> Tesla offers a “Root access program” on their bug bounty program. Researchers who find at least one valid “rooting” vulnerability will receive a permanent SSH certificate for their own car, allowing them to log in as root and continue their research further.
Pretty interesting. Sounds like Apple's Security Research Device Program[0], where you're loaned a rooted iPhone, but with a clear qualification criteria.
It strikes a nice balance, because to qualify you have to 1) show you have the skills to get root access anyway and 2) show you're willing to participate in the bug bounty program and get things patched.
I would of course love root on everything I own, but I can understand Tesla's motivation here since root for everyone would make vulnerability discovery easier for malicious actors. And if everyone had root on their Tesla, it'd be much easier to make naughty modifications that might catch the ire of regulators. (like disabling driver attentiveness checks in self-driving mode).
The underlying tension is that "you own the car" means something very different from "you own the software running the car." Tesla treats the firmware as licensed software rather than property you can inspect and modify. The bug bounty program is a PR-friendly way to say "we support security research" while keeping full control over who gets access and under what terms.
Right-to-repair legislation is chipping away at this but slowly. The EU's right-to-repair directive covers physical repair and doesn't really touch software access. The real test would be a regulator taking the position that restricting root access on hardware you own constitutes an anticompetitive tying arrangement, since you can't use the car's data for your own purposes without going through Tesla's APIs and paying their fees.
John Deere has been the main battleground for this argument so far. Farmers can't repair their own tractors without paying for dealer access to diagnostic software. Tesla is the same pattern applied to consumer vehicles, but the consumer advocacy pressure is weaker because fewer people feel the pain directly.
>> Tesla is the same pattern applied to consumer vehicles
No i'd push back on this because the entire workshop manual is available for free without even registration required. You can literally google and land in the relevant sections and it is of a far higher quality than ford, VAG or bmw as three examples i'm pretty familiar with. I haven't seen the John Deere stuff.
Tesla does have "special tools" for some repair procedures, a practice as old as the auto industry but they don't rely on them to the same extent as BMW for example. Anecdotally, the special tools i'm aware of are genuinely useful - for example, the tool for disconnecting seatbelt anchors saves time vs the traditional bolt - where special tools on other marques are often clearly to workaround a failure of packaging or engineering resulting in tight access for a regular tool.
Their online API access is a little bit annoying, or at least unfriendly to casual home user, specifically the workflow to register an OIDC client, but not insurmountable.
> "Tesla is the same pattern applied to consumer vehicles"
It really isn't. Unlike John Deere, Tesla is actually pretty good on right-to-repair. All of their technical and repair manuals are available for free to anyone. The service/diagnostics software ("Toolbox") is also available to anyone, albeit for a (not entirely unreasonable) fee.
(There is also a service mode built in to the car which can do many basic diagnostics for free)
> All of their technical and repair manuals are available for free to anyone.
That should be the bare minimum. Ford charges you 40 dollar an hour for it and unless you know exactly what you are looking for you will spend several hundreds on it.
Too bad ford killed their old site, the print form was unauthenticated and you could print the entire schematics to pdf if you knew the internal model number. Or do what I did and run a script to dump it to higher res PNGs.
charm.li covers Fords and many other makes too up to 2013 ish. It is a pirate archive site holding workshop manuals for thousands of cars. Very useful. Very free. Long may it stay hidden.
More legitimately, alldata.com has repair data, workshop manuals for most marques up to today and will sell you either single vehicle (called "DIY") or a package aimed at independent mechanics where you can access anything. Same manuals either way, but you pay per vehicle with DIY (and have to contact support to switch.)
ETIS is dead and Ford finally pulled the plug, though since the current backend is some semi-custom IBM bloat I would not be surprized if you could get by that without too much hassle (took them three years to find out I was downloading all my car's travel and charging logs before they banned the dummy account, but now they track it and discontinued most of it anyways).
I won't go into details but searching around with the "forum" keyword and etis might get you somewhere (at least that did the trick a few years ago, now with LLM slop I don't know, and what the other person posted).
I would love to lobby to change how the law works for these cases: for some definition of "firmware" (informally "software that ships with hardware and is not intended to be selected by the consumer like a computer operating system"), add a copyright exception so that modifying the firmware in situ is treated like modifying the physical hardware, because in practice they are in fact the same thing: a single component that does a single thing.
With this, the John Deere approach to gatekeeping vehicle repair would no longer be legally protected by the DMCA or by copyright law. All the other protections afforded by copyright law would still apply: you cannot rip the firmware off the hardware and distribute it, the manufacturer is under no obligation to help you modify it, etc.
However, tools which patch or circumvent antifeatures of the firmware would now be legal to use on hardware you own: it would be legal to patch out software locks, retune engine computers, etc.
>The underlying tension is that "you own the car" means something very different from "you own the software running the car."
What does that mean? "The software" is a specific configuration of the hardware you own. How can you own the hardware and not the specific copy of whatever data is on it? Note that I'm not confusing the copy of the data with the IP rights to it.
Because American courts have entertained utterly moronic claims for decades now and the DMCA eliminates any sanity in consumer rights around IP products.
When you bought a DVD, you didn't "Own" the movie, but you had a legal right to do things with that data you didn't "own" anyway, like format shifting and selling that physical object on to another person. You could copy that data off and do things with it. I think technically it would be a copyright violation to then put that movie file into Movie Maker and cut up your own personal highlight reel, but good luck finding a judge willing to hear that case if you don't upload it to youtube.
Now, thanks to the DMCA and courts being absurdly credulous of bullshit arguments from corporate attorneys, you no longer have basic consumer rights. If you try to even inspect the code that runs to protect your literal life, that can be a crime. You own the literal hardware, but if you try to act like you own it, that's a crime. You technically still have the right to format shift a BluRay for example, but bypassing the math protecting that data overrides that "right" and you are guilty of a crime. A CEO's wet dream.
If the DMCA was older, IBM could have prevented the existence of the Clone PC market and ensured a locked up market. We would all be stuck on absurdly shit hardware because that's what was more profitable for IBM.
Pre-DMCA, Sega was told that their trademark rights were overridden by the innate market right to interoperate with their product. IP rights used to be fairly weak! Sony could not prevent a company from selling a software product that ran playstation games. To this day, Nintendo simply pretends these court cases didn't happen.
