Kingsong s20 Cut out FIRE (New York)

[supercurio]

The S20 was fully charged initially.

U-Stride was conducting this max acceleration/max breaking stress test on a nearly full charge, the cut-out happened 1:15 after the beginning of the ride.
Enough to drain the battery a little but not that much.

It's frankly a bad idea to repeat hard breaking test on a nearly full battery and with any wheel.
The voltage from regen can shoot way beyond the 126V spec in that case.

It changes my perspective a bit, to me there's a component of user error here.

However everything else:

• lack of beeps to alert of over-voltage condition
• escalation to any kind of fire has no reason to happen

Edited March 28, 2022 by supercurio

[RagingGrandpa]

22 hours ago, Tawpie said:

No flaming batteries flying around is also a good thing, the rubber seals will fail to relieve the pressure build up in the battery case (and make a nice blowtorch) but it looks like the case was able to contain all of the cell explosions.

17 hours ago, Paul A said:

Appears the aluminum battery cases worked.

There didn't appear to be any ejecting cells.

Umm... flaming shit went everywhere  

If your couch was 10ft away from your EUC, your couch (and therefore house) is going up. Sorry. (So, let's prepare.) 
 

12 hours ago, Tasku said:

E-bike fires are a thing, but they do not shoot like flamethrowers to 2.5meter away..

Oh, they sure do!

(Nearly any instance of Li-Cobalt cells doing thermal runaway in a sturdy box, will result in a big jet of flame...)
 

10 hours ago, Paul A said:

Could a burning control board generate and transfer enough heat, in twenty seconds, to cause thermal runaway in the battery packs?  

The cells are encased in the aluminium boxes.

Almost certainly not. It's why we think there is an electrical cause here, not a conducted-heat cause.  
 

11 hours ago, Cress said:

can confirm the battery manufacturer / model ?

It's this... but nearly all Li-Ion cells of this chemistry will burn if you overheat or short-circuit them  
 

3 hours ago, supercurio said:

The controller was not ravaged by flames to begin with, which likely rules out fire of the controller propagating to the battery packs as root cause

Umm... are we looking at the same picture?  

The PCB is destroyed.
Perhaps the only normalcy here is: "it's still a fiberglass rectangle with some copper here and there."
Controllers can absolutely pass massive, damaging currents, without physically disintegrating. 
 

11 hours ago, Roadpower said:

I am forming a pending order for two fire blankets. So far I'm leaning towards a two pack of 49 inch blankets.

Some forum members have experimented with fiberglass blankets for dealing with battery fires... with mixed success.

Check this out:
https://forum.electricunicycle.org/search/?&quick=1&tags=fire safety

On 6/24/2021 at 10:44 AM, RagingGrandpa said:

Nice demonstration!

 

[techyiam]

21 hours ago, Seba said:

...and this is why I recently bought an 18XL to be a successor of my 18XL. And in light of all recent incidents and my own experiences with S20, I'm happy that once again I choose a well known, matured, tested and absolutely reliable wheel.

Your choice is very intriguing to me. You already have a V11, yet you recently bought a 18XL to replace an existing 18XL.

Looking at your profile, you obviously like Kingsong. And if I am not mistaken, you have said in your posts that you like the quality of Kingsong controller boards. So I get that. Since your most powerful wheels are the 18XL, and the V11, speeds above 45-ish kph is probably not important to you.

I am wondering whether you bought the 18XL the 2nd time is because there is something still missing from the V11 that you still like on the 18XL? Is it just because of reliability, or is there more to it?

I have never ridden a 18XL before, so I don't know how it rides. There must be something really special about it.

As for me, I started on a T3, which I still like very much. I got a 2nd wheel primarily because the bike routes that I ride include plenty of bumpy side streets that greatly reduce my speeds when I am on the T3. So hence, logically I should have bought a suspension wheel. As it turns out, I didn't. I was close to buying the V11, but at the last minute, I decided I wanted more headroom, and to postpone buying suspension wheel until the 2nd generation. I bought the V12 instead (although the 18XL was always on my mind). The V12 has a more refined and premium feel to it then the T3. Using car analogy to contrast the two, I would say, to me anyway, the V12 feels like I am driving an Audi S4, whereas, the T3, a Honda Accord. When the tuning is set right, the V12 is a lot more compliant over bumps than the T3. If it wasn't for the mosfet debacle, the V12 works well for me.

