. BG has an advantage, they've kept things brutally simple: lean=go. Maybe this will hurt the board or battery, maybe it won't, but damn, does it go! We don't need no stinking redline governer.

Edited April 22, 2022 by Tawpie

[supercurio]

Thanks I have to say that although it is hard to be satisfied with KS moves, I'm really appreciative of the discussion here and everything learned in the process

[Twerd68]

Any op ions from you all on whether the Samsung 40T cells should also help? Seems unrelated to this thread but maybe it provides more safety margin? Ewheels seems to think it may help. Thank you

[fryman]

I would ask the same question.  Only benefit is 35A discharge. Negatives only 4000mah and 250 life cycle. 50e has 5000mah and 500 life cycle. 

[chanman]

25 minutes ago, fryman said:

Negatives only 4000mah and 250 life cycle

The cycle life difference is fake, we discussed this in the master thread. The 250 cycles on the 40T is with 6A charge and 35A discharge, while the 500 on the high capacity cells cells is something like 2A charge and 5A discharge. Under real world conditions for wheels we can expect much better performance from the 40T cells.

There was a nice graph posted of test data on more real world conditions https://endless-sphere.com/forums/viewtopic.php?f=14&t=114473 last one on the page. (Although this one is a bit unfair to the high capacity cells because it's discharging them above their specs, it does show 40T performing well at 20A discharge and 2A charge, having something like 85% capacity at 1000 cycles)

Edited April 25, 2022 by chanman

[mrelwood]

4 hours ago, fryman said:

I would ask the same question.  Only benefit is 35A discharge. Negatives only 4000mah and 250 life cycle. 50e has 5000mah and 500 life cycle. 

4000mAh * 3.7V * 120 = 1776Wh. Assumed range of roughly 70km per cycle times 250 cycles equals roughly 17’500km (10’938 miles) for the capacity to fall to 80%. I personally wouldn’t be troubled by the “low” cycle life at all.

[Miko.cz]

Delivery update from my distributor (Ekolka) - source web: FB
Shortened translation from 23rd of April: Upgrade 20->22 has been done (issues fixed). EUC manufacturing already running. Expected 10th of May departure from KingSong, expectations 45-55days to Czech.

My note: Not sure what has been changed and/or fixed. Not sure if Czech post service will not ruin the delivery times (long tradition). I will share with you in case of any news.

[Forwardnbak]

yes my dealer also confirmed S22 in production now and will ship in next couple of weeks. 

[Tasku]

4 hours ago, mrelwood said:

4000mAh * 3.7V * 120 = 1776Wh. Assumed range of roughly 70km per cycle times 250 cycles equals roughly 17’500km (10’938 miles) for the capacity to fall to 80%. I personally wouldn’t be troubled by the “low” cycle life at all.

That would only be like 1 year of ride experience for some of us(me). I had 15000km after a year. Based on my devices battery health it looks I can do another year and another 15000km, totalling closer 30k before noticing performance loss.

I do like safety but not certain, if the dendrite formations and high use case is safe thing. Any performance loss means device battery life is getting worse and end of life scenario for batterys  draw closer. For safety, batterys change is key or stop using the device before performance goes too bad.

Edited April 25, 2022 by Tasku
typos

[supercurio]

@EcoDrift posted on Telegram this: (automatic translation)

Quote

We, KingSong Intell Co., LTD hereby declare: We are starting shipment of the KingSong S22 unicycle.
We apologize to all interested parties for the delay in shipments, which was associated with the improvement of the design features of the unicycle.

The list of changes is attached.

The front bumper is made of a more durable material.
Improved controller waterproofing.
Changed the setup and location of the pedals.
Additionally processed wiring.
Improved fastening of motor-wheel covers for greater reliability and ease of maintenance.
The angle of illumination of the headlight has been changed and its fastening has been improved.
Rear handle mount changed.
Improved battery cover for strength and easier maintenance of adjacent hardpoints.
Added suspension fasteners for greater structural reliability.

They added this comment:

Quote

🥳 Today KingSong plans to ship the first 50pcs of S22 (nee S20) to our delivery company.
And, now guess, was this list of improvements worth returning to the manufacturer's warehouse?🤷🏽‍♂️🤪

KS clearly omitted to say what prompted them to recall the shipment, which is a regrettable lack of transparency.

[chanman]

I don't think they actually recalled the shipment? I read "was this list of improvements worth returning to the manufacturer's warehouse?" as kind of a joke "we hope the improvements were worth the delay" some ambiguous language from the translation there but the emojis tell all.

[wstuart]

From KS's list of improvements the one I wonder about is "additionally processed wiring"....... wonder what that is?

 

I'm still on the fence about maintaining my S22 pre order, and I'm a tired of reading posts about glowing nickle strips

This is all I would need to see to make me feel instantly better about my KS pre order:

Take 5 of your newest, shiniest production S22s with all the software/hardware upgrades and send them to that Ustride guy who blew up the first one.  Have him do his absolute worst to them.  I don't even care if he over torques and smokes the main board on all 5, as long as they don't explode I'm happy.  I'll never be able to ride like that guy so I know I won't fry that board, I just want to see that the battery won't explode in a real world STRESS test. 

10 hours ago, supercurio said:

@EcoDrift posted on Telegram this: (automatic translation)

They added this comment:

KS clearly omitted to say what prompted them to recall the shipment, which is a regrettable lack of transparency.

 

[Elliott Reitz]

wstuart:  "I'm still on the fence about maintaining my S22 pre order, and I'm a tired of reading posts about glowing nickle strips"

Yea, metoo. 

As someone who already has an S22 I'm wondering if I should get an old metal garbage can to keep it in until I can replace its dead controller and questionable batteries.  I like the high surge batteries option talked about too.  I don't push range and do have speed anxiety vs underpowered wheels into high headwinds like the S18 (and like S22 compared to my 3YO MSX100V).  

Meanwhile, I just put an ewheels deposit in for a Begode Master.  When that arrives I think I might sell what's left of my S22 (obviously there might be a market for S22 parts to keep the 38 remaining prototypes alive).  

And yea, of the 40 prototypes 1 = destroyed by fire and 1 = killed by firmware declared fault (mine). And yea, killed by firmware fault since I can measure my batteries are  both charged over 124V.  (both are >125 now charging that its only charging 1 at a time).

Edited April 25, 2022 by Elliott Reitz

[Tawpie]

28 minutes ago, Elliott Reitz said:

both are >125 now charging that its only charging 1 at a time

I'm a doofus, confusing Master's voltage with S20's. My apologies.

Edited April 25, 2022 by Tawpie

[dycus]

Why is 125V too high? For a 30S pack that's only 4.167V per cell. It's fine up to 126V assuming all the cells are balanced.