This is part of why China has so much success in manufacturing and product development IMO. They don't need to develop purposely worse versions of things just so some other company can sit on their hands for 20 years collecting rent. If you want a fast moving market, the ability to lock things down for 20 years is fundamentally unacceptable, only enriching a few owners, and outright harming our country. Basically every time in history that IP rights are weakened or nullified, you see a burst of development and advancement in products and solutions.
If you're going to sell the car with the modified firmware, fine.
But at least in my jurisdiction, I can mechanically modify the car in any way I please, as long as it still has seat belts, brake lights, and bumpers of a certain height. It doesn't even still require a steering wheel; that's not specified in the law as far as I've been able to find. (Now, if I removed the muffler and made it louder than proscribed by law, I could be cited for a noise violation, but only at such a time as I womped on the gas and actually made the noise. The car itself being _capable_ of the noise is not, inherently, illegal.)
This blew my coworkers' mind once as I unplugged the passenger-side airbag while mounting a bunch of new stuff there. Apparently in some places, it requires paperwork and certifications just to unplug a connector? Weird.
Tesla absolutely does not apply the same patterns as John Deere.
Everyone can fix Teslas. Parts are easy to obtain. Never had issues with them.
John Deere on the otherhand is the absolute evil of right to repair.
> The underlying tension is that "you own the car" means something very different from "you own the software running the car."
How is this different from the 2000s, or the 90s, or even before, when the normal thing to do with commercial software was to purchase a license to use said software and a physical medium containing a copy? You'd also then not "own the software", but you owned the right to install a copy on your own computer and use it. That worked without having to hand over the keys to your own computer.
Sure, the physical delivery medium is gone, but that's just a detail. Why do we now think that just because we license software for use, we can't be in ultimate charge of our own devices?
In 1990 Ford couldn't turn off your Mustang because you plugged a TwEECer into the J3 port and screwed around with the tune. Best they could do was void your warranty and deny you further upgrades (i.e. tunes flashed as part of a recall or TSB).
These days unauthorized access tends to lose you effective use of the hardware you bought because the hardware requires software features to work and that software often unnecessarily phones home so if the OEM toggles a field in a DB somewhere you lose access to back up assist or whatever other fancy tech features that you a) paid for b) don't strictly need to have dependencies that phone home to work but do "because reasons".
It's not whataboutism, it's a legitimate question. How does it increase safety on the road to reject local SSH connections by a dumb user, when that same user can mess with the car physically?
Simplest example: a driver could probably disable attentive driving checks by pasting a script in from a web search in a few minutes. Nothing like an inattentive 3750 lbs weapon.
Yeah and they could hire a professional driver or a engineer and IPO for billions a life sized driving AI powered crypto robot too. Look, like clearly google + ctrl-v scripting or running an one click deployment exe on your computer on a whim is different than physically ordering/picking up something and then installing it into a vehicle?
Of course they're different, but you're trying to argue that the former takes objectively less effort than the latter, and it doesn't. One or the other may take less effort depending on who you are and what you know.
In most cases I agree with this, but maybe not for potentially dangerous things like cars? What if someone roots into their car and disables some essential safety feature - maybe even a legally mandated safety feature?
More concretely, the expertise-required-to-access-root is in a different field to the expertise-required-to-make-wise-changes. i.e. you might know how to hack a car, but that doesn't mean you know how cars operate.
Up until v recently cars were not remotely accessible and part of a command-and-control network which Teslas are (perhaps other modern cars are too, I only know Tesla because I have one).
I know that the car reports practically all user events to Tesla in real time over the cell network (eg, open door), and I know it has root access. I don't know if that root is available remotely and I don't know if foundational commands like steering, acceleration and brake are accessible via the CLI (they are computer controlled actions locally)
THUS I would not want to drive a Tesla if there was the possibility of all cars being rooted and remotely controlled by an unauthorized actor.
Given electric cars are responsible for much bigger responsibilities than combustion cars (avoid driving into that bicyclist), there are new concerns here which beg extra consideration.
I actually think we should be asking more of safety regulations here with regards to the design of electric/computerized cars.
Think of it this way: every concern you have about a teenager having root on their electric car is the same as any sociopath hacker (AI enabled for modern nightmare fuel) who finds a root vulnerability and decides to not be a good person with it. If a teenager can mess with the collision avoidance, e.g. Israel can modify it to murder anyone who talks shit about Israel in the car. Or the CIA could turn it into a weapon. Or one day some dev could push a bad OTA update. Et cetera. Our safety regulations should mandate design features to prevent a malfunctioning computer from posing any greater safety risk than any other modified part in the car.
Then why wasn't it a problem before? People have always been able to install aftermarket or possibly even hacked together physical parts. If there was liability you'd expect some sort of shield blocking access to, for example, the hydraulic system for the brakes.
As it turns out though blatant irresponsibility is quite rare (depending on your definition anyway) since people have a strong self interest in not endangering their own lives or wallets. It's similar for homeowners - many states explicitly carve out a requirement that insurance companies cover DIY modifications that are within reason and this generally works out since you have a strong vested interest in not destroying your own house regardless of any insurance policy.
It is. Thousands of people have died because of aftermarket headlights. Harder to assess, but probably much larger, is the number of excess deaths from nitrous oxide etc. emitted by modified cars.
There are about 3000 deaths per year in Sweden attributed to position from cars, and 300 physical accidents. So it is a really big issue, but it is almost impossible to make people understand that their car use and modification mains people.
Modified cars can release 1000x more polution, on streets with 800 daily cars that will have an affect.
You can ban modifying your car to pollute more (which we do) without banning modifying your car.
This isn't complicated FFS.
The difficulty against this in the US is the unfortunate reality that the people coming to these shops to enable their stupid trucks to roll coal are the people who should technically be raiding and shutting down these companies. This can be fixed.
Physically, you can already modify your car to be controlled by a stupid program and that has been possible since at least the 90s. You can do the supposed harm by not being aware of damage to your exhaust system.
The solution to exhaust harms of ICE engines is electric cars, not a reduction in consumer rights.
People get killed by changes to exhaust, height (lift kits), bumpers (bull bars in particular), etc pretty often, though. And I can imagine software changes (exhaust is part of that actually) could kill people too.
Maybe you think daytime running lights are stupid and want to disable them for instance.
