Near annual replacement of 12V battery is typical according to Tesla Service Tech

[arg]

No need to hope. Look at the recent history of microprocessors and memory. I can guarantee those things will happen. We are talking about "consumer electronics" here. Okay, this item has wheels and a big electric motor. But otherwise it's basically the same thing.

The problem isn't a silicon problem - it would undoubtedly be possible to do much better with the devices they have - so it has to be an architectural problem. And it's easy to see how they got into this hole, much less easy to see a way out.

Presumably, whoever specified the overall 12V system and the mechanical constraints flowing from it assumed the vampire drain would be much lower than it is now - a reasonable assumption, and these mechanical decisions were probably being made long before most of the software existed. Those assumptions leave us with a battery that's hard to exchange (because nobody back in 2010 thought it would need changing often) and no space to go up to a bigger size as a 'quick fix'.

Then there's the question of what went wrong with the vampire drain. Getting a complex system to have good standby power is hard, and needs a lot of planning as there are huge numbers of dependencies between modules - you can only put a module to sleep if nothing that's active depends upon it, and you have a means to wake it up again and you have a means to store or regenerate the data it holds. Evidently in the rush to get the car out of the door in 2012, power management work was thrown overboard (early cars were delivered without it), leaving them with the problem of retrofitting the power management in later software releases. Very likely, some wrong decisions were made as a result: if they were software ones, they can theoretically get fixed but it gets harder and harder as more software is layered on top; if there were mistakes made in the hardware then we are stuck with them at least in existing cars.

 

[Matias]

Thanks for the analysis guys. It seems like this issue is settled. Is it time to move on to speculate what Tesla is going to do with the processors to reduce vampire drain? I would assume all accessory processors for windows, etc are in a deep sleep state or even off. I would think just the basics are on: things like the key fob radio, 3G radio (oh, there is a power hog), and charge monitoring/charger. Anyone care to guess what they have now and how it could be improved?

In any case 50W energy consumption in sleep is huge.

To put it in perspective, check these consumptions

Computer Power Usage | Penn ISC Computing Resources

Average notebook consumes less when in use and about 1W when in stand by

 

[Cottonwood]

In any case 50W energy consumption in sleep is huge.

To put it in perspective, check these consumptions

Computer Power Usage | Penn ISC Computing Resources

Average notebook consumes less when in use and about 1W when in stand by

Exactly!

The MacBook Air that I am typing on now is using WiFi, has a bright screen on in a well lit room, is keeping a tethered iPhone alive with its screen on and running a few programs. All of that is using about 8 Watts while active!

As another example, I have an off-grid site in Southern Colorado, that has a 1/2 mile WiFi radio network link, two HD network cameras each running Linux, a solar controller that uses 1W when the sun does not shine, a separate charge monitor for the 500 A-Hr, 12 V battery, an ethernet switch, etc. All of that runs on 20 Watts.

The Tesla vampire at 50 Watts is rather piggish...

 

[AmpedRealtor]

Exactly!

The MacBook Air that I am typing on now is using WiFi, has a bright screen on in a well lit room, is keeping a tethered iPhone alive with its screen on and running a few programs. All of that is using about 8 Watts while active!

As another example, I have an off-grid site in Southern Colorado, that has a 1/2 mile WiFi radio network link, two HD network cameras each running Linux, a solar controller that uses 1W when the sun does not shine, a separate charge monitor for the 500 A-Hr, 12 V battery, an ethernet switch, etc. All of that runs on 20 Watts.

The Tesla vampire at 50 Watts is rather piggish...

But you can't compare a car to a MacBook Pro. Your Macbook has two processors, some RAM, a motherboard and a screen. The Model S has many, many more interconnected systems and is a far more complicated system. There is n't just one computer in Model S.

 

[SteveS0353]

Great insights Steve!

One follow up question with regards to cycling of a regular 12V battery: If there is a constant power draw from the terminals of a 12V battery but at the same time a steady charge applied that balances out the draw would that even result in any cycling? Basically what I mean is that it's all about voltage delta right, so if shore power was somehow directly charging the 12V battery or if the contactor between the HV battery to DC-DC converter to 12V battery were left open all the time while the car was plugged in, it would be like the short power powering all the 12V electronics directly, without causing wear and cycling on the 12V, right?

