12v battery issue, Tesla unsatisfactory response

[FalconFour]

Ouch! That’s nice. Sadly you have only addressed a portion of my argument. You didn’t even address my other post.

El touché. Pagination is a b*tch here. Really chaps my ass that everything I wrote on page 14 will be lost to the sands of pages with no way to really call attention back to it (other than me, here, reminiscing).

So, my biggest issue here is with taking numbers out of sterile scenarios and trying to use them to create other stats and figures. Where does this 7W figure keep coming from, anyway? It's a reasonable number, sure.

But let's also consider that: the car has complete control over when it wakes up and recharges the battery. That means there's no need for it to go to a low SOC on the 12v battery, ever. Since SOC is very accurately tracked by near-rest voltage level, it's pretty easy to say "oh, 12.25v? yeah we'd better wake up and charge that".

What would be great to see is some battery-monitor stats/logs with a knowledge of what I've gone over before - regarding how/when it charges and such. Maybe try and get a current reading on the battery. See what the draw is like while it's asleep. See how low the voltage goes before the car decides to kick in and charge it. Voltage maps fairly cleanly to SOC - 12.0v is quite low (near 0%) and 13.0v is quite full (near 100%). So, see what the voltage is over time... how badly is it really being beaten up?

 

[AlanSubie4Life]

That means there's no need for it to go to a low SOC on the 12v battery, ever.

No it doesn’t go to a low SoC. That’s not the issue. The issue is that you are drawing energy from the battery. That is what is wearing it out. The less energy you draw, the longer it will last, all else being equal.

What would be great to see is some battery-monitor stats/logs with a knowledge of what I've gone over before - regarding how/when it charges and such.

Here is what mine is doing right now. I’m burning about 5 rated miles a day in the garage, so that is 5rmi*234Wh/rmi = 1.2kWh/day, or 50W.

Plugging into my formula that yields about 16 hours of sleep time (which looks about right). Note this is NOT how a normal car behaves - wearing out the contactors!

Maybe try and get a current reading on the battery.

Vampire Current from 12V Battery Tracking

@rrolsbe has looked at this. It looks like it is actually perhaps 4-6W. So 7W is in the ballpark (there will be times when it draws slightly more before turning on the DC-DC).

So, see what the voltage is over time... how badly is it really being beaten up?

My battery is discharged from 13.5V or so to 12.8-12.9V about 20 times a day. Again, this is unusual. Others can post more typical behavior from their car sitting unused. Three times a day is more typical, and it is allowed to discharge lower. I think it works out to 100-150Wh per day. In that ballpark anyway.

 

[FalconFour]

No it doesn’t go to a low SoC. That’s not the issue. The issue is that you are drawing energy from the battery. That is what is wearing it out. The less energy you draw, the longer it will last, all else being equal.

Now don't get my quote snip confused that I'm not reading everything, but I must add to this point: I don't think "total energy moved through the battery" is that big a factor. At least, it seems like the chemistry in action might even be more complex than lithium batteries (again going back to "why are they even using lead batteries with all these issues"). Far as I understand it, lead batteries are most happy at full charge, thus to reduce or eliminate sulfation. Just as long as they're not constantly held at full 14.4v all their life (like a UPS battery).

I have noticed strangely sub-optimal usage of that battery, though - when the car is awake, it seems to constantly and intentionally bring the voltage down, start drawing power (under 10a, not the whole car, by virtue of the system voltage balancing between PCS and 12v draw) from the battery while awake, then bring the voltage back up, which honestly probably puts the most cycling on the battery - when really, it should just be holding it at 13.5v and idle. Instead, as in your graph, it plonks up to the top, 14.4v, then down to 13.0v or lower, which drains some charge out of the battery, and back up again. Totally unnecessary since it's entirely within software control to just have the PCS power the whole car all the time (it's energized and ready, just being commanded to lower the voltage), and leave the 12v alone until it's ready to sleep. It ought to be: wake -> charge 12v to 14.4v -> idle battery at 13.5v indefinitely -> sleep.

Instead, it's like: wake -> charge 12v to 14.4v -> discharge to 12.9v -> charge to 14.4v -> loop forever -> charge to 14.4v -> sleep.