Sure. Point is nothing has really changed. Largely there's no problem and to the extent that bad things happen it isn't something novel that's only just come up. It's not in and of itself an excuse to erode private ownership. If intervention is required then regulation should be passed deliberately by the legislature.
Well both of those examples could potentially electrocute you or start a fire and both can be done by a homeowner if he feels like it.
I don't disagree that it's a bit different in certain ways but I feel like that's drifting off topic. It shouldn't be up to manufactures to determine these things unilaterally but rather the legislature. Particularly any justification to the contrary rings hollow in this case because there's a very strong conflict of interest.
I don’t think that’s the reason, seeing as a car is already endangering everyone around it by existing. More likely about keeping the tooling to diagnose issues proprietary and expensive.
That kind of thing is always the stated justification but never the real reason.
Almost invariably when that excuse is trotted out, there are are usually many things that are much more common that are also far more dangerous. For example, texting while driving or driving with bald tires in the wet are both 100x more dangerous than anything almost anybody would do by modifying the car's software.
Four 9/11s worth of people die every year from drunk driving. If we can't even get that under control, I don't see why being able to modify your own car is a big deal.
Enforcement is abysmal for stupid reasons. Courts are reluctant to remove the ability for people to drive because America purposely made itself dependent on cars, and cops are reluctant to actually arrest a lot of people for drunk driving because they tend to be buddies, or worse. You can find plentiful examples of off duty officers trying to get out of drunk driving simply by being a cop.
This is what you get when you can vote on the sheriff and judges who insist they are "Tough on crime" because they sentence a dude smoking a joint to years in the joint while ignoring real problems like, you know, murder and theft and violence and all the shit their buddies are doing. The "Tough on crime" people are the ones drunk driving often enough.
Even as a well trained software engineer who works on transportation software including ECUs (heavy equipment not cars), I'm not sure there is much I could do with root. IF I had full source code to my car's radio I might try to add android-auto back in (it has android-automotive so I know it can do it), but if that isn't easy I'd probably give up. Without source code and a lot of time doing anything is impossible - as anyone else who works on complex software knows.
You can feel that way, but plenty of car configuration has always been locked away and walled off, and manufacturers make a tidy profit selling software licenses to dealers and mechanics to perform basic diagnostics. Proprietary software is big business what can you do.
Definitely not always. It used to be that a mechanic or a skilled owner could tune, modify, repair or replace absolutely anything in your car. That was basically since the invention of the car, up to somewhere in the 2000s. And even then, various hackers and pirates made sure almost anyone could get their hands on the software. In fact, many mechanics these days use 3rd party software because the manufacturer refuses to sell them their version or even that version doesn't have all the features.
That is the recent (and gradually worsening) situation but it is not in and of itself a justification. Effectively you're saying "it's currently this way therefore it's okay for it to be this way".
Manufacturers have increasingly restricted control over products as they've gradually been digitized. Prior to the digital era anyone could do anything to personal property (regulations notwithstanding ofc); more expensive items typically came with circuit diagrams for the purpose of repairing them.
Having shell is extremely handy for further discovery. SO handy that if they were just gonna patch the bug and lock you out, you would simply not disclose it.
This is what happened. Tesla security received tons of bug reports that required root access to identify, yet they got a vanishingly small number of root vulnerability reports. This policy fixes that misincentive.
That’s quite a weak confidence in their own platform security if finding a root level vulnerability is not one-off event, but it’s a program expected to have multiple people routinely finding those.
My read of the output in the post when they tried to SSH to the device was that Tesla are actually doing the right thing here and using an SSH certificate authority, which allows issuing certificates signed with a private key authorising access to a subset of devices (optionally for a defined period of time). https://en.wikibooks.org/wiki/OpenSSH/Cookbook/Certificate-b... has more information, but in summary unless the private signing key is compromised in some way this is entirely legit. I'd hope that they also have some mechanism for distributing a new public key if the signing key does get compromised but who knows.
And as we all know, if you're smart enough to get root access, your neighbours children playing football in the street should be subject to the risk of you driven a car that claims to have full self driving with custom code on it.
I used to work for a company that made third party scan tools. We had racks of ecus disconnected from the car with just a diagnostic connector and power. nothing got to a real car without first trying it on the rack. I remember on time we figured out a bmw (pre obdii) had the bytes offset from the standard documentation (it was a semi-standard protocol that some other cars used at the time), we went from we communicate but nothing is wrong to a very long list of dtcs on that controller. (All our competitors also showed nothing wrong, but the official bmw tool showed dtcs)
That's super cool, I'm currently struggling with scan tools for a 1999 Mercedes E300 Turbodiesel. I had one that worked OK for about a decade (Autel something or other) with a 38pin connector, but it recently bricked itself with a message like "connect via USB to Updata" which I assume means its firmware somehow erased itself. Cannot figure out how to "updata" it, doesn't seem to connect via USB, the Autel software runs under Wine but doesn't appear to recognize the device... gave up and bought an iCarsoft device which sorta kinda works. It can talk to every module except for the ECU (Bosch MSA 25.1 I believe?) however if I tell the device that my car is a different model (1995-1997 naturally aspirated) I can blindly clear ECU DTCs, which is good enough because this thing is barely more complicated than a toaster. All that is to say, this space is ripe for some open hardware/software love.
Interesting...1999 is probably a bit early for that Bosch to be running one of the usual ECU update protocols like UDS. It sounds like it's in the bootloader and looking for a valid executable. So the FW updater is likely in the bootloader.
If you can open it up and find the JTAG pads, it should be simple-ish to use a JTAG reader to dump the image and then you can figure out the update protocol from that. It's unlikely to be complicated.
Not sure about your specific car, but a lot of the “consumer friendly” options like OBDeleven, Carly, etc are fantastic. You often have to pay, but a lot of work goes into them and they often just work.
> All that is to say, this space is ripe for some open hardware/software love.
There's just so many computers and what-not in modern cars that this is a very tall ask. You'd need a project on-par with HomeAssistant to get anywhere.
Yeah, it seems like more modern technology has settled on standard protocols (maybe a naive impression--someone will shout at me if that's the case) but there's probably a very long tail of bizarre false starts if you want full coverage of models back to the early 90s when computers became more commonplace.