Currently (excuse the pun) shore power charges the HV battery, which via the DC-DC converter (400V -> 12V), charges the 12V. Carefully designed, considering the load and it's dynamics, yes, it's would be possible to feed the vampire from shore power and thus remove or at least significantly reduce the 12V cycling. That's a major architectural change from the way the current hardware is implemented.

 

[Cottonwood]

But you can't compare a car to a MacBook Pro. Your Macbook has two processors, some RAM, a motherboard and a screen. The Model S has many, many more interconnected systems and is a far more complicated system. There is n't just one computer in Model S.

You are correct when the car is on, but when it is off, the only tasks remaining are to check a few vitals (temperatures, SoC's, physical position, etc), log them and communicate them back; also, wait for a FOB or network wakeup. Those sleeping functions should take far less power than an active Mac with a screen lit up in a bright room. I am a retired systems and communications engineer and have implemented everything from large communications centers to small, low-power telemetry sites, and this stuff is pretty straight forward.

When you have a 60 or 85 kW-hr battery, 1.2 kW-hr per day does not seem too bad, especially when there is a "meet the schedule or the company fails" goal for delivering the Model S. All I am saying is that for many reasons including the life of the 12 V lead acid battery, lengthy storage, and the background load that reduces the overall efficiency of the car, the piggish vampire needs to be tamed. A combination of some straight forward hardware redesign and intelligent software could easily reduce the vampire load by a factor of 10 or much more.

- - - Updated - - -

Currently (excuse the pun) shore power charges the HV battery, which via the DC-DC converter (400V -> 12V), charges the 12V. Carefully designed, considering the load and it's dynamics, yes, it's would be possible to feed the vampire from shore power and thus remove or at least significantly reduce the 12V cycling. That's a major architectural change from the way the current hardware is implemented.

It would be a bandaid (see above), but all that is needed to do this is a little, ~80W charger that sits behind the J1772 interlock, and charges the 12V battery when the main 10-20 kW charger is inactive and AC power is available. Cost for such a 12 V battery charger (maintainer) is probably less than what a 12 V battery costs, and if done reasonably well, would probably make the 12 V battery last as long as most Teslas, or at least as long as the 12V battery in an ICE car.

 

[SteveS0353]

The problem isn't a silicon problem - it would undoubtedly be possible to do much better with the devices they have - so it has to be an architectural problem. And it's easy to see how they got into this hole, much less easy to see a way out.

Presumably, whoever specified the overall 12V system and the mechanical constraints flowing from it assumed the vampire drain would be much lower than it is now - a reasonable assumption, and these mechanical decisions were probably being made long before most of the software existed. Those assumptions leave us with a battery that's hard to exchange (because nobody back in 2010 thought it would need changing often) and no space to go up to a bigger size as a 'quick fix'.

Then there's the question of what went wrong with the vampire drain. Getting a complex system to have good standby power is hard, and needs a lot of planning as there are huge numbers of dependencies between modules - you can only put a module to sleep if nothing that's active depends upon it, and you have a means to wake it up again and you have a means to store or regenerate the data it holds. Evidently in the rush to get the car out of the door in 2012, power management work was thrown overboard (early cars were delivered without it), leaving them with the problem of retrofitting the power management in later software releases. Very likely, some wrong decisions were made as a result: if they were software ones, they can theoretically get fixed but it gets harder and harder as more software is layered on top; if there were mistakes made in the hardware then we are stuck with them at least in existing cars.

System design for low-power applications is tough as there are dependencies between the hardware and software, but fundamentally, the hardware needs to have the hooks for the software to manage power. Early microprocessors did not have snooze or sleep states which consume lower power. These functions were added later as the processing power increased, driving the supply power up with it. Early computers used dynamic RAM which needs to be refreshed to maintain memory contents even when the processor is idle. Static RAM does not require refresh, and as long as power is applied, the clock can stop and memory contains are retained. Flash memory will maintain contents even when the power is removed. Most modern battery powered applications make use of these and other hardware features, like duty cycling the power to "islands" and clock gating, to tame power consumption. The Model S is a mobile battery powered application, and I would expect already incorporates many low-power hooks in the hardware. IMHO, careful attention to (power) detail, a systematic approach to low-power system design in the software can bring significant improvements in vampire power consumption to existing vehicles.

 

[scaesare]

.. the vampire drain when the car is off is ~1.2kWh per day on average.

