12 volt battery died without warning

12 volt battery died without warning

Yesterday after work, I got a surprise. The screen on my TM3 did not turn on. After getting Tesla support on the phone, we figured out it was the 12 volt battery. Also found out the Tesla App could open the frunk even without the 12 volt battery. After figuring out how to get to the 12v battery and a jump start, I got home. Jumping the TM3 was easy but I had to do it again to get to work. Now it is a waiting game for the road ranger to call me to schedule an appointment for the new battery. I think I will buy an emergency portable jump start battery for TM3.

Tronguy | 19 February, 2019

Aaanndd.. Here's where we get into the infamous bathtub curve and FIT rates.
Build something, anything. Probability of failure right off is rather high, which is why many, many things are tested right at manufacture. Something new might have a 5% failure rate; something designed a while ago might have a 1% failure rate, because the manufacturing engineers get better with a particular widget, whatever it is, over time.
That's the beginning of the bathtub curve. As time moveth on, these early failure eventually drop into the noise.
The noise: Random failures. Take a box of 100 lightbulbs. Plug them all in. One might blow on first turn-on; probably one that escaped the initial testing at the factory. Over time, one or two might fail and random. That's the "noise" factor. There are processes that kill incandescent light bulbs. When a light gets hot, electrons actually fly off the wire, wander around a bit, then get back on the wire, which is why when a light bulb is glowing, it's not just the filament that's white-hot, there are emissions from the flying-around electrons, too. Further, not only do electrons fly off, sometimes atoms in the filament do, too. Some of those recondense on the filament; others on the inside of the lightbulb. After time, the filament gets thinner; where it gets thinner, it gets hotter; and more ions fly off. Eventually it gets hot enough (or fragile enough) to melt/break, and one has a dead lightbulb.
So: the end of the bathtub curve is the wear-out section, where the probability of failure rises dramatically.
The whole business is probabilistic. Maybe the wire has an occlusion that makes it die early; maybe it's been built thicker by chance, so it dies 'way late.
The failure rate in the bottom of the bathtub is random, Poisson distributed, and is usually given as a rate in FITs, which means, "Failures in 10**9 hours".
A single resistor is normally figured to have a failure rate, at the bottom of the bathtub, of 1 FIT. Which means if you've got a billion of the suckers, you'll get one failure every hour :-).
Integrated circuits have FITs on the order of 10-1000, depending upon the complexity. A motherboard on a PC might have a 30,000 FIT rating. But there's 8,760 hours in a year, so the probability of failure of a motherboard in a given year is 8760*30000*1e-9 = 26.2%.
So.. That PC has a 26% chance of failure per year; which means it might fail in the first month, not in that year at all, or 10 years down the pike. And there are wear-out mechanisms, all of which are accelerated by heat; metal migration inside of ASICs, for example. And, if one has a hard drive with bearings, there are mundane processes like bearing wear-out that can kill those, too.
The point behind all of this: Batteries are subject to FIT rates, too, and have wear-out mechanisms. It's not unheard of for a 12V battery to fail due to the early wear-out mechanism; that's why aftermarket batteries have warranties. Many of those warranties are pro-rated.. Because a customer will be upset if a random failure hits a battery one year into use, but if the manufacturer gave a free replacement, they'd have to charge more, be uncompetitive compared to their rivals, and so forth.
So: Not wildly unusual on a new car that a 12V battery failed, and bad luck, probably no worse.
And this is why manufacturers keep a quality group handy, tracking failure rates, so that if something pops up higher than expected, they can _do_ something about it. Change suppliers, change manufacturing processes, and push the failure rate down.
(And this is one of the reasons CU says never to buy the first year of a given model.)

derotam | 19 February, 2019

@kiwank... thanks for your Feb 19th diary entry. Don't know what the point was.

@Tronguy, TLDR

hokiegir1 | 19 February, 2019

When I had my miata, the 1 year old 12v battery died while I was parked at the airport for a week, despite everything being off/closed correctly. It happens. A ump will get you back on the road, and it can be replaced relatively quickly -- and likely under warranty at no cost.

kiwank | 19 February, 2019

detrotam: I guess the point is that a new 12 volt battery can fail and also you can access the frunk using the app to get to the battery to do a jump start. I am a TM3 fan and I wasn't aware of this until it happened to me yesterday. My appointment to get the battery replaces from a road ranger is at 7:00 AM tomorrow. I am just trying to share what I found out.