After 2006/2007 nearly everyone did CAN. I think that is even mandatory in the US, though I have no clue how to look that up (I assume there are details and exceptions) However before then everyone did their own thing. Often with custom chips that haven't been made since 2004 (or even 1999): good luck finding one that works if it breaks. CAN is cheap and allows a lot of power while hiding most of the protocol complexity. The things before that were often not as powerful as CAN, while being in practice a lot more complex because the complexity wasn't hidden.
I remember getting that era working. I concluded Mercedes was trying to be clever in making a protocol so complex nobody else could understand it (thus ensuring you had to use a dealer) - and then discovered they couldn't debug it.
each body model (nothing to do with year or style) was different so clearing dtc but nothing else is not a surprise.
i did get that working, but I last touched it in 2007 so I don't remember enough details to be helpful. good luck.
You don't know anything about late-90s Lucas/SAGEM GEMS ECUs do you, or Range Rover BeCMs?
I'm currently picking apart the firmware in those because it is now impossible to get replacement ignition key fobs, and it just can't be that difficult...
It was 16 years ago, and I only worked with what got to the US. I don't remember much and not those at all. I saw a few how to program key fob documents but we decided that was a dealer service and so I never implemented it. still generally just send the right 4-8 bytes and press a button on the fob in a minute. In any case it sounds like you want a different end: making a fob or bypassing them was never something I got anything on.
I spent the last week successfully reverse engineering my car / various scan tools to get the right information to diagnose a fuel pump problem (and to do so without the incredible awkwardness of many of the tools)
It's pretty amazing what Claude + Ghidra + knowledgable coaching can accomplish. It was basically just setting direction, setting up an incremental workflow with the right kind of documentation, and questioning some of its theories and assumptions from time to time.
I'd love to release a lot of it but I'm torn between releasing artifacts created with expensive software I paid for and thinking that many of those things should really be freely available to anyone (specifically the things which definte the protocol to talk to the car and mapping of what various things are reported vs what they actually mean.
> I'd love to release a lot of it but I'm torn between releasing artifacts created with expensive software I paid for and thinking that many of those things should really be freely available to anyone
Release it or not, but either way you’re almost certainly going to get paid back the same amount of money: $0.
I’ve recently built a disassembler and emulator using Claude to help reverse engineer a 90’s ECU based on an Intel embedded cpu.
It was quite impressive to watch when Claude started to use the emulator to help understand how bits of the code worked.
If you release it people expect you to support it an answer questions. Some of them are not even nice about it. It pays to release this only if there is a group of people who will be constructive in helping make it better, otherwise it is a thankless effort.
Sorry, what are you talking about? Just release it? Are you talking about trying to make money off it? Are you claiming you reverse engineered ecu tuning software you paid for?
You really must be new to this, huh?
Expensive software that you paid for?! Claude?
Yes, the question is whether you want to share knowledge that cost you literally nothing, and will bring humanity one microscopic step in a better direction - or not, feeling superior in that only you have access to that knowledge.
You have a choice!
It's funny to hear LVDS be described as an "automotive" cable when all of my run-ins with it are for connecting laptop displays to their main-boards! (though that has a very different connector on it, and its a very general term for the signalling protocol from what I remember)
Not saying there's anything wrong with your perspective (lots of terms get in muddied waters, it's common and not a problem if everyone is on the same page), but this is what I just found on Wikipedia:
"Early on, the notebook computer and LCD vendors commonly used the term LVDS instead of FPD-Link when referring to their protocol, and the term LVDS has mistakenly become synonymous with Flat Panel Display Link in the video-display engineering vocabulary."
The cable in the article is pretty much doing the same conflation of terms that Wiki is talking about - the automotive one is a proprietary cable that carries some protocol that uses LVDS as its signalling, so at the most basic level both it and the display cable in the laptop are 'LVDS cables' but that's also the most generic term that gives you no information about the protocol actually being carried by the cables.
Yeah I saw that too which is why I posted my comment, it's surprising to me :) LVDS for display cables was an incredibly term in that context. Even still is sometimes despite them mostly being eDP (embedded-DisplayPort) now, which is quite incorrect hah
And eDP is a differential signal at 200 or 400 millivolts so I don't see how that's "quite incorrect". It's not "the" LVDS but it's still in the category.
Most modern laptops no longer use LVDS for connecting the screen, but they use eDP (embedded DisplayPort).
So LVDS is more likely to linger in automotive displays, while in less obsolete devices it has been replaced by either eDP or by MIPI DSI (used e.g. in smartphones).
Very cool. Over a year and a half ago I installed a towing brake controller in my Tesla Model Y. Found the location of the plug, how to access and the pinout online (confirmed via a voltmeter..) so the car's side felt straight forward. But then I needed to find a brake controller that can work with the higher voltage (14.4v vs the normal 12v). Then built a cable from the brake controller to the connector that plugs into the car that I found on eBay. I velcro'd the controller under the dashboard. It works pretty well. I towed my small camper several times with it last year with no issues. Yay! However my little project is nothing compared to this post. Love people hacking away. So cool.
>then I needed to find a brake controller that can work with the higher voltage (14.4v vs the normal 12v)
Put a voltmeter on the battery terminals of a regular car at 2000rpm and note the voltage. You'd be surpised (the alternator can produce as high as 15V on some cars).
Automotive transients can be wild. I did a bringup with a board that had specified 100+v range specified for transients and finicky quality requirements on the output. The power supplies took up most of the (very large) board.
14v is not a transient, if your voltage was 12v with the car running, there's something wrong with the charging system (DC-to-DC in an EV, alternator/generator in an ICE)
13-14v is normal in all 12v automotive systems as the charging voltage
If I recall correctly, a fully charged lead acid battery has an open circuit voltage of 13.6V.
So the alternator has to put out at least something higher than if it’s planning on recharging the battery after 500 to 700 amps have been pulled from it for a few seconds to start the engine.
Yeah, max CV charging voltage is ~14V, max charging C rate is ~0.2C, open circuit voltage at 100% is that 13.x range. And lead acids like to stay at 100% unlike Li-ion which likes 50% +/-30%, so "12V" ICE cars just use a bus voltage of 13-14V and wire the battery there. At any given moment, the car's "12V" bus voltage MUST be above 13.x and below 14.4(absolute max).