Nice thorough post.. i have been thinking about vampire draw for a couple weeks and had considered posting a bit about it. Now that we are talking about it's effect ont he 12V battery, I'll wade in.

Do you find the loss to be 1.2kW per 24 hours? I typically will lose 3-4 miles will sitting at work for 9-10 hours or so. While my car is typically plugged in and thus recharges by the time I see it in the morning... I suspect my overall vampire drain may be more like twice your estimate. This is with power-save on, but "Always Connected.

That's somewhere in the neighborhood of 100W of draw on average constantly. By way of contrast, the MAX power draw for my i7 laptop is 65W (which wold include all ports active, max HDD and CPU & GPU usage, full screen brightness, WiFi, BT, and Ethernet ports, sound, etc...). I suspect the average draw is likely a 3rd to 1/2 of that... say 25-35W.

So the car seems to be pulling a good amount of power... even for keeping things like WiFi on. Specifically because the main console is based on a Tegra3 chip, which is a low-power device, and the same one used in one of my older tablets. It has a 5th special "low power" core for background tasks. I'd bet that the computing draw of the console itself is relatively light.

With the LCD screen itself off, I've been considering what additional power draws are...

1) 3G and/or WiFi: This is part of the console computer... I'd be really surprised if this used more than a watt or two.

2) BMS computer: we know the car's "operational control" computer is separate from the console. Assuming the BMS is part of it, I'd bet that's running constantly. I have no idea what it's power draw or capability to throttle down to a low-power state is.

3) Thermal systems: Obviously the coolant pumps, battery warmers, etc... would take quite a bit of juice... but I'm relatively certain that these rarely come on with the car just sitting and not charging, unless it's at a signifcant temperature extreme.

4) Keyfob RF detection: This is obviously on all the time... but I'd expect it to be relatively low power, as it's receive-only.

I'm not sure what other systems would be on much with the car off. The dash is a sperate Tegra2 based computer, but I expect it's not active at all when the car is off.

So I'd guess #2 is the largest culprit...

 

[SteveS0353]

But you can't compare a car to a MacBook Pro. Your Macbook has two processors, some RAM, a motherboard and a screen. The Model S has many, many more interconnected systems and is a far more complicated system. There is n't just one computer in Model S.

Actually, huge motor and the power electronics aside, the low voltage (12V) electronics in the Model S is actually fairly similar to a smart phone architecturally. There are many more processors in the Model S than a smart phone, but like a smart phone, the Model S has several radios. Don't discount the power these radios consume. In common with a smart phone, the Model S has 3G (less power than an LTE radio), WiFi and Bluetooth. The radios the Model S has that the smart phone does not are the key fob transceiver and the TMPS receivers (4 of them).

As a semi-retired medical device engineer (designed battery powered cardiac pacemakers and defibrillators, hearing aids, swallowable cameras), I think I know a thing or two about designing ultra low-power systems. I'm fairly confident that the vampire drain can be reduced by an order or magnitude (10x), certainly with new hardware designed for the application, but even with intelligent software system design on existing hardware, paying careful attention to power-down and power-up sequencing, deeply duty cycling radios and their support hardware, duty cycling CPUs and memory if they are not required when the car is off and so forth. It takes laser focussed attention to detail, chasing and justifying every µA of supply current, but it can be done.

- - - Updated - - -

Nice thorough post.. i have been thinking about vampire draw for a couple weeks and had considered posting a bit about it. Now that we are talking about it's effect ont he 12V battery, I'll wade in.

Do you find the loss to be 1.2kW per 24 hours? I typically will lose 3-4 miles will sitting at work for 9-10 hours or so. While my car is typically plugged in and thus recharges by the time I see it in the morning... I suspect my overall vampire drain may be more like twice your estimate. This is with power-save on, but "Always Connected.

That's somewhere in the neighborhood of 100W of draw on average constantly. By way of contrast, the MAX power draw for my i7 laptop is 65W (which wold include all ports active, max HDD and CPU & GPU usage, full screen brightness, WiFi, BT, and Ethernet ports, sound, etc...). I suspect the average draw is likely a 3rd to 1/2 of that... say 25-35W.

So the car seems to be pulling a good amount of power... even for keeping things like WiFi on. Specifically because the main console is based on a Tegra3 chip, which is a low-power device, and the same one used in one of my older tablets. It has a 5th special "low power" core for background tasks. I'd bet that the computing draw of the console itself is relatively light.