Going back to your purchasing of the 18XL the 2nd time, is the 18XL good enough on bumpy roads that you would prefer it over the V11 for your commute? And you use your V11 primarily for offroad? Or is reliability also a factor in the equation? Just curious? Thanks.

Edited March 29, 2022 by techyiam

[techyiam]

1 hour ago, supercurio said:

The S20 was fully charged initially.

U-Stride was conducting this max acceleration/max breaking stress test on a nearly full charge, the cut-out happened 1:15 after the beginning of the ride.
Enough to drain the battery a little but not that much.

It's frankly a bad idea to repeat hard braking test on a nearly full battery and with any wheel.
The voltage from regen can shoot way beyond the 126V spec in that case.

It changes my perspective a bit, to me there's a component of user error here.

It looked like his max accel / max braking stress cycle was becoming more frequent just before the cutout too. It looked to me like he was doing a pendulum type stress test, but at speed as oppose to stationary. Perhaps this type of user-error induced stress test can become his stress test trademark, not unlike Overheat Hill stress test for Marty, except Marty does pause when the euc gets close to overheating.

Nasty outcome, but no one got hurt, and huge flaws exposed, after all. Mind you, don't know if he can repeat such tests again without getting hurt in the future. 

Edited March 28, 2022 by techyiam

[Cress]

We should thank EUC Distributers for negotiating with EUC manufacturers.  Well-informed distributors are positioned to influence EUC manufacturers to make better choices.  Reference @RagingGrandpa the EUC manufacturers are choosing high watt/hr-capacity cells without protecting charge/discharge rates.  It's a follow-the-money issue where the Manufacturer is biased to protect his profit from the batteries and profit from (miss-leading) EUC performance specifications.

This note is from Jason @eWheels:

eWheels Support <support@ewheels.com>
To:richard cress
Cc:Ewheels Support
Mon, Mar 28 at 11:37 AM
Hello Richard,

Yes, it's a total fiasco, 100% King Song's fault. For the past year, I've been trying to have them have the whole system evaluated by a 3rd party, particularly the battery packs; it's the typical situation where these manufacturers think that they have all the answers, fail to understand risk complexity, and here we are...  

The outcome of this disaster is that it now gives us a better negotiating hand for using high-powered cells, outsource pack production, require our own battery stress testing plan, battery pack certification & more extensive reviewer testing.

You can imagine that we're rather inundated today.

Thanks Richard,

Jason 

[chanman]

I don't think the high power cells is something we absolutely need to be advocating for, Realistically the current limitations of the high capacity cells are quite high compared to regular demands on the wheel. (for S20 4P 10A per cell gives 40A sustained output, which is already loads of power). For most people switching to the high power cells will be sacrificing capacity without any real benefits. I don't think high power cells necessarily would have had a better outcome in this situation, even though the cells may support higher discharge they will still go thermal if shorted.

Demanding for battery packs with safety features that actually work verified independently with actual tests, 1000%.

[techyiam]

13 minutes ago, Cress said:

eWheels Support <support@ewheels.com>
To:richard cress
Cc:Ewheels Support
Mon, Mar 28 at 11:37 AM
Hello Richard,

Yes, it's a total fiasco, 100% King Song's fault. For the past year, I've been trying to have them have the whole system evaluated by a 3rd party, particularly the battery packs; it's the typical situation where these manufacturers think that they have all the answers, fail to understand risk complexity, and here we are...  

The outcome of this disaster is that it now gives us a better negotiating hand for using high-powered cells, outsource pack production, require our own battery stress testing plan, battery pack certification & more extensive reviewer testing.

Right there tells me not only not to buy first batches, but also not to buy any batch until all important flaws are uncovered, and are properly fixed.

[Tawpie]

Well, I guess we'll hear what we hear on April Fools day, right?

A thought on KS figuring it's a SW problem (while still saying they're looking for root cause), there are a couple reasons that might explain why KS is hoping this is true:

1. it's less costly to roll out a change (dumb reason, but when you're hoping you might as well hope for lower cost)
2. "Hardware will break, software comes broken" (unknown, but a mantra of mine)

I encourage us to continue to formulate our personal theories on what happened, but to keep in mind that we only have video to go from. Video is very telling, but it mostly tells an investigator where to start looking for the actual root cause. What we see are the results of something going wrong—but video alone doesn't give us a clear window into why it broke in the first place. Unfortunately, to gain assurance that the problem is actually addressed and not papered over, we don't have much choice but to trust the intent of the manufacturer. That trust must be tempered by acknowledging that "business concerns" are real and exist for all manufacturers, including KS, and these are often in conflict with the good of us riders. Hopefully KS is the kind of company that will be able to find a solid balance between our safety and their business needs. In the past, they've erred on the side of rider safety, I really hope this attitude hasn't changed.