It's a bit perplexing that those lead acid systems are referred to as "12V" systems when that figure is effectively the 0% voltage, whereas 3.7V for single Li-ion cell is the 50% voltage.
e: also, ICE transients can be in kV range, coming from ignition mechanisms. I've heard that you can literally measure engine RPM by selecting 1/dt on an oscilloscope and dividing that by cylinder count.
The nominal range for automotive systems is 10-16v. If you are designing anything for automotive use that doesn’t work reliably in that range, you are manufacturing problems for people.
This. Most cars nowadays come with the so-called "smart" alternators that vary voltage wildly depending on the current driving conditions.
One minute you might be accelerating and the onboard voltage drops as the battery supplies most of electricity. Then, as you reach the crest of a hill and start engine-braking, the car frantically tries to convert all the available kinetic energy to electricity, raising the onboard voltage to quickly charge the battery.
>This. Most cars nowadays come with the so-called "smart" alternators that vary voltage wildly depending on the current driving conditions.
Which in practice means that they do a very miserly job charging the battery and are a ton more sensitive to a battery being in less than tip top shape so you can expect your battery lifetime to go down.
But it's a "win" because they pushed the serp belt change outside of whatever interval the reviewers who calculate TCO care about and they saved .000003mph in the process.
nit: Some vehicles can use a two stage charging system where if the ECU is not trying to charge the battery and the power draw is otherwise low, the voltage sits in a lower range rather than constantly float charging the battery. This can surprise you if you're trying to diagnose a battery issue!
Saw up to 800A on units like the FSD for the short time until the caps were full. Slow starting a SoC is a software problem, slow starting the Cs and keeping the impedance low at the same time a non-trivial hardware problem.
I typically fault anything above 15.6V as “that’s a bit high, your alternator might be on its way out” when working on automotive / caravan / camper van appliances and accessories.
you're correct. a '12v ICE' alternator generates up to 14.8-15.2v. Most automotive stuff can operate between 9ish-16ish-v , of course totally depending on the product.
of course this is just a modern interpretation. older stuff runs at 6v and some weirdo offbeat cars have a 24v/48v rail sitting around somewhere. Cop cars often had alternators that put out weird voltage ranges for certain equipment, or dual 12v for high amperage output.
Even just a "12v" automotive battery itself is mostly dead if if actually reads 12.0V. Fully charged is around 12.6 or 12.7. If a car had an electrical system that actually ran at 12 volts, the battery would always be dead.
"12v" in reference to anything automotive is very much a nominal reference.
> Whilst cranking, an ICE car will drop to around 6 volts (then maximum power is extracted according to thevenim's theorem).
> That means all computers etc will work at 6v.
Not necessarily all of them. Plenty of stuff will drop out while cranking; hopefully not the computers that run the fuel injection and ignition, though.
Not a car engineer, but those motors can be pretty high A, so this could also just be a feature that helps the starter get as much power as it can while cranking.
The specs say no less than 6volts. In the real world when the temperature drops down to -70F or colder and batteries get old the voltage goes well below that: deal with it.
You are probably right. Surprisingly the first controller I tried didn't work. I assumed the voltage was too high since it worked in my other (much older) car. I found a reference online of people that tried a particular brand/model and that's what I went for. Thankfully my car isn't the model with the internal 18v battery.
Voltage isn't the whole story, controllers also need to survive current spikes and power transients, and Tesla's rails may not look like generic 12V gear.
> " I needed this because both the computer and a screen were being sold with the cables cut a few centimeters after the connector (interestingly most sellers did that, instead of just unplugging the cables)."
Can't you just solder some extra wires onto the cut off bits, rather than having to try and find a compatible cable? They've left the connectors in, and that's the hard bit, the rest is just wires
He does that in the write-up, though it's unclear from the photos if he actually soldered it or just twisted up pairs. The discussion of part of the wire melting also raises an eyebrow.. no idea about the authors hardware experience, but I've seen software guys use piddly 24 AWG on a multi-amp circuit
LVDS implies differential signals and are designed to minimize EMI and can be hard to splice while still maintaining signal integrity. They can support high data rates (ethernet cables also use twisted pair LVDS). Theoretically this should be feasible up to 100s or even 1000s of mbps
From messing around with these units from various cars, which often need more enablement than these, sometimes it’s nice to just know your interpretation of the wiring diagrams isn’t at fault when chasing down “no lights at all” issues.
Excellent detective work. I had no idea you can get a Tesla's computer off market. I wonder if these may be the last decade that we may be able to get root access to our on hardware consumer products. Keep the good work up.
The part about reverse engineering the boot process just to get a display output is wild — most people would have given up three rabbit holes earlier. Curious what the latency feels like running actual Tesla UI on desk hardware, does it feel snappy or is there noticeable lag compared to in-car?
ECU software development is sort of my day job. If you're going to go down this path, I seriously recommend getting the specialized plugs and connectors and making your own wiring harnesses to whatever size you need. It's absolutely easier than manhandling a full wiring harness or cutting one down. Cheaper, too.
I work on automotive software (not Tesla), and it's like this partially because it makes development _way easier_. Rather than needing to get a whole car to the dev team, you just give them the specific part that they're working on. Anything that needs outside features usually just fails gracefully (e.g. no speedometer or no location for maps). These are usually mocked for testing, or you add the specific ECU that provides it for your testing setup if needed.
Modern cars have tens of ECUs, so if you had to have all of them for testing, that would get unwieldy extremely quickly. Not to mention that cars are pretty resilient to having random parts failing, you don't want to lose the entire dashboard just because the ECU that provides camera data failed, or something.
I work on that software too (again not Tesla). One other thing - often the hardware we get is per-production and has a list of errata that will be fixed before production. They don't want to make too many of these for engineers because it is done in a fast turn-around higher cost factory and it is expected that once in a while things won't work at all (someone forgets to connect power to the CPU pin or something similarly stupid that requires a lot of manual wiring to fix on each one). As such you have to justify getting any controller and often share. Once the product is shipping they make many of them and it isn't a big deal to get one on your desk - but you have now moved onto the next thing and so the problem returns.
Yeah, I expected some gigantic writeup about tricking it into thinking all other systems are connected to it but maybe it's made this way so it's easier to repair without needing the whole car
It's funny how the biggest problem turned out to be a mostly mechanical part, the rather trivial 6-pin connector.