I don't think those repeated 14.4v spikes in your tracking graph are caused by sleep/wakes to charge the battery, but instead more likely are the time the car is awake and doing its charging/discharging loop over and over. You think it ever got to sleep during that day?

(Also, the issue of contactors closing and opening, though it doesn't appear to be happening often, isn't really an issue as long as there's no arcing/sparking when it does that. It's the arcing that causes wear-out, and EV tech goes to extreme lengths to prevent arcing. Without opening under load, those things could open and close virtually infinite times and never wear out!)

Still, I'm only really observing this after having taken an interest in 12v behavior when 2020.28 started causing issues with lithium 12v's... and with that came changes in how the 12v is treated and monitored, apparently. Maybe this is "better" than what it was doing before (an unknown "X-prior-state" so to speak). I dunno.

 

[FalconFour]

Also regarding the sleep behavior... I hadn't at all thought about the need/desire to continue pumping coolant to adjust the main battery temperature. I know the car targets 87F (80-something-high, don't recall exact digits) when it's operating, and tweaks the flow configuration and radiator bypass to get there. It also runs the interior fan for quite some minutes to dry the evaporator after driving. Maybe these factors play into how hard it is to get to sleep. Today, while tinkering in the garage after charging, I actually got to observe that absolutely blessed moment when I heard the car "clunk-clunk" itself to sleep, just a couple minutes after unplugging the charger!

I really want to get one of those handy dandy Bluetooth 12v monitor boxes now, as Scan My Tesla goes dark when it goes to sleep - so the 12v can't be monitored while it's dreaming. It'd be nice to know what that chart looks like with the characteristics of a lithium 12v battery...

 

[EnrgyNDpndnce]

Doesn't that mean the battery you bought is not an AGM ?

Must not, because I definitely bought an AGM. Of that I am 100% positive.

 

[scottf200]

... I really want to get one of those handy dandy Bluetooth 12v monitor boxes now,...
It'd be nice to know what that chart looks like with the characteristics of a lithium 12v battery...

I would second that. I'm considering a lithium 12v (ohmmu) when I replace my Xs 12v on a schedule after my last one failed during an OTA.
I monitored the 12v battery voltage during an OTA update

 

[AlanSubie4Life]

it seems to constantly and intentionally bring the voltage down, start drawing power (under 10a, not the whole car, by virtue of the system voltage balancing between PCS and 12v draw) from the battery while awake, then bring the voltage back up, which honestly probably puts the most cycling on the battery - when really, it should just be holding it at 13.5v and idle. Instead, as in your graph, it plonks up to the top, 14.4v, then down to 13.0v or lower, which drains some charge out of the battery, and back up again. Totally unnecessary since it's entirely within software control to just have the PCS power the whole car all the time (it's energized and ready, just being commanded to lower the voltage), and leave the 12v alone until it's ready to sleep. It ought to be: wake -> charge 12v to 14.4v -> idle battery at 13.5v indefinitely -> sleep.

So far, from my observations, I've only ever seen ~13.4V or ~14.3V when the car is awake.

Any lower voltages you see are observed (13.2V and below) when the contactors are open, so that is the only time significant current is drawn from the battery, that I can tell. As I said, seems like my battery has very low capacity, since it drops voltage quite quickly - a little hard to tell most of the time now since it quickly gets charged again (for many people it's allowed to go below 13V - for my battery it seems reluctant to let it get below 13.1V)!

I don't think those repeated 14.4v spikes in your tracking graph are caused by sleep/wakes to charge the battery, but instead more likely are the time the car is awake and doing its charging/discharging loop over and over. You think it ever got to sleep during that day?

I'm sure of it. When the voltage drops below the 13.4V level, the contactors are open. The contactors are opening and closing all the time now. It has not always behaved this way, of course.

Today my contactors have opened and closed more than 15 times so far. I hear them clunking all the time, and it is completely correlated with the drops to the 13.1V level. That’s the “low load (5-7W)” voltage of my battery - 13.1V when loaded with the sleep load, right after a recharge, and then it decays from there. Every now and again it starts at 13.6 and slowly decays, which I cannot explain - the only possibility I can think of is that THAT is actually 5-7W, and what is happening when it drops immediately to 13.1V, the load is actually higher than that. I should really set up a current probe loop!

Hope they are good for those million cycles they are rated for!