I do want to shout out to eWheels, my distributor of choice. They've demonstrated that they are a no nonsense advocate for rider safety when it comes to system reliability especially when it comes to fires. My single order doesn't get KS's attention, but eWheel's container loads do. I want to encourage everyone to buy through reputable distribution—they have very large levers and if they use their influence as eWhees does, there is real hope for change.

[fryman]

Can we get another S20 tester to test their wheel with the same riding style?

[RagingGrandpa]

14 minutes ago, fryman said:

Can we get another S20 tester to test their wheel with the same riding style?

Please!

And if it's the same configuration as the demo wheel that burned, retrofit additional, lower-rating fuses in the battery wires. 
A controller failed. And then the pack burned. Let's study them separately, so the evidence is not all destroyed next time. 

[mhpr262]

2 hours ago, Seba said:

...and this is why I recently bought an 18XL to be a successor of my 18XL. And in light of all recent incidents and my own experiences with S20, I'm happy that once again I choose a well known, matured, tested and absolutely reliable wheel.

Sadly buying an older "proven" wheel doesnt protect you from the manufacturer secretly switching to cheaper components, cutting quality control measures or some supplier cutting corners and selling inferior, faulty or fake parts to KingSong. I remember reading that there recently was a ton of issues with the Inmotion V10F.

[supercurio]

1 hour ago, Cress said:

We should thank EUC Distributers for negotiating with EUC manufacturers.  Well-informed distributors are positioned to influence EUC manufacturers to make better choices.  Reference @RagingGrandpa the EUC manufacturers are choosing high watt/hr-capacity cells without protecting charge/discharge rates.  It's a follow-the-money issue where the Manufacturer is biased to protect his profit from the batteries and profit from (miss-leading) EUC performance specifications.

This note is from Jason @eWheels:

eWheels Support <support@ewheels.com>
To:richard cress
Cc:Ewheels Support
Mon, Mar 28 at 11:37 AM
Hello Richard,

Yes, it's a total fiasco, 100% King Song's fault. For the past year, I've been trying to have them have the whole system evaluated by a 3rd party, particularly the battery packs; it's the typical situation where these manufacturers think that they have all the answers, fail to understand risk complexity, and here we are...  

The outcome of this disaster is that it now gives us a better negotiating hand for using high-powered cells, outsource pack production, require our own battery stress testing plan, battery pack certification & more extensive reviewer testing.

You can imagine that we're rather inundated today.

Thanks Richard,

Jason 

Awesome @Jason McNeil replied here, and thanks for broadcasting it here.

Now we can understand why suddenly Kingsong mentioned 3rd party evaluation. It's 100% under pressure of the major distributors, and only after a catastrophic failure.

1 hour ago, techyiam said:

It looked like his max accel / max braking stress cycle was becoming more frequent just before the cutout too. It looked to me like he was doing a pendulum type stress test, but at speed as oppose to stationary. Perhaps this type of user-error induced stress test can become his stress test trademark, not unlike Overheat Hill stress test for Marty, except Marty does pause when the euc gets close to overheating.

Yes pendulum is an overheating type of test since there's high motor current and essentially no airflow.

Max accel / max braking test will break some more wheels today, and/or might damage a few cells if completed at full charge.
The amount of energy to dissipate for hard braking is huuuge, and although taking 7-10kW out of a pack is one thing, putting it back near max voltage is between impossible and dangerous.

Do this accel/braking test on an incline and many will go poof.
My Sherman did  - not even at max voltage (although I caught it before it would catch fire, but 1 cap and a few more components were busted)

 

An exercise in speculation, connecting things: will probably be proven wrong

• During hard braking, regen pushed the board & BMS voltage beyond 126V
• Kingsong S20 BMS might have shut off output when reaching 126+V to prevent cell over-voltage
• With nowhere to dump the energy from regen besides the tiny capacitors, voltage shoots up to 150+V beyond the MOSFET or 160V+ capacitor capability

If that's a realistic scenario, then Kingsong would have a firmware quick fix to let the BMS allow higher voltages on regen, instead of blowing things up.
But at the expense of cells health (overvoltage potential)

In any case, this failure didn't happen randomly: it's directly connected to hard braking on near-full battery IMHO.

