What would a battery swap cost?

[Raphie]

Yup. It’s fully up to Porsche what they want to monetize and what will remain free.
the current renewal prices already give you a heads-up of what’s coming.
smart planning charging gone, entertainment? who knows. Several icons, working now will be greyed out after the initial term. Some getting a free new certificate others being paywalled.

Sponsored

 

[WasserGKuehlt]

What we do in the controllers is put the FW hash residual in an immutable register that anyone can see. That is how the version and signing can be verified by a central authority. Is that what you mean by key id?

How would an update work? (if the register is immutable, I mean) That sounds like a fine mechanism for ensuring the integrity of the component itself, but doesn't seem to address its provenance (or I'm not fully grasping the solution).

In my speculation, which may be completely wrong:
- each module has a TPM of sorts, and can do some crypto such as signing a message with a given private key it stores
- the car's TA has its own TPM/HSM, and can generate keys (the HSM is write-only, and would allow a trusted entity to invoke a decrypt/sign operation with a given key, resolved by an identifier)
- on pairing a module with the car, the TA issues a challenge authentication to the module, which responds by signing it with its endorsement key. (The module's response may include any other info in the measurement, such as hw/fw/sw Merkle hash.) If the TA accepts it, it generates a 'utility' key for that module, which the module would then store and use for subsequent communication within the car. (Here there would be different solutions, depending on whether the utility key is symmetrical or not, and whether the car really uses an HSM or can actually export private key material.)

As you said earlier, it'll have to be a flat hierarchy, but you don't want the TA to be a single point of failure, either. It'd be fascinating to know more details about how exactly car manufacturers are approaching security - in the realm of services, it's very easy to fail if confidentiality or integrity can't be guaranteed, but that's a luxury the physical world doesn't have. Failing has to be safe, and the differentiation between minimum-required functionality and what should fail/stop working following a breach must be crisp and consistent.

 

[whitex]

Yup. It’s fully up to Porsche what they want to monetize and what will remain free.
the current renewal prices already give you a heads-up of what’s coming.
smart planning charging gone, entertainment? who knows. Several icons, working now will be greyed out after the initial term. Some getting a free new certificate others being paywalled.

I get the subscription stuff, even if the subscription is free for now. I'm talking about non-subscription scenarios, say you paid for Innodrive or surround view at build time, if you don't pay for connectivity, it all goes away?

 

[whitex]

How would an update work? (if the register is immutable, I mean) That sounds like a fine mechanism for ensuring the integrity of the component itself, but doesn't seem to address its provenance (or I'm not fully grasping the solution).

Typical ways to authenticate content such as firmware using crypto is to burn a public key, or its hash, into a device (immutable for life of the device, such as fuses). Then, each firmware comes with a signature which is checked against that public key (or, if only the hash is burned in, the public key comes along with the firmware image too, it's checked against the hash, and then, if that passes, the firmware signature is checked against the public key).

More advanced devices, such as Tesla AP chips, use PUF (Physically Unclonable Function) technology instead of fusing public key or hashes.

 

[WasserGKuehlt]

Typical ways to authenticate content such as firmware using crypto is to burn a public key, or its hash, into a device (immutable for life of the device, such as fuses). Then, each firmware comes with a signature which is checked against that public key (or, if only the hash is burned in, the public key comes along with the firmware image too, it's checked against the hash, and then, if that passes, the firmware signature is checked against the public key).

The way @OTPSkipper described it (or how I understood it), it was the hash of the code/FW being burnt in. The way you describe it sounds more like the endorsement key - in that case you'd have to trust _any_ change, so long as it's signed with the expected key. I don't see how you can ship a change that matches a predetermined hash. Either way, rolling the key is effectively bricking the component, and that's what a hierarchy of keys is supposed to address.

More advanced devices, such as Tesla AP chips, use PUF (Physically Unclonable Function) technology instead of fusing public key or hashes.

Very cool, and funny you should mention that; I was thinking of PUFs earlier, as a way to deter/avoid 3P devices. Those really are unique 'keys', and servicing would most certainly require not only live manufacturer authorization, but really only OE parts, known to/inventoried by the car's manufacturer. I guess that may be ok for central systems, not so much for 'peripherals'.

 

[OTPSkipper]

Typical ways to authenticate content such as firmware using crypto is to burn a public key, or its hash, into a device (immutable for life of the device, such as fuses). Then, each firmware comes with a signature which is checked against that public key (or, if only the hash is burned in, the public key comes along with the firmware image too, it's checked against the hash, and then, if that passes, the firmware signature is checked against the public key).