SteveWin1 | 19 February, 2019

Wait, so you don't have to pop the little tow hook cover off of the bumper and apply 12V to those cables to get the frunk to open? You can just open the app and open the frunk? The high-voltage battery can pop the frunk on its own? How does that work? If there's power somewhere that can pop the frunk, then why can't it just flip the contactor that connects the high voltage to the rest of the car -- including the dead 12V to charge it back up?

Neomaxizoomdweebie | 19 February, 2019

@Tronguy, Good read! Thanks for the interesting explanation of normal failure rates.

derotam | 19 February, 2019

I have a feeling there was enough juice to allow the app to still connect to the car and pop the frunk, but not enough to turn on the display system.

Curious though... did you even attempt to put the car in gear and drive?

Magic 8 Ball | 19 February, 2019

@Neomaxisoomdweebie Do you believe a PC has a 26% chance of failure per year?

1FIT = 10^9 hrs MTBF
30,000 FIT = 10^13 hrs MTBF

kiwank | 19 February, 2019

derotam: yes I tried to put in gear and nothing. Even the window had problems opening and closing. The car was pretty much dead. The jump start allowed the car to turn on and everything was normal except for the warning to replace the 12v battery. Yeah the app opening the frunk was a surprise because I was told by support on how to open the frunk through the tow hook cover and I did it on my own because I was playing with the app. The app was showing connecting and never connected but it did open the frunk with the command. Not sure how that works.

derotam | 19 February, 2019

30,000 FIT = approx 10^4.52 hrs MTBF
MTBF = 1,000,000,000 x 1/FIT

Neomaxizoomdweebie | 19 February, 2019

M8B - No, I think some of his math is wrong and/or he is making up arbitrary numbers as a "for instance". I enjoyed the read because it's been a while since I thought about the statistics of product failure. I did a Lean 6 Sigma blackbelt course about a decade ago - Good times.

Neomaxizoomdweebie | 19 February, 2019

Although... Every time you turn off your computer you stop the clock on failure rate. Maybe a computer that runs only 8 hours out of the day has a 26% failure rate every 3 years.

Magic 8 Ball | 19 February, 2019

@derotam is correct, I forgot it was the reciprocal. Yeah I did the whole blackbelt thing but have already forgotten more than care to admit. I still remember that if I play let's make a deal that I should change my mind when it comes to picking between three curtains, tho' ; ).

Magic 8 Ball | 19 February, 2019

I think the 26%, if I follow, should be the percentage chance that the component will make it all the way to the calculated MTBF.

Neomaxizoomdweebie | 19 February, 2019

I initially want to say 26% is high because I've never observed roughly a quarter of my workplace computers failing every year.

Neomaxizoomdweebie | 19 February, 2019

MTBF - Mean Time Between Failure - half of your sample exceed MTBF, half do not, by definition.

1agkirk2 | 19 February, 2019

Tronguy Good response, thanks for taking the time to remind us of the importance of Reliability Studies.

kiwank | 19 February, 2019

I must not be geeky as you guys. I thought I was pretty geeky electrical/software engineer but no way compare with the discussion on MTBF on why my battery failed.

Magic 8 Ball | 19 February, 2019

The take home is that most failures will occur either as infants (a low percentage) or fail after a normal, predictable, time.

stonerje | 19 February, 2019

So is it ok then to receive a jump, if the 12v dies? I thought it was a bad idea to give others a jump.

surfpearl | 19 February, 2019

@stonerje - Those are two different things. The Manual (page 165) clearly says it's OK to receive a jump if the 12 V battery dies, but not OK to give others a jump.

httran26 | 19 February, 2019

Was the main battery below 20%?
I'm always uncomfortable with leaving the car sit when the main battery is below 20%. This is the point where the main battery no longer charges the 12v battery.