56 minutes ago, fryman said:

Can we get another S20 tester to test their wheel with the same riding style?

I talked to a reviewer and he was forbidden to ride it now: probably won't happen.

Edited March 28, 2022 by supercurio

[.....]

Now the dance begins. Ks gets to dance and explain why this happened and how it can and has already been fixed. Of COURSE its something that can easily be fixed by simply moving a one or zero in the coding.  Along with KS dancing, we will see pre-order customers doing the dance. You know, the dance that ensures its a fluke, everything will be all right, and 'hope' overcomes reality. No worries for those that have already signed up to the s20 party. Your reluctance to lean on the reality and maintain hope, is admirable. Misplaced, but admirable. Probably the same people who still think the s18 is a wonderful wheel. Obvisouly these arent people who have owned PROVEN wheels of the past. If your expectations are low enough and hope is deep enough, youll be happy riding a frozen turd while bragging about its smell. Brakes, who needs brakes? If a wheel catches fire because you try to apply moderate braking, well... Don't lean back and try to slow down, problem solved, nothing to see here....

I'm content with my 18xl. I'd like to join the dance with everyone, but I already know better. I buy wheels to have fun and ride around. I buy torches to start fires. I think its a neat idea to combine the two tools, but its simply not for me. Of course, this kind of thing NEVER happens and we have been shown we can fully TRUST these manufacturers, to repair and properly test these wheels. YOu now, same as they always do, prior to releasing a wheel.

Fwiw, a fire like that out here in rural america= acres and acres of property damage. Even in the middle of my two lane road, flames like that would easily catch the ditches on fire and escalate, LONG before the volunteer fire dept shows. So, how much IS 200 acres of appalachian mountain land worth? I'll tell you, its worth more than the gains from rolling the dice when buying ANY of these newer wheels. How many companies and how many times do we need to see this crap, BEFORE we realize that quality and safety have long since been off the agenda. Its all about sales and marketing and keeping up with the Jones'.... duh!

Edited March 28, 2022 by ShanesPlanet

[fryman]

18 minutes ago, supercurio said:

In any case, this failure didn't happen randomly: it's directly connected to hard braking on near-full battery IMHO.

Unless they had a different charger than what everyone received with their S20, they would have only been able to charge to 96%. Something else to think about

[fryman]

35 minutes ago, RagingGrandpa said:

Please!

And if it's the same configuration as the demo wheel that burned, retrofit additional, lower-rating fuses in the battery wires. 
A controller failed. And then the pack burned. Let's study them separately, so the evidence is not all destroyed next time. 

Have EUC World running and save the data.

[Paul A]

Lots of news, R&D, investment, pilot production etc, in the space of solid state batteries.

Major car and electronic companies all racing for breakthroughs and market dominance.

Fall safe will eventuate.

Consumers want everything and now.

[supercurio]

18 minutes ago, fryman said:

Unless they had a different charger than what everyone received with their S20, they would have only been able to charge to 96%. Something else to think about

On hard braking, the system voltage can easily pear 2V+, so the wheel doesn't necessarily have to be charged to 4.2V/cell. to reach 126V+ system voltage or quite a bit more.
That's also by pushing significant currents as well.

Edited March 28, 2022 by supercurio

[RagingGrandpa]

34 minutes ago, supercurio said:

the wheel doesn't necessarily have to be charged to 4.2V/cell. to reach 126V+ system voltage

I don't see the connection to overvoltage... 

Considering the cells by themselves, there is no strict threshold voltage after which they suddenly burn. 4.2V is a safe charging guideline, but real trouble doesn't start until well beyond 5V. And I've personally taken M50T's to 6V per cell, without causing venting. 
Heating is always a concern, during discharge and charging. But this S20 fire was only a few minutes into UStride's first test ride, on a cold day. I find it hard to believe that there was enough stress to the system to significantly heat the cells, that day.

Also bear in mind, there is very little regeneration braking available at low speeds. Instead, the controller is applying battery current to create reverse torque, the same as it would do to drive the motor in reverse. This "plug braking" mode is discharging the pack, not recharging it. A voltage recording would help illustrate this... but if we consider aggressive braking from 20mph, at the beginning it's mostly regen, and at the end it's mostly reverse-drive. 
Pendulum maneuvers (all below 10mph) have almost no regen.