Mobile Service will be by to take care of some other things (first actual service for the car in two years - replacing the charge port insulator pins, which are fine, but are also like fragile wine glasses) on Friday, so we'll see what they say about this behavior.

Service said 12V "all looked ok" from a quick remote look.

 

[neurocutie]

You get an automatic “Like” for referencing Fox Mulder, one of the greatest TV characters ever!

And I agree that Tesla likely has the data we’d all like to see, and would love to see them publish it. Not gonna hold my breath, but I’d love to see it happen.

If Agent Mulder can't get the inside data, maybe Agent Denise Bryson can...

 

[FalconFour]

I would second that. I'm considering a lithium 12v (ohmmu) when I replace my Xs 12v on a schedule after my last one failed during an OTA.
I monitored the 12v battery voltage during an OTA update

Excuse me whilst I smash that Add to Cart button with gusto.

I'm sure of it. When the voltage drops below the 13.4V level, the contactors are open

Odd. I've seen mine playing with the 12v level, well, just as I describe, oscillating between charging and discharging, all while awake.

This was late August when I was tinkering with it and saw it discharging the 12v while awake. Haven't really dived deep into monitoring it, as I wasn't involved in this thread's debates until recently

Today my contactors have opened and closed more than 15 times so far.

Holy wow, that is so far outside the realm of my experience I can't even describe it! Is that even a Model 3? What FW?

I get MAYBE 3 sleep/wake cycles a day... even as I drive around, park, do stuff, come back, drive some more, park, etc etc etc... it sleeps/wakes so rarely, it gets fewer open/close cycles than I even have trips. I dunno. Very, very strange...! I'd be curious to take out your 12v battery and take it to AutoZone...

Come to think of it, the car will totally be okay on its own with the 12v removed, if you want to go to an AutoZone, pull out the 12v battery, take it inside, get it tested, bring it back out, and plug it back in. All you've gotta take with you is a 10mm wrench. The car would just pop up an alert saying "?? the 12v battery was removed, idk? ??" and continue operating normally.

 

[AlanSubie4Life]

I've seen mine playing with the 12v level,

Your picture shows 13.4V which is a typical "maintain level" that you would see when the car is in idle mode and not sleeping. DC-DC is on.

It should be normal to just have 2-3 contactor cycles a day if the car is left undisturbed, as far as I can tell. We'll have more data from more people with monitors soon (see @power.saver data for example).

Come to think of it, the car will totally be okay on its own with the 12v removed, if you want to go to an AutoZone, pull out the 12v battery, take it inside, get it tested, bring it back out, and plug it back in. All you've gotta take with you is a 10mm wrench. The car would just pop up an alert saying "?? the 12v battery was removed, idk? ??" and continue operating normally.

Yeah it's a Model 3. I agree this would be interesting, but since I'm under warranty, I'm not going to be doing this. I'm not opposed, it's relatively simple stuff, but under warranty I will let Tesla take care of it - as long as it does not require a trip to the service center (which is dangerous for the vehicle).

My service appointment details came through and on Friday it looks like they plan to replace the 12V. I assume the fee will be waived. (If not, I will not allow the replacement, and I will go buy a battery directly from Tesla, or just buy the battery directly from the technician if allowed.)

For the record, note that the cost of the Model 3 12V battery is $85 - nice and cheap, as you would expect for a non-AGM battery. Good news. Makes the apparent fact that they seem to only last a couple years a bit easier to bear!

Again, not going to pay for the labor to replace, it if it's not being covered properly under warranty (as it should be). $39 is way too much!

I'm not insisting the battery is bad - so I only want them to replace it under warranty if they deem it to be bad. But for $85, I'll buy one out of warranty, if they claim the battery is fine (I wouldn't believe them, and the consequences of a failure are far too dire to take chances!).

 

[AlanSubie4Life]

I get MAYBE 3 sleep/wake cycles a day... even as I drive around, park, do stuff, come back, drive some more, park, etc etc etc... it sleeps/wakes so rarely, it gets fewer open/close cycles than I even have trips

What's the Ah capacity of the Ohmmu battery? Not that that alone would prevent it from sleeping. I would expect a lower capacity to mean more wake/sleep cycles, not fewer. 3 sleep/wake cycles a day is pretty normal, and desirable I think. But most of the time should be spent sleeping. If it didn't sleep most of the time, you'd have 15+ miles of vampire losses a day (would be close to having Sentry on all the time).