Bighorn | 19 February, 2019

False. Where’d you get that?

kiwank | 19 February, 2019

httran26: No the battery was about 70%.

kiwank | 19 February, 2019

magic 8 ball: Yeah about 7 months and 15000 miles. Not sure if it is an infant but failed pretty early.

Magic 8 Ball | 19 February, 2019

@kiwank Lead Acid batteries aren't what they used to be. It seems to still be a mystery as to why car mfrs. don't use something else for 12V. I agree yours does not seem like an infant and would be a rarer case of mortality in the flatness of the tub.

Tronguy | 19 February, 2019

My apologies, guys: Should have put up the units.
OK, so 30K FITs may be a little high. Let's try for 7000 K FITs. Probability of failure in a year:

(7000 Failures/10**9 hours) * (24 hours/day) (365 days/year) = 0.061 failures/year, or a 6% probability of failure.

Or, if you want to play MBTF, 1/(failure-rate) = 1/(0.061) = 16.31 years/failure.. On average.

Put in 30,000 FITs and you'll get 26.1% and whatever 1/0.261) is.

Now, this is assuming one is playing with (a single busted whatever) = (totally dead PC). As you all know, life ain't simple. Suppose one is using ECC memory? Lose a RAM, you don't lose the PC.

And, speaking of electronics a bit more. It's been a couple decades since I Did This Stuff For Real, but FIT rates for ASICs can be gathered. Your typical 8-pin op amp has a FIT rate of around 10; something with a couple hundred pins has a FIT rate around 200 or so, and there are actually documents that will give you the FIT rate for a given technology and pin count. (The more pins on a device.. the more likely one will break :-).)
A typical 12' by 7" circuit pack, the kind that I used to work on, would range anywhere from 1000 FITs if it had minimal components and above 20,000 FITs if it was a spicy meatball with hot components. There's this equation, the Arrehenius, that relates failure rates to temperature, and there's an exponential in there. One way to _get_ the FIT rate of a part is to heat a batch of them up to just under the level where silicon dies and let it run for a month. Play statistics with the results, and FIT Rates R Us.
So, take a random computer on, say, the Tesla. One can look at one of these and actually make a reasonable estimate of the FIT rate by counting up the number of chips on one's hand, guessing at the pin counts, and calculating away with the more-or-less standard tables. I've actually done this with a Prius engine controller when somebody posted a picture of one they'd taken apart from a junker on a forum. I came up with around 5% a year failure rate.
Fun fact: Everybody knows that the _actual_ failure rates are two to five times less than the calculated failure rates. But, so long as everybody is doing the calculations the same way, then one can safely compare apples with apples.
These kinds of calculations show up in interesting places. For example, sparing. If one knows that box X has a calculated FIT of something-or-other, and there's yea many dealerships over the landscape, and bunch_of_cars_all_with_that_part driving around, then one can stock the dealerships (or not, if the FIT is low) with those parts and be assured that said parts won't gather too much dust. And keeps the customers happy, since they won't have to wait forever and a day for the part to show up. You get the idea.
Where it gets to be really fun is when one works with fault tolerant hardware. With two totally redundant computer boards, say, with 30K FITs each, and perfect (ha!) fallover, the probability of both of them failing in a year (resulting in an outage) is something like
(30K*30K)/(1e9*1e9) * (365*24) = 0.00078%. No, I didn't drop a digit, there :).
You can bet the NASA guys (extreme case, here) where repair in outer space is extremely unlikely, do fault-tolerant protection like one wouldn't believe.

kiwank | 20 February, 2019

Road ranger came on time at 7:00 AM. There is a process for replacing the 12v battery on a Tesla. You just can't take it to anywhere to replace it. My driver side window had to be re-calibrated because of the loss of power. Road ranger is a great idea and hopefully after my warranty expires, it still will be available for a reasonable cost.

Neomaxizoomdweebie | 20 February, 2019

kiwank, had your 12V battery by chance leaked any fluid that resulted in white crystals under your car?

kiwank | 20 February, 2019

Neomax: the 12V battery showed no sign of any leaks. I haven't noticed any white crystals in my garage floor.

Bighorn | 20 February, 2019

My ranger drives 700 miles and has never charged me anything despite being way out of warranty.