With the LCD screen itself off, I've been considering what additional power draws are...

1) 3G and/or WiFi: This is part of the console computer... I'd be really surprised if this used more than a watt or two.

2) BMS computer: we know the car's "operational control" computer is separate from the console. Assuming the BMS is part of it, I'd bet that's running constantly. I have no idea what it's power draw or capability to throttle down to a low-power state is.

3) Thermal systems: Obviously the coolant pumps, battery warmers, etc... would take quite a bit of juice... but I'm relatively certain that these rarely come on with the car just sitting and not charging, unless it's at a signifcant temperature extreme.

4) Keyfob RF detection: This is obviously on all the time... but I'd expect it to be relatively low power, as it's receive-only.

I'm not sure what other systems would be on much with the car off. The dash is a sperate Tegra2 based computer, but I expect it's not active at all when the car is off.

So I'd guess #2 is the largest culprit...

YMMV, but the 1.2kWh number comes from several sources on this forum, but is just an average number to do some calculations. It will be temperature dependent, and I would not be surprised to see variations over a factor of 2 depending on various factors.

The BMS resides inside the HV battery pack, and certainly does contribute to the vampire drain, and may be a large portion of it depending on the SoC and state of balance of the HV pack. I don't know if the BMS loads the 12V, which is the source of the cycling on the 12V battery and the cause of its wear out.

To reduce power from the little key-fob battery, I believe the key-fob is a polled system. The car transceiver "pings" and then listens for the key-fob. The key-fob listens most of the time (to reduce power), and when it hears a ping from the car, it wakes up and transmits its code as a handshake.

 

[scaesare]

YMMV, but the 1.2kWh number comes from several sources on this forum, but is just an average number to do some calculations. It will be temperature dependent, and I would not be surprised to see variations over a factor of 2 depending on various factors.

The BMS resides inside the HV battery pack, and certainly does contribute to the vampire drain, and may be a large portion of it depending on the SoC and state of balance of the HV pack. I don't know if the BMS loads the 12V, which is the source of the cycling on the 12V battery and the cause of its wear out.

Yeah, I'm sure there is quite a bit of variance... with all of them being "a lot of power" it seems.

To clarify, although there are BMS modules are on the batteries, I think the teardown that wk057 did demonstrates that the individual modules are wired to the data connector on the pack, which ostensibly communicates back to in-car systems. Those systems in turn would also monitor the temps of the chargers, motor/inverter combo, comand the pumps and valves as well as chiller for the heat exchanger, etc....

That's the system I was referring to in my earlier post.

 

[santana338]

There are many more processors in the Model S than a smart phone, but like a smart phone, the Model S has several radios. Don't discount the power these radios consume. In common with a smart phone, the Model S has 3G (less power than an LTE radio), WiFi and Bluetooth. The radios the Model S has that the smart phone does not are the key fob transceiver and the TMPS receivers (4 of them).

I've worked on chips for smartphones and can tell you the hardware is absolutely designed to meet a strict power budget. So unless Tesla is making their own chips (extremely unlikely) the hardware hooks for low power operation are already there. I think this is a matter of the software architecture not taking advantage of the hooks because of the rush to market as has been stated earlier. I do hope we are able to see some improvements in the "classic" cars over time. I'm afraid the Model X and Model 3 are going to prevent that from happening. I just hope the software team has this problem under control for the first Model X.

I have enjoyed this discussion. Thanks to everyone for posting.

I have one more question. When the car is out of warranty and/or Tesla decides to stop replacing 12V batteries out of good will, would it make sense to put a 12V charger in the frunk and just plug it into a 120V AC outlet when parked in the garage? Short of Tesla addressing the vampire drain issue through software changes or the charger change mentioned up-thread, this could be a quick and dirty way to prolong the life of the 12V battery.

 

[SteveS0353]

Yeah, I'm sure there is quite a bit of variance... with all of them being "a lot of power" it seems.

To clarify, although there are BMS modules are on the batteries, I think the teardown that wk057 did demonstrates that the individual modules are wired to the data connector on the pack, which ostensibly communicates back to in-car systems. Those systems in turn would also monitor the temps of the chargers, motor/inverter combo, comand the pumps and valves as well as chiller for the heat exchanger, etc....

That's the system I was referring to in my earlier post.