I don’t know that I would call what gets put in Otp (one time programmable ) memory as “public”. When that secret is known, you can create new FW that the chip will run. It’s different than a public/private key sceeem that is used to pass a session key. It’s more like the session key. It is the shared secret between the FW signer and the FW user.

 

[OTPSkipper]

How would an update work? (if the register is immutable, I mean) That sounds like a fine mechanism for ensuring the integrity of the component itself, but doesn't seem to address its provenance (or I'm not fully grasping the solution).

Download a new image into flash. Then reset the chip (usually a FW command). Then the boot rom reads the image and puts it in sram. Then it copies the key from Otp to the crypto HW and start the crypto HW. If the image Checks out, it jumps to it. If not, it repeats using the old image in flash.

when an image checks out the crypto hash residual goes in the FW status register and is locked down so it can’t change till the next reset.

 

[whitex]

I don’t know that I would call what gets put in Otp (one time programmable ) memory as “public”. When that secret is known, you can create new FW that the chip will run. It’s different than a public/private key sceeem that is used to pass a session key. It’s more like the session key. It is the shared secret between the FW signer and the FW user.

In the approach I described it is in fact a public key (or more commonly its hash since it’s smaller) in OTP and it is not a secret at all. A corresponding asymmetric private key is used to sign the firmware - that one is a secret.

 

[maximumpanda]

I can’t help but feel like most of you are overthinking it (at least in relation to the battery).

the easier workaround is that at some point there is an analog, physical connection. as long as the new battery outputs the same current and charges from the same current (can controlled / managed by a secondary BMU) the taycan has no way of knowing that it’s no longer using The original battery.

You might need something that sits between the two to do some translation (for instance reporting battery health from the new batteries relative to parameters of the old ones, or if new batteries have better charging characteristics, tricking the BMU into allowing more juice to flow).

absolutely worst case, there are umpteen bypasses which could be devised once you have physical access to the system. Most of these security measures are related to avoiding a software hack remotely, all of that goes out the window once you have prolonged physical access to the device.

 

[ben1]

I can’t help but feel like most of you are overthinking it (at least in relation to the battery).

the easier workaround is that at some point there is an analog, physical connection. as long as the new battery outputs the same current and charges from the same current

Indeed. If you have broken cells in a module, you open the module and replace the broken cells by similar lithium-ion cells.
As long as there is nothing that registers that you opened the module, it should work.

I regularly read horror stories of prices of battery replacements of 50000 to 75000 $/euro. But in reality, for the real examples, from real people, really having their battery replaced, the prices I heard were more like 10000 to 18000 euro.
I'm taking those stories with the very high prices with a grain of salt. Sounds a lot like spreading FUD about EV's.

 

[tchavei]

Well porsche quoted 30k for a Taycan battery so directly from the horse's mouth but like I said previously, someone gonna figure out in a few years (if Porsche doesn't step forward) how to replace the cells with the latest solid state or whatever technology that is around in 5 years and just make it work.

There's definitely a market for lithium "upgrades". It's been around for phones for years now. Cars will follow soon.

 

[simcity]

What do we think a swap at say 120K miles would cost? My thought is how this would compare to ICE maintenance. In my experience, putting that kind of mileage on a car always leads to some major maintenance bills. How would the battery swap compare to all the thing I had done to keep my 140k miles boxster on the road?

The 800V battery should or ought to be totally fine for that mileage.

The battery heath on my 80,000 mile 5 year old Model X show 6.4% degradation. It will probably level off around 10%.

What I would budget for is replacement of the 12V LiFePO4 battery which is IMHO ridiculously priced and seems far more inclined for failure, at around $1000 a go.

 

[whitex]

The 800V battery should or ought to be totally fine for that mileage.

The battery heath on my 80,000 mile 5 year old Model X show 6.4% degradation. It will probably level off around 10%.

What I would budget for is replacement of the 12V LiFePO4 battery which is IMHO ridiculously priced and seems far more inclined for failure, at around $1000 a go.

Don't forget the tiny lithium ion battery inside the "conbox high" module (connectivity), not sure what the cost is I did confirm with service techs it needs replacement every few years (they have them on other cars, not just Taycans).

 

[simcity]

Don't forget the tiny lithium ion battery inside the "conbox high" module (connectivity), not sure what the cost is I did confirm with service techs it needs replacement every few years (they have them on other cars, not just Taycans).

Is that some sort of keep-alive button lithium cell or small cylindrical Li-SoCl2 battery?

 

[OTPSkipper]

In the approach I described it is in fact a public key (or more commonly its hash since it’s smaller) in OTP and it is not a secret at all. A corresponding asymmetric private key is used to sign the firmware - that one is a secret.

I’m going to check this out with our security expert. This sounds like an older approach , but let me check it out.

Sponsored