Atoms | 21 February, 2019

Yep, lead acid batteries fail and some earlier than others. Tesla mobile service should be able to repair within hours wherever you are. Likelyhood of another failure is low unless there is another issue.

SalisburySam | 22 February, 2019

For the record @Tronguy I hate actually learning something before completing my 2nd cup o’ joe, but I have to admit your posts were fascinating. Thanks for the details. Helped explain why my new Keurig died and I had to drive to Starbucks in my pajamas last week.

slevinn | 19 March, 2019

As I read this post and others that point out it's "ok" that the 12V battery failed in the Model 3 because it is not a common occurrence I think we are not seeing the full picture.

I have had a Tesla since 2013 (Model S, Vin #8xxx). During the past 6 years I had many issues (documents on an earlier post on the Model S forum). Among them were several 12V battery replacements. I did not think it was unusual to have 12V battery issues or any of the other issues I dealt with in those days because it was a new car and a new car company. Additionally, Telsa service was beyond accommodating. They picked up and returned the car and usually responded the same day or next day in the worst case.

Fast forward 6 years - my wife got a new Model 3 in October 2018. She drives it every day and has had a few issues but nothin serious. This evening we got home from a 3 day vacation and her car is dead. After speaking with support we know the 12V battery died. Tee car was plugged into a Tesla charger and the main battery is fine.

I am assuming the Tesla logs showed the 12V battery was dying (because service told me they saw it in the logs). There was no notification from Tesla to let us know either in the car before the past 3 days or via a call or email from Tesla.

Time frame for mobile support to come out and change the battery: 4 -6 weeks. So much for mobile service - they are overwhelmed.

Tesla is going to pick up the car tonight (9PM in the east) and tow it the the very busy NJ service center. I will find out in the morning if they are going to fix it in a day or hold it for longer. I fully expect Tesla to do the best they can under the circumstances (overwhelmed service center in NJ and more cars delivered every day and I appreciate the pickup tonight.

The problem is the 12V battery issue has been known for at least 6 years and it appears that Telsa is no closer to solving it then they were in 2013. My wife's driving pattern is mostly local driving (about 3,000 miles on the car since October) and I know from reading many posts that this causes more stress on the battery than long distance trips.

If Tesla service had sent a ranger out tonight or tomorrow morning to fix the problem I don't think I would have as big an issue (in my house) because it would have felt like this particular problem can be easily dealt with. She does not think it's ok because she happens to be on the wrong end of the bathtub curve.

If her driving pattern means that Tesla is not a good fit for her, then I think Tesla has an even bigger problem of they want to sell to the masses.

jjgunn | 19 March, 2019

4-6 weeks for mobile service?

I vehemently disagree.

I had a body control module replaced. This controls the self-presenting doors on a Model X. SC also reprogrammed my one key fob. 2nd key fob was not in their possession so I needed it re-keyed. Proximity worked but not the buttons.

Mobile Service came out within 48 hours & re-keyed both fobs correctly.

I've heard of people occasionally having the 12v battery die but not multiple times over 6 years.

If true, your case is very rare & not representative of 99% of the Tesla fleet. If true, it sounds like something is draining the 12v. A short or something is causing your car to "eat" 12v batteries.

Best of luck ...

beaver | 19 March, 2019

Anyone know why they don’t make a stand-alone Lithium ion 12v battery that can last 10 or more years so we don’t have to worry?

beaver | 19 March, 2019

Ok so I did some searching. The answer is cost, you can buy 12 v lithium ion batteries today but they cost $500-1,000

Lithium Pros C680 12V Lithium-Ion Powerpack Battery with Top Mount Battery Terminal

GeorgeA | 19 March, 2019

I have always wondered why the main battery cannot be utilized with a small section of it having a step down in voltage instead of using the small 12 volt battery that at times needs replacement.

If the main battery cannot replace the 12 volt battery in this way, then how about using main battery (create a step down in voltage section ) to jump the 3's 12 volt dead battery automatically by just pressing the brake 3 times for example or using the phone app. to initiate the jump process. An update to the phone app could add a 12 volt battery level indicator or a sensor notice when it needs to be replaced avoiding being stranded or needing immediate mobile service (which has always been timely in my experience). These three ideas could be considered in future model iterations if they are feasible, needed or even justified.