With my car's current behavior, vampire losses have increased to in excess of 5 rated miles a day, as you'd expect from the observed duty cycle in the images above.

 

[iwannam3]

Why are the smartest battery guys in the world screwing up lead acid? Did they boycott the obsolete lead acid classes thinking they would never need to know old "buggy whip" tech? Are the lead acid experts 80 years old and retired or something?

 

[AlanSubie4Life]

Why are the smartest battery guys in the world screwing up lead acid?

I don't think they are screwing it up, really. My contention is that they are just using the 12V battery too much. Much more than a regular ICE vehicle, when you work out the total energy pulled from the battery, in typical conditions.

Can't work miracles. If you use the 12V battery a lot, even if you charge and discharge it perfectly and very mildly, with shallow cycles, it still wears it out. Entropy!

So, my claim is it's probably not really the fault of the 12V battery treatment, really. It's more fundamental. They need to use it less (which is actually a much easier problem to solve than inventing a 12V lead acid battery that lasts forever)!

All the other manufacturers have already solved this problem (except BMW probably)! No reason why Tesla cannot. Get it to under 1W when sleeping and unattended. Not really a difficult job. The battery monitor uses 12mW max and you can communicate just fine with it over Bluetooth. That is 500x less power than 6W. They can wake up periodically and use the main battery however they wish (though that could be more efficient too!). But a fire hose of 5W of power while asleep is just really hard to accommodate & recover from - and it forces more frequent wake ups to top off the 12V - topoffs which are super inefficient periods (12500x more than 12mW, though about 70% of that power is pouring into the 12V battery, probably - 95% of which is recoverable, probably), and add insult to injury.

 

[AlanSubie4Life]

12V battery is ~570Wh
10% depth of discharge (required to maintain life for a shallow-cycle, non-AGM battery like this) is 57Wh
At 7W (estimated), that necessitates charging every 8 hours minimum (7W*8h = 56Wh).
Awake state consumes ~170W (overhead, estimated from wall charging efficiency vs. wall charging power curve), with unknown pareto of consumption. It may be that when charging the 12V the awake state is closer to 250W.
If we assume a big portion of this 170W overhead is 14V @ 10A into the 12V, that would be 140W for recharging. So, 30W true overhead (minimum - again, true overhead might be as high as ~110W; I am not sure).
To recharge 57.6Wh at 95% efficiency, requires 60Wh. At 140W (14V@10A), that is ~26 minutes of charging. 140W*0.43h = 60Wh.
The overhead is completely wasted. So, if it's 30W, 30W * 0.43h = 13Wh is wasted for charging. Inefficiency of charging wastes ~3Wh. So 16Wh wasted per charge cycle, minimum.
Totals per day: 24h/8.43hr/cycle = 2.84 charging cycles per day.
Total energy consumed from main battery/day, minimum: (60Wh+13Wh)*2.84 = 207Wh (162Wh 12V, 9Wh 12V Charging, 37Wh overhead)

That's a minimum of 0.9 rated miles a day (207Wh/234Wh/rmi = 0.9rmi). Since 2-3 rated miles per day is more typical, obviously there are usually other losses. 1) To some extent, some of these may be issues with my estimate of 7W - on average it may be higher (worse for 12V). 2) The other source may be that the awake state uses more than 170W sometimes (could be as high as 250W when charging 12V, for example). Inconsequential for 12V longevity. 3) We know it is true that when the battery is not charging and it goes to "maintain" state of 13.3 to 13.4V, the car sometimes sits and runs in idle mode and is consuming between 30W and 150W for however long it is in that state. Inconsequential for 12V longevity, very consequential for vampire.

But about 1 rated mile per day is the absolute minimum with current implementation, assuming optimal sleeping patterns with waking only to charge the 12V. And with that optimally low vampire behavior, 2.84*57Wh = 162Wh per day is being used by the 12V, which really adds up. As has been pointed out, increasing vampire drain a lot by staying in idle is actually probably good for the 12V. But to get to 2-3 rated miles a day, which is more typical vampire drain, it is just 2-3 extra hours of idle mode (150W*3h/234Wh/rmi = 2 rated miles), so it really does not reduce the stress on the 12V much - just a couple fewer hours per day than I have calculated above.