Indeed, sophisticated HV battery management will take power, and all of the systems in the Model S are interconnected and interdependent. It would be terrific to see an ordered list of power consumption by module, but AFAIK, that's not in the public domain.

 

[Moonwick]

Do we know if anyone's slapped a logging ammeter onto the 12v battery leads and let the car sit for a day or two to confirm that this is what's happening? It's hard to imagine anything else going on, but the theory that the 12v battery is getting cycled 2-4 (or more) times per day is seeming pretty likely.

If that's really what's happening, then engineers at Tesla had to be aware that they were using the 12v battery in a way that couldn't be expected to last more than a couple of years. They've been great about keeping up with replacements (even being proactive in many cases) but it's obviously not sustainable to keep doing this for much longer. I've got to assume that fixing the vampire draw (or at least finding a more durable way to power it) is a pretty high priority over at engineering HQ.

 

[SteveS0353]

I have enjoyed this discussion. Thanks to everyone for posting.

I have one more question. When the car is out of warranty and/or Tesla decides to stop replacing 12V batteries out of good will, would it make sense to put a 12V charger in the frunk and just plug it into a 120V AC outlet when parked in the garage? Short of Tesla addressing the vampire drain issue through software changes or the charger change mentioned up-thread, this could be a quick and dirty way to prolong the life of the 12V battery.

It's been good discussion, and something about which I'm rather passionate. My whole life's work has been designing systems to consume the lowest power possible. I absolutely love my Model S, but the vampire drain is a niggling issue for me.

Your suggestion for a wall powered charger to tame the vampire (similar to Cottonwood's suggestion up thread), I think would work. By definition, an automotive battery charger will provide voltage within the range of the 12V battery while it's charging from the DC-DC converter, so I don't expect would present any danger to the other electronics connected to the 12V bus.

 

[rpo]

Do we know if anyone's slapped a logging ammeter onto the 12v battery leads and let the car sit for a day or two to confirm that this is what's happening? It's hard to imagine anything else going on, but the theory that the 12v battery is getting cycled 2-4 (or more) times per day is seeming pretty likely.

If that's really what's happening, then engineers at Tesla had to be aware that they were using the 12v battery in a way that couldn't be expected to last more than a couple of years. They've been great about keeping up with replacements (even being proactive in many cases) but it's obviously not sustainable to keep doing this for much longer. I've got to assume that fixing the vampire draw (or at least finding a more durable way to power it) is a pretty high priority over at engineering HQ.

I second this. It would be great to have actual data on the power draw. It could be monitored while driving and while sitting with the various power saving features on and off.

 

[SteveS0353]

Do we know if anyone's slapped a logging ammeter onto the 12v battery leads and let the car sit for a day or two to confirm that this is what's happening? It's hard to imagine anything else going on, but the theory that the 12v battery is getting cycled 2-4 (or more) times per day is seeming pretty likely.

If that's really what's happening, then engineers at Tesla had to be aware that they were using the 12v battery in a way that couldn't be expected to last more than a couple of years. They've been great about keeping up with replacements (even being proactive in many cases) but it's obviously not sustainable to keep doing this for much longer. I've got to assume that fixing the vampire draw (or at least finding a more durable way to power it) is a pretty high priority over at engineering HQ.

Hans posted this in another thread -> #618 Waking and recharging every 3 hours?

 

[Moonwick]

Hans posted this in another thread -> #618

Huh, that seems like pretty strong evidence. If they're really using the 12v battery like this it's pretty stunning that they started out shipping cars with standard (instead of deep-cycle) batteries; it should have been pretty obvious from the get-go that a normal battery wouldn't have been suitable for this sort of treatment. It's arguable that even a deep-cycle battery isn't suitable for it, either.

 

[santana338]

I just ordered this voltage data logger to see when the 12V battery is getting charged. Then I will put my 12V trickle charger on it and see what differences I see

 

[Great Dane]

Do not just slap any charger on the 12 volt battery, it has to be highly regulated power supply. I have seen the battery at 13 volt for hours at a time
while working on the car, installing lighted T, so if we put a regulated 13 volt - 13.8 volt ( float charge voltage ) on the battery it would stop the
dc to dc inverter from kicking in and prolong the 12 volt battery life, preventing cycling while in sleep mode at home.

 

[santana338]

I bought this Yasua float charger last year and it is the one I was thinking about using. I was going to monitor the car first, then the charger on my mower/snow blower, and if they looked compatible, I would try it on the car.