But it does seem ironically unfortunate that the enormous main battery stored power is useless if / when the tiny 12 volt battery fails. Perhaps the Tesla engineers have an innovative new design that resolves any 12 volt concerns since it could be a useful improvement to dependability to the entire model line fleet. My experience over the 6+ years I have owned a Tesla is that the 12 volt batteries have improved and become more dependable.

frizenyeh | 7 February, 2020

I've been Tesla Model 3 owner for a bit over a month. 12V battery already failed on me. I installed 2x my sealed lead acid batteries that I got from computer backup power system just to make the car run. It runs OK (though the warning shows up after few days).
Detail story:
I went out of town on Jan 26th 2020. came back on Jan 31th. Car was plugged in to 120V outlet whole time. First drive, it shows the replacement warning. I scheduled Tesla service on Feb 6th immediately (that was the soonest available). Next day (Feb 1st, 2020), on my phone app it shows some weird stuffs (for example, it tells me to check my power source. and the charging amp goes to 0/45A while there's only 2V). I checked at my car, tried to open door. It did open. but window kept rolling down every 5 second. I tried to press brake. The brake had weird sound and system (center screen) did not start. I can't do anything. I got out, and turns out I need to jump open frunk to access 12V battery. I did with my adjustable power supply. Battery voltage dropped to 6.5V. I pull it out and replaced with a sealed lead acid battery from computer back up system. Car screen lid up and i was able to start the car. I tried to charge the AGM battery that was originally in car to close to 13V. After 1 hour of stand by (nothing drawing power from it), it dropped to 10V. I'm guessing there is internal resistance or there is a short. I brought the battery to O'Reilly and hope to buy a new one. They didn't have it. I brought the battery to Tesla service center. They weren't open on weekend. I contacted Road Assistance on my phone. They told me they could only offer tow to service center. I had to put in 2 of the backup battery (actually 1 should be enough but I put two just to make sure there is enough current) in car just to keep it moving.
On Feb 6th, the service guy came. I told him I used another battery just to keep the car moving. He told me that voids warranty and that I have to pay full price for the battery and/or labor. I told him to just get me a new battery and take the old one. I also ask him if there is a way to check if something in car was wrong that caused 12V battery failure. He told me that Tesla checked the logs and it's just the battery failure. I got an Email of invoice for the battery ($85, $92.65 with tax).

My feel: Terrible, but amusing. Turns out a small backup battery (similar to the one on Amazon for $22) can power up Model 3 computer just fine. I'd imagine that the DC-DC converter will keep getting on more frequently just to charge the battery. But then I'm a bit concerned to go on any trips since the original battery may die on me.

Suggestion from my story: right before service guy came, you should replace back the bad AGM battery to keep it under warranty.

I'm going to buy the Ohmmu lithium battery if this ever happen again, although I'm expecting not to replace the new AGM battery within 3 years. It is said that charging normal lithium ion batteries are bad in cold weather. But they seem to be confident in extreme weather and offer 4 years warranty.

For anyone that's interested, here is a picture of what I did to temporary make it run.

derotam | 7 February, 2020

Sounds like you had a standard random infant mortality on the battery, it happens. I had one go bad in a previous brand new vehicle after 3 months.

Yes, it is always advisable to remove any related modifications before service looks at it. Giving to much information is not always a good thing.

Just like you said about Li-Ion batteries, that Ohmmu is fine, but charging it cold will not be good for it and the space where that battery is definitely isn't heated in any way.

Good work adapting to the situation though to be able to use the car while you wait for service to be open!

billtphotoman | 7 February, 2020

12v lead batteries and flat tires are the two items most likely to strand an EV owner. Fortunately both can be somewhat mitigated (Li Ion battery jumper, tire plug kit).

frizenyeh | 10 February, 2020

Update to my adventure:

I put the new AGM battery that I got from service guy yesterday (Feb 9th 2020). It still shows "12V battery requires service" on screen.

Not so sure how to make it disappear. I remember when I first put in backup 12V battery, the notification didn't show up in first two days.