 

[FalconFour]

The battery monitor uses 12mW max and you can communicate just fine with it over Bluetooth.

Okay, not to be an arse, but you're doing it again... starting with one fantasy number ("this is all they need! why don't they just do this!") and then splattering it around through various permutations of math to produce fancy-looking numbers that are utterly devoid of a shred of reality. Politely stop doing that

First of all, the battery monitor is integral to the main pack. It's basically the lowest-level system that exists. If the pack is pulled out of the car, and it just exists as a pack on the ground, then ... yeah, great. 12mW target, go for it. But that number is impossibly small for any kind of embedded system short of a clock battery on a computer. Even just adding Bluetooth (... where are you associating the BMS with Bluetooth here?), that'd add hundreds of mW right there. The Tesla BMS doesn't employ Bluetooth at all.

That's just a battery pack out of a car sitting on the ground, though. Now let's add

• security/lock (BT "phone as key")
• sleep current for MCU to preserve RAM and not have to be fully rebooted (as a PC in sleep using under/around a watt to keep the RAM refreshed)
• cellular standby to be able to be woken from the app
• RFID periodic pings to see if you're holding the card to the door
• monitoring the 12v battery level and other environmental factors to know if it needs to wake up for something

These aren't optional things that can just be disposed of for the sake of saving a little battery. Bet your sweet butt the Tesla engineers have poured untold hours of effort into optimizing away every possible non-essential system. It's offensive that you go around with magical-thinking math to parade around how dumb you think Tesla is, based on these things.

If you want to criticize a system, then great... find out what it is, with evidence, and criticize it. But please don't pull a magical number out of a hat and then derive a hundred other convincing-looking calculations from it. Less-informed people are watching.

 

[AlanSubie4Life]

First of all, the battery monitor is integral to the main pack.

This raises a good question, just for establishing knowledge. Do we know whether the BMS is pulling energy from the 12V when the contactors are open? I would think that the BMS would be inside the pack (on the other side of the contactors?). And I think that’s what was shown in the various tear downs. So it wouldn’t really be relevant to the discussion about energy pulled from the 12V battery (it would be allocated to vampire drain instead). But I am not sure. I think it must be disconnected from the 12V, because they would want constant monitoring just in case something goes wrong with the pack, and it would be impossible to isolate the pack if they ran the BMS from the 12V.

This is a question, with implications suggested, based on a particular answer:

If it is disconnected from the 12V when the contactors are open, we can exclude the BMS consumption from this discussion.

12mW target, go for it.

That’s not what I said was a reasonable target, please do not suggest that I said or implied such a thing. The 12mW was given to provide an order of magnitude idea of what is possible with a given amount of power.

Bet your sweet butt the Tesla engineers have poured untold hours of effort into optimizing away every possible non-essential system. It's offensive that you go around with magical-thinking math to parade around how dumb you think Tesla is, based on these things.

I’m an engineer. I know engineers will work to do what is possible within a given budget. Just suggesting that the observed behavior suggests maybe a somewhat lower target would have been better. It does kind of add up, as you can see, even setting aside the effect on the 12V.

cellular standby to be able to be woken from the app

What is your phone’s standby power?

security/lock (BT "phone as key")

What is your phone’s Bluetooth power consumption?

sleep current for MCU to preserve RAM and not have to be fully rebooted (as a PC in sleep using under/around a watt to keep the RAM refreshed)

What is your phone’s sleep power consumption?

But please don't pull a magical number out of a hat and then derive a hundred other convincing-looking calculations from it. Less-informed people are watching.

I think I have made it pretty clear I don’t know where the power is going, and that there is some small uncertainty on the exact magnitude (which would be easily resolved if someone put a clamp current probe on the 12V line).

 

[FalconFour]

Do we know whether the BMS is pulling energy from the 12V when the contactors are open?

I'm not sure myself, but I peeked through a few things. Looks like there's two halves to the system, and at least one of those halves is powered directly by the HV battery, but it needs 12v to be "woken up" in manufacturing or repair, not much else.