Given the presence of the wiring schematics and the mechanical dimensions, I'm surprised that the author did not try to 3D-print the mechanical parts of the connectors, givem that the electrical parts extracted from the BMW connectors did fit.
That particular statement is also wrong. For this particular setup you can also buy the correct cable (making sure to not get the one that inverts some pins!!!) for cheaper. Mine was around 15$.
Violet HSD Code D 4+2 Pin Female to D Female Jack Connector 6 Pin HSD LVDS High Speed DataTransmission Harness Wire LVDS Cable
https://a.aliexpress.com/_EuGOh9e
Parts and labor combined that would be a 4000-5000 eurobuck job. A Nissan importer tried to weasel out of a warranty replacement on a friend's X-Trail so they first offered to completely refund options that had become unusable.
The car was full of issues and probably spent more time in the shop than on the road. Nissan finally had to buy it back after service techs failed to tighten the oil drain cap after an engine overhaul.
> Turns out that actual cars don’t have individual cables. Instead they have these big “looms”, which bundle many cables from a nearby area into a single harness. This is the reason why I could not find the individual cable earlier. They simply don’t manufacture it.
Typical setup for cars (and lawn mowers). As a software guy my first instinct is, computing power is cheap enough, seems like a CAT5-like thing running between all components would do it. Speaking as a software guy - meaning I'm probably missing a lot of the big picture. On the other hand, it's a lot easier to safety-check a mechanical lockout that physically opens a circuit, than something running on software.
I read somewhere that the reason they don't typically use IT networking cables / tech is because normal IT infrastructure is a lot less strict with things like packet loss. It's actually not a huge deal to drop packets here and there, especially if any given component is at capacity. But in a car, some devices are super chatty and you can't be dropping packets much at all.
That said, I'm sure there's gotta be a better way to solve it with less copper. And I think they did something like that with CyberTruck.
> ...in a car, some devices are super chatty and you can't be dropping packets much at all....there's gotta be a better way to solve it with less copper.
I know CAN is a thing for a while now, and in the aviation world they have ethernet-derived standards like AFDX etc. But for some reason cables abound.
Congrats, OP has recreated a test/development bench, the bane of developers working on automotive software development all around the world. They're so close to being a real vehicle that you think you'll be able to get a lot of work done, but they're not, so you don't.
Honestly I love it. Few things develop a more fun camaraderie than a bringup bootcamp with two precious/priceless new samples on a large conference table, and everyone being very careful to keep cups/mugs very far away.
And a soldering robot with a specialist a few rooms away to beam down the latest errata into physical form, at times.
Tracy Kidder just died, and Soul of a New Machine was a favorite of my formative years as an engineer. Once I started in headunit ECU development it felt very familiar to me at times.
I'm a software guy, but the gear has a lot of allure.
This is awesome. Curious if these are plug and play and if that's the case where is the memory that tells you what the mileage is. If it's attached to the computer than the mileage would be off if you switch/repair it.
Completely unrelated. Would be interested if you figure out how to retrofit the new adaptive shocks on performance models to the older cars. Something I would love to do if I had hobby time. I'm pretty sure they fit physically, but needs to be connected to the main computer. I likely would never touch the main computer unless I got root access. In my brain I was thinking about a separate system made with raspberry pi's.
You can, at the very least, retrofit the Juniper suspension onto the old car [1]. I haven't ridden in the new Performance yet, but I recently got a 2026 Model Y and the suspension is night and day compared to my 2024.
Hey, I just remembered my school used to have ages ago some cool power supplies (I think from Agilent?) that were very idiot proof, they had current limit with a dial that I think didn’t went over 1A or perhaps even less, and they would instantly disarm on short circuit (and indicate it with a led), and also the voltage dial I think wouldn’t go over 25V. I remember it was very big and heavy, but it survived countless students that used the lab daily.
Nowadays, is there any power supply available that is that resistant or is the recommended approach to get an used old one? Does anyone have a power supply at home that is also used by kids with a brand/model they would recommend? Thanks!
What you're describing is a lab power supply. (The "instantly disarm on short circuit" is overcurrent protection, which is a standard feature.) The name brands like Keysight or Rigol are kind of expensive, but there are a lot of no-name models on Amazon which will do the job well enough.
Thanks! I had a BK Vision or something similar at some point and it just blew up. I will give it a search for these brands, sometimes I find a well-cared used one from the more expensive brands at good prices so that’s what I will look for first. :)
> We ordered the chip and took the board to a local PCB repair shop, where they successfully replaced it and fixed the MCU.
What is a "local PCB repair shop"? All the guys who used to fix TVs and radios are gone. Anyone else (not living in China) having trouble locating such an outfit in their neighborhood?
When I’ve brought atypical stuff in to be repaired at one of those shops they have been absolutely willing to solder whatever, however they did have just one “soldering guy” for every shop in the metro area who only came as needed. So just keep that in mind if you’re in a hurry or want to talk an atypical task through with someone. Probably call ahead.
> A REST-like API on :8080 which returned a history of “tasks”
I am curious to know what kind of historical tasks- since it's a media control unit; does it show what kind of media was being played in the last trip? does it reveal any other info about the driver?? There might be a privacy angle here that you could exploit and share it with Tesla.
They hit Odin. Odin is the diagnostic tool of Tesla.
The tasks they've seen are like "TEST_BRAKE_X_STIFFNESS-TEST-PRESSURE-BURNISHED" and are used to test different components of the car. They're also used for example to reset FSD strikes.
In Tesla terms, the infotainment does much more than just playing music - it has full access to the rest of the car.
I remember back when Chrysler did that and researchers were able to shut a Jeep down mid-drive by attacking the internet-connnected infotainment. This doesn't sound great.
You need to be physically connected to the ethernet port and service mode must be enabled. On top of that, to run these you need service mode plus, which requires a subscription (signed JWT). Additionally, IIRC, most of these can't be run if the car is not in park.
People need to request the source code.. There’s a ton of open source they use that forces Tesla to give you source if you’re a customer and you ask. I don’t get why security people aren’t doing this already.
You get the linux kernel and a bunch of other things you can find on github anyway. You can't do anything useful with that, except what you already can get from any linux package manager.