So the only 'overvoltage' concern when the system has a momentary transient above 130V is: the electronic components. But there were no obvious problems with S20's hardware design: main capacitors rated to 160V... FETs to 150V... 

My opinion, echoing others: this looks like an overcurrent (and therefore local overheating) problem in the controller. Not overvoltage; and not related to pack temperature. And then a second problem regarding fault-current-interruption, which allowed the pack to be destructively overloaded by the failed controller. 

This also could have been made worse by single-component defects (in the cells, BMS, or controller), which are always statistically possible.
The cells and BMS seem rather innocent from 'random defects' though, since both the left- and right-side packs burned simultaneously, despite having completely separate containers, connections, and fuses. A design issue, then 
 

Edited March 28, 2022 by RagingGrandpa

[Seba]

51 minutes ago, mhpr262 said:

Sadly buying an older "proven" wheel doesnt protect you from the manufacturer secretly switching to cheaper components, cutting quality control measures or some supplier cutting corners and selling inferior, faulty or fake parts to KingSong. I remember reading that there recently was a ton of issues with the Inmotion V10F.

If a wheel is well designed, switching to cheaper components won't make it less safe. Also, 18XL is a classic design with proven 18650 cells. It doesn't matter if it's made with LG MJ1 or Samsung 35E; battery will be safe, strong and will exhibit great discharge characteristic. Moreover, 18XL has no record of using cheap components, the only reports of faulty units are from poor workmanship, something that can be found in products of every EUC manufacturer. In fact, 18XL has been improved with each batch. Unreliable analog lift sensors were replaced with microswitches. Telescopic stems has been improved. Mainboard has been improved. Headlamp PCBs has switched from glass epoxy to aluminium substrate. Actually, everything that could be improved has been improved.

Now I have two 18XL - one has almost 30 000 km on the clock and still is in mint condition in terms of key components - battery, controller and engine. I just wanted another wheel as a spare and for local rides. First one still serves me well and will still serve as my first wheel for long-distance rides. Finally, I decided to get another 18XL so I don't have to change my habits. And it seems that it was the best decision.

S18 is an inherently broken design and I'm afraid that at least in some areas S20 is inherently bad too  Looks like adding a smart BMS didn't resolved any issues or weaknesses of classic, bleeding BMS. In fact, it may be that smart BMS in its current form/implementation contributed to this accident. For me, recent wheel designs are more and more distant from a key feature of an ideal EUC - SIMPLICITY. Unfortunately, wheel manufacturers decided to make their new wheels more and more complicated. In some areas, I would call them OVER-ENGINEERED. Which is bad, really bad, especially if you don't have real-world, well-funded R&D.

[Finn Bjerke]

I added music to the fil for the sarcasm: 

 

[RayRay]

[supercurio]

9 minutes ago, RagingGrandpa said:

Considering the cells by themselves, there is no strict threshold voltage after which they suddenly burn. 4.2V is a safe charging guideline, but real trouble doesn't start until well beyond 5V. And I've personally taken M50T's to 6V per cell, without causing venting. 

@RagingGrandpaYes I agree with you, cells won't burst in flames if they're charged a bit over 4.2V, especially if not at full capacity at the time.

I only meant as an answer to @fryman that yes the controller can be pushed to higher voltages than 126V briefly during hard braking even if the charger provided is only charging the packs to 125V.

9 minutes ago, RagingGrandpa said:

Heating is always a concern, during discharge and charging. But this S20 fire was only a few minutes into UStride's first test ride, on a cold day. I find it hard to believe that there was enough stress to the system to significantly heat the cells, in this case.

Agree also very doubtful that it would be cells overheating with a bit of load in, but not that much in less than 1:30

9 minutes ago, RagingGrandpa said:

Also bear in mind, there is very little regeneration braking available at low speeds. Instead, the controller is applying battery current to create reverse torque, the same as it would do to drive the motor in reverse. This "plug braking" mode is discharging the pack, not recharging it. A voltage recording would help illustrate this... but if we consider aggressive braking from 20mph, at the beginning it's mostly regen, and at the end it's mostly reverse-drive. 
Pendulum maneuvers (all below 10mph) have almost no regen.

Yes I know it exists, although it's usually not reflected much in phase amp (can't?) or battery amps on KS wheels.
It would be really cool to see the ratio of energy/regen/plug braking happening on real-time data.

Including to optimize your speed when you want max regen going downhill.