That’s not what I said was a reasonable target, please do not suggest that I said or implied such a thing.

But you did, there was a whole paragraph on it:

That is 500x less power than 6W

(12500x more than 12mW

You might be thinking you wrote this like "look, relative values", but it sure as heck reads like you're saying 12mW is all they should be using, and they're using 12500x more than they should.

I’m an engineer.

Nice to meet a fellow engineer

What is your phone’s standby power?

Not sure, but a hell of a lot less than my laptop, which is a much closer approximation given that the MCU is x86-based. (honestly, a move that I strongly disagree with - the Atom CPU they chose is a piece of $#!+ that would be out-performed by a good CPU from 2007 or a flagship phone of 2016).

What is your phone’s Bluetooth power consumption?

Virtually nothing, but we're talking about completely different systems here. MCU vs. security key. Yeah, you might want to say "but they could do BETTER if they wanted" but we're talking about a company that builds all its own tech... any more integration and (a) it'd be unserviceable, and (b) the company would implode upon itself as they integrate the car into the car to save energy.

What is your phone’s sleep power consumption?

wait you just asked that. literally that. why do you keep talking about phones, anyway? Mobile devices suck, lol. Don't even get me started on how much I hate mobile platforms (talking Android/iOS and the ecosystem here, not "mobile hardware").

I ordered that 12v battery monitor, so we'll get to have some fun on this shortly. Meanwhile, as we crawl through more pages, I watch page 14 sail further and further away...

 

[CertLive]

Ok this is an interesting read from a technical point of view. But can I add one thing. The 12V battery being used is nothing different than one you can typically buy in an ICE car. Now I am not saying it is being treated in the same fashion but the expected life of the average car battery is four to six years. The longer being what I have been used to. So in that respect with the amount of 12v battery replacement complaints 2 years is far too soon to wear it out. Now it is not your fault as the customer to expect better and its also not your fault as a customer that the model 3 is killing these 12v batteries quicker. Unless there is a large product recall due on the 12v batteries due to a batch being made bad. This actually goes against sustainability going through them at a higher rate creates waste.

In my opinion Tesla should replace these batteries at no cost. It seems quite obvious whether it can be fixed via a software update or not there is a common fault here. So yes demand a free replacement with proof from here that it is quite a wide spread issue.

 

[EnrgyNDpndnce]

So in that respect with the amount of 12v battery replacement complaints 2 years is far too soon to wear it out.

Do you have some actual numbers on “the amount of 12V battery replacement complaints” other than just perusing this forum? If so please share, because I haven’t seen any actual data on the topic. We know roughly 500k Model 3s have been sold (at least that’s the number I’m pulling out of memory from an InsideEVs article I read a month or two ago, and I’m 48 so the memory isn’t what it used to be) so if you have hard data on early 12V failures we can calculate the early failure rate. Then all we need is early 12V failure rate for the average car and we can compare the two in order to determine if this is really an issue worthy of over 300 posts to this thread or if this is typical.

 

[AlanSubie4Life]

But you did, there was a whole paragraph on it:

Get it to under 1W when sleeping and unattended. Not really a difficult job.

We can quibble about whether it is a difficult job, but I insist that we not quibble about what I said was a target.

Get it to under 1W when sleeping and unattended. Not really a difficult job.

I agree that with selection bias it is impossible to actually say whether Tesla 12V are failing earlier. It’s just anecdata.

So instead, we have to use science, to put forward a reasonable hypothesis for why Tesla Model 3 might be wearing out the 12V faster. Above.

Some anecdata:

Subaru #1: Battery lasted 5 years, and then I sold the car.
Subaru #2: First battery failed after about 5 years. Note the STi has a very hot engine compartment so would have been a stressful use case. Subsequent non-OEM batteries lasted about 4 years each.
Highlander: 4 years so far. Still going. Unfortunately I have inadvertently killed the battery twice in the last 6 months, and the resulting sulfation will likely kill it sooner now. Guess I need a desulfating charger...though they probably don’t work.

Spark EV: 4 years and counting.

Tesla Model 3: 2 years and counting. Not clear at the moment whether there is anything wrong with my 12V.

Big plus: Tesla battery is cheaper and smaller than any battery I have had in my prior cars. It’s a non-AGM, small, relatively cheap battery. $85.