Sure someone should do it just to verify the process works, but it isn't really useful. (in general companies are very careful to not violate GPL license terms, often refusing to use GPL3 at all)
I would love to use the drive units from a Tesla in a conversion project. Unfortunately, they're cryptographically paired with the main computer, and there's no way to use them.
That is done to kill the "chop shops" criminals used to steal cars and then break them up for parts. You can't do that with computers because of that pairing and so stealing a car pays much less and in turn is less common. (it still happens, it just doesn't pay as well)
The trick is to pick up the main computer and the paired drive unit(s) by picking up a whole vehicle (with a salvage title). There are shops in LA, and elsewhere who do conversions this way.
I _do_ find it weird that the LCDs from crashed cars are so expensive. I wonder if newer models have better screens, so people with older cars upgrade? Or if they're a common failure point?
I have a Model 3, but I can't say I follow the forums.. but I've never heard of screens failing -- I'm sure it happens but I think if it was common problem I'd have heard of it.
I'd guess they fail not on their own, but because they are human interface devices and take the brunt of abuse... e.g. iPhone screens are a popular repair despite being reliable components.
I have a 2023 model 3 and my screen had a small defect develop, a slightly darker area in a roughly half cm diameter area. I think most people would have never noticed but I pointed it out to Tesla service and they replaced the screen.
Some newer models have better (bigger) screens, and some are incompatible since they've slightly changed the connector. Old models (pre highland/ jupiter facelift) have used the same display shown in the article for a very long time across M3 and MY. What usually happens is that they physically break because people are not that careful, so the touch screen ends up breaking - although you really have to put a lot of force to break that display.
My 2016 Model S LCD panel developed the well-known fault of delamination and leaking some kind of sticky fluid.
Turns out the early Model S vehicles used consumer grade LCD panels that weren’t designed for the prolonged high heat you get in a metal and glass box left outside in the sun all day.
Tesla since upgraded their vehicle screens to proper automotive-grade LCDs which are excellent.
My point is, automotive-grade hardware is higher spec than regular consumer computer hardware, hence the high prices.
As an aside, I upgraded my whole computer and screen from MCU1 to MCU2 and it was worth the upgrade.
Credit to Tesla for building a retrofit computer upgrade for old vehicles. Thats a non-trivial thing to engineer and I appreciate their effort. Other car manufacturers would prefer you were compelled to buy their latest vehicle instead.
Ha! Reading this comment made me curious, so I went back and looked at the article and there does seem to be a full sized HDMI connector. I wonder if it is enabled, or just for Tesla internal testing?
Granted, I think it would be valuable to look at all sorts of automotive ECUs. I always wonder how the tuning industry does their thing; I shudder to think they're just sitting there flipping hex codes directly in running software...
Anyone finding this fascinating, please check out Openinverter Forum [0]. Ton of work has been done in decoding CAN messages, DBC files are floating around, open source firmware and controllers are available for Tesla and others components, mostly inverters and chargers but there are overlaps with the VCU and displays as well.
I'm amused reading the terms and requirements the author mentions in the bug bounty program for researchers gaining root access (under 'Vehicle Targets') - https://bugcrowd.com/engagements/tesla
"To promote further security research, Tesla offers security researchers the opportunity to retain root access on their infotainment system even after their reported vulnerability has been patched. In order to qualify, a researcher must send in a valid report describing a novel way to gain root access on a Tesla infotainment system. Upon confirmation, Tesla will instruct the researcher on how to use their existing root access to enable the researcher SSH feature, along with an SSH certificate for the researcher's public key (tailored to their specific hardware ID). The certificate restricts SSH access to the local diagnostic ethernet link. Tesla may renew the certificate as long as the researcher continues reporting vulnerabilities."
> Turns out that actual cars don’t have individual cables. Instead they have these big “looms”, which bundle many cables from a nearby area into a single harness. This is the reason why I could not find the individual cable earlier. They simply don’t manufacture it.
I was really surprised to read this at the end of the article -- how could someone be this deep into a project of this depth and not realize this?! Not only because all cars (...er... all vehicles) are wired this way, but also because the documentation they were referencing has plenty of detail to show this... there's even a whole picture of it (and to Tesla's credit they have amazing free docs): https://service.tesla.com/docs/Model3/ServiceManual/2024/en-...
Even if you know that cars consist of a single wiring harness, it's not implied that they aren't modular and the individual cables cannot be purchased separately.
Cars usually consist of multiple harnesses -- as it is in this case as well. The harnesses are the cables in a car. That is the part you can purchase because that is the part.
> and to Tesla's credit they have amazing free docs
Not to Tesla's credit, they had to be dragged kicking and screaming into it (primarily by Massachusetts) and their right to repair legislation through a solid chunk of malicious compliance:
1. When told that they had to have a site for people to order parts, Tesla put up a site that had every single item as "Call us", including the most simple of bolts. And when a few places called, "Sorry, that's not available to you".
2. The service manual was originally only available in a few locations in MA, and had strict conditions: you had to book in advance, there was a $100 fee per booking, and you could only view the manual on premises, and could not bring electronic devices into the room with you, just pen and paper.
The docs they have are great, and who knows how their attitude would have changed over time, but they absolutely didn't want you to have it, initially.
> The access story has been inconsistent over the years. Tesla has opened up free access to both the service manuals and diagnostic software in the past, but that was apparently a mistake, and loopholes were quickly closed.
"Always ... all free to use". Not so much.
And before that, even less available.
I will grant you for number 2, there seems to be some ambiguity - some people claimed it was only if you needed to actually use their diagnostic tools, because Tesla wouldn't sell them to anyone at the time (which is also in contradiction to your "everything you need, all free, always").
I will say I’m surprised how far apart the two boxes are in the car. I guess they’re not where I thought. I would assume they’re both up near the dash.
The passenger side kick panel or behind the glove box are two very common places for vehicle computers -- some cars have them under the hood, which I always thought was a bad idea.
My RAM truck with the Cummins diesel engine has the engine computer mounted on the engine block. You'd think the heat and exposure to the elements would make that a bad idea, but I suppose Cummins knows what they're doing.
Sounds alright until you realize after spilling a bunch of flower vases in the trunk (hatchback) that the computer has literally no case on it and immediately shorts out while driving. Or a passenger spills a drink in the rear seat cup holder.
There is now a recall notice to pull the back seat out to install a $5 plastic cover over the thing.