9 minutes ago, RagingGrandpa said:

So the only 'overvoltage' concern when the system has a momentary transient above 130V is: the electronic components. But there were no obvious problems with S20's hardware design: main capacitors rated to 160V... FETs to 150V... 

Yes that's the one I'm thinking about.

I read somewhere that the feedback from KS was that here's a chip on the board recording data and that it looked intact, so they might be able to extract telemetry from it.
First time I heard it mentioned. I hope it's true tho!

9 minutes ago, RagingGrandpa said:

My opinion, echoing others: this looks like an overcurrent (and therefore local overheating) problem in the controller. Not overvoltage; and not related to pack temperature. And then a second problem regarding fault-current-interruption, which allowed the pack to be destructively overloaded by the failed controller.

👍 Yes same local overheating that blew 1 capacitor and a few more components on my Sherman board when going up and down a hill, with some acceleration and braking to test my pads setup at the time (foolish me!)

 

Edited March 28, 2022 by supercurio

[Tawpie]

3 minutes ago, Seba said:

I would call them OVER-ENGINEERED

Over-featured, yes. Adding features adds complexity, and adding complexity opens more doors open for Engineering oversight just because there are more things that could go sideways. But it's semantics. KISS still applies!

[.....]

15 minutes ago, Seba said:

If a wheel is well designed, switching to cheaper components won't make it less safe. Also, 18XL is a classic design with proven 18650 cells. It doesn't matter if it's made with LG MJ1 or Samsung 35E; battery will be safe, strong and will exhibit great discharge characteristic. Moreover, 18XL has no record of using cheap components, the only reports of faulty units are from poor workmanship, something that can be found in products of every EUC manufacturer. In fact, 18XL has been improved with each batch. Unreliable analog lift sensors were replaced with microswitches. Telescopic stems has been improved. Mainboard has been improved. Headlamp PCBs has switched from glass epoxy to aluminium substrate. Actually, everything that could be improved has been improved.

Now I have two 18XL - one has almost 30 000 km on the clock and still is in mint condition in terms of key components - battery, controller and engine. I just wanted another wheel as a spare and for local rides. First one still serves me well and will still serve as my first wheel for long-distance rides. Finally, I decided to get another 18XL so I don't have to change my habits. And it seems that it was the best decision.

S18 is an inherently broken design and I'm afraid that at least in some areas S20 is inherently bad too  Looks like adding a smart BMS didn't resolved any issues or weaknesses of classic, bleeding BMS. In fact, it may be that smart BMS in its current form/implementation contributed to this accident. For me, recent wheel designs are more and more distant from a key feature of an ideal EUC - SIMPLICITY. Unfortunately, wheel manufacturers decided to make their new wheels more and more complicated. In some areas, I would call them OVER-ENGINEERED. Which is bad, really bad, especially if you don't have real-world, well-funded R&D.

Stop Seba, you're making too much sense. Why would I want to buy a wheel I can trust, when I can buy the newest fancy thing that I can't? Same rules apply in golf. Old clubs work fine, but I surely can't been seen playing with my rich buddy who has the newest clubs. Performance is only a second consideration.  If a child can barely handle basic math, OF COURSE you would expect them to handle calculus w/o issue or studying. Add some complexity and what was already too tough to manage, will be easier, wont it? 

Maybe the problem you have @seba, is you simply havent reached the level of faith it requires to embrace the success of owning an 80lb flammable paperweight. Live a little! 18650 cells are too reliable. Why on earth would you invest in something reliable, when you can buy newer battery designs and circuits that arent? Cmon man, keep up with the program. People don't want proven and reliable, they want NEW and exciting at any cost. You should work on your ability to ignore the obvious.

Edited March 28, 2022 by ShanesPlanet

[Tawpie]

A thought on *cheap* capacitors... when they blow up, it's pretty much always because they got too hot and the electrolyte boiled. This heating usually happens two ways: an internal short due to failure of the dielectric separation layer (over voltage tends to trigger this but manufacturing defects come into play as well), and excessive ripple current (current going into and out of the capacitor) that converts to heat due to the cap's internal series resistance. The ripple current rating of the cap is every bit as important as its voltage rating and cannot be ignored in choosing the part. Some cap architectures are better at handling ripple current than others—electrolytic (cheap) ones generally... not so much. Some caps can self-heal a short in the separation layer, but those are typically physically large and expensive because they involve very thin film construction techniques.

Edited March 28, 2022 by Tawpie