And yep, it’s the main computer for the car which controls the electronic transmission etc. Immediate full on engine-shuts-off at speed on the freeway and you require a flatbed to tow it away level of broken. I’m sure the engine ECU is in the engine bay, but holy hell what a surprise!
I had a car with an all wheel drive computer in a similar spot in the late 2000s.
I had a small crack in the rubber seal around my sunroof from parking outside in the elements. When it rained, water seeped in, made its way down the a-pillar, pooled under the seat, and fried the computer.
Expensive fix but I was able to drive it to the shop.
Hehe I was thinking about FCA/Stellantis vehicles when I wrote that. I know it works and there are components made to work in that environment but it always felt intuitively wrong to me. Especially when the other side of the firewall is a much better environment and not far away
It’s because when placed inside the engine bay, the large wiring harness is shorter, which is not only cheaper, but also shorter wiring helps with the consistency of electrical timing and reduces noise.
Yes they do. They can tolerate engine bay heat, but not exhaust heat. They are usually shielded from getting soaked.
Some Mazdas put the metal-cased engine computer in a plastic air box that feeds cold air from the front, to help ensure the engine computer stays cool enough.
In general, I believe the cooling airflow from the frontal air and the cooling fans keeps engine bay in check.
Yeah, on the Cummins the ECU is mounted on the intake side of the engine away from the exhaust and turbo and toward the front right under the fuel injection pump so it gets lots of cooling air.
This thread is interesting to me 'cause I'm also a software guy and recently took a job dealing with building fighter jets and the amount of engineering going into the wiring and computers on those things is insane. It's been a very interesting learning experience.
> I was really surprised to read this at the end of the article -- how could someone be this deep into a project of this depth and not realize this?!
Usually, for most other vehicles, the connectors are either standardized (e.g. radios, ISO 10487 [1], high-current chargers by VG 96917) or the foundation plugs, sockets and re-pinning tools are readily available by the vehicle manufacturer or by aftermarket suppliers.
Tesla truly went out of their way to make the life of third parties (such as wire harness repair shops) more miserable here.
When canbus is already two wires, and by definition, is a bus, so you can just keep stringing those two wires to any module you need. I know Ethernet BUSes exist, but what advantage would those have to canbus then? They're both two-wire buses.
Tesla also went to a 48v wiring harness in some of their vehicles to allow them to power more equipment with less copper. It might be one reason why they use nonstandard connectors, so people don't attempt to hook 12v equipment to the system and also the higher voltages might require connectors rated for it.
Now they just have to take the next step and have everything in the vehicle running on PoE.
Software people tend to overestimate their knowledge of other disciplines, writing it off as "easy" or work beneath them. Being overpaid compared to your peers certainly doesn't help dispel this feeling. Some people have built entire careers around designing wire looms.
A professional scientist I know (tenured, professor) recruited me to set up a backtesting framework for a predictive finance model. When the results were not as they expected (this person does not work in finance and never has), they asked to see the code, then told me that claude had found a problem with the way some of the calculations were done (there was actually no problem), supplied the claude comments, and told me to change the code to match what they thought was correct. I did it anyway. Had they had more expertise in the domain (finance), they likely would have been able to leverage claude as a tool rather than inadvertently pursuing a very stupid mistake. Domain experts tend to doubt their ability to excel in other domains which is amplified by LLMs.
This sounds rather similar to the form of scientific fraud where you first create a conclusion, then invent/manipulate the data until it supports your conclusion.
They suddenly act as if Claude has awarded them with a second PhD in CS. Now they know everything and everything you tell them gets filtered through Claude.
It's like "software dude thinks he can do hardware", but on steroids. They don't know what they don't know and they think they have a panacea in their hands.
Don't you know? Software is beneath them and the fiddly bits are just standing in the way of them getting their BigImportantWork™ done.
Consider whether this is an uncharitable comment --- someone with little expertise in a discipline has made a rookie mistake and didn't realize that the wires weren't produced individually.
Professionals overestimating their knowledge is a very common thing!
Try working on a software project as a non-developer and see if you still respond so negatively to their sentiment. I can’t tell you how many times developers tried to arrogantly and dismissively explain design principles to me, as an experienced, degree-holding designer, because they skimmed a whole Tufte book at some point.
I was a developer for a decade before I went to school for design, so I’ve seen it from the other side. It’s not all bad: that overconfidence can lead people to tackle problems they’d abandon if they really understood the domain’s complexities. But often it presents like developers acting like their genius developer brain allows them to solve difficult problems in completely different fields with a few glib analogies and a few brief thought experiments.
It actually stands for "lizard brain"... it is (or at least was) an Infineon Aurix control and monitoring microcontroller, they may have changed to a newer one.
I am surprised that they are surprised that car wiring diagrams are online. People wouldn't accept cars without online service manuals and schematics, and some states mandate them by law. I just looked up this subsystem for my car via my public library. https://appcontent.chiltonlibrary.com/chilton_images/Honda/E...
i wish the ui on those things was more visually appealing. between the cheap looking gloss finish on the display itself and the unextraordinary ui, it's just kinda blah. one can have a debate about to screen or not to screen or whether to use vfd displays or whatever and i get the importance of cost control but it should look good and it really doesn't. the graphic of the car looks like a cartoon.
i think a lot of people do. i don't know what it is, there's maybe just something about the car graphic that doesn't sit right with me. the front/side view when parked just seems cheesy for some reason. maybe because it's meant to show unclosed doors or something and when everything is set the car's status is car which is redundant.
It does show open doors etc. but if not that then what would you show on the screen? You can already shrink it so the rightmost 3/4 of the screen is the map, leaving just 1/4 of the screen for the car visualization and indicators.
maybe it's the quasi-photorealistic nature of the car image that bothers me. it's not a photo, it's not a schematic, it's not a diagram. it's too artificial to look like a photo, yet too realistic to look like a schematic. or maybe the physically implausible lighting.
Animations could probably be faster and touch areas for opening trunk/frunk could be larger.
But then I'm driving brand new Fiat RV with CarPlay this week. Cruise control by itself has 2 or 3 bugs, and that's not even trying to be picky. Or how's this - can't hotspot from me while phone is in carplay. Can't pinch-zoom or pan maps in 2026 and myriad other things that makes me cringe when people moan they don't buy Tesla because lack of carplay.
