Hacker News Reader

Comment by liminalsunset | Hacker News Reader

Seems like there's been relatively less attention on the 48V/800V architecture and the Ethernet comms/steer by wire setup, but I think those are what make this particular vehicle quite interesting. Up until recently I think 48V power electronics have been fairly rare outside of telecom/other high cost situations, but I suppose one major difference is that at 48V, transformer-based DC-DC conversion is usually a better choice for digital electronics compared with a simple buck, which historically seems to be a significant jump in cost and faff to design. Perhaps PCB planar transformers will help here, but this may be one reason we don't see 50V EPR USB C devices yet (disappointed that the Cybertruck didn't decide to be the first 240W USB-C device, still stuck at 65W)

Seems like if there were to be some kind of electrical failure, now with full steer-by-wire it could be quite a harrowing situation.

Based on the images posted from the event [1] it seems to have two discrete power steering motors (we know from previous vehicles that these have two motor driver channels internally for failover), so there's probably a decent amount of redundancy here. There seems to also be some kind of a sensor that's only populated on one side, though I can't tell what it does from the picture (position sensor?)

Given that the steering is variable gain now, seems like the vehicle speed data would also be important, as in, the communication bus is now quite important for the function of the steering module.

With the Ethernet architecture they're using now, it's likely that it's not as problematic as with CAN bus (where water intrusion can take down the whole bus) but it would be interesting to see what the design choices there were.

[1] https://twitter.com/DriveTeslaca/status/1730305183572177107/...

View on HN

Lexus is already doing steer by wire, they used it to make a vehicle with a yoke that didn't suck to operate (as much anyway): https://www.motortrend.com/reviews/2023-lexus-rz450e-steer-b...

What happens if the engine shuts off (and the alternator stops spinning)? Does the steering just run off the 12v system?

That Lexus is an EV.

Engineering Explained did a really good video about this: https://www.youtube.com/watch?v=agMrewRJTow

To save going through that whole video for the answer: they have redundant power supplies (3 batteries total) and redundant actuator motors.

But the fact is if there is a bad enough electrical/wiring fault, especially as the car gets older, you are toast and you have no steering.

Of course the same thing can happen with mechanical steering (eg a broken linkage) it comes down to the quality of the design and build.

The whole design will be such that a single electrical fault can't make it fail. It probably has at least 3 channels.

Probably won't even allow you to start the car if one of them is faulty.

This kind of thing will almost certainly be designed to comply with ISO 26262.

I've had the power steering cut out on me in a traditional ICE car, you're pretty much hosed in that situation, too, as far as steering is concerned.

Hasn't VAG had steer-by-wire for nearly a decade now?

Only as part of four-wheel-steering systems in combination with mechanical steering for the front wheels.

Nissan has steer-by-wire with a backup mechanical system.

I had a BMW from 2007 that had something called "active steering", where an electric motor in between the steering wheel and the rack-and-pinion sort of "adjusted" the steering, which gave it a variable steering ratio (add according to speed) and also stability control interventions (it would countersteer in a slide even if I didn't turn the wheel).

The stability control part is probably the biggest safety benefit of this steer-by-wire system.

Other manufacturers have had similar levels of electronic steering "augmentation" of one kind or another for a long time (e.g. electronically steered rear wheels). In this kind of thing, the safety case is probably that the system only provides small adjustments so that:

1. The driver can overcome any errors in the electronic augmentation system using the main steering system

2. If the system fails, it can switch off and the driver can still steer the old-fashioned way.

Yes as well as some BMW’s, McLarens, at least some of the electric Hyundai/Kias and I’m sure a bunch of others I’m not as familiar with.

The shock over it on HN is just because of how few people here work in automotive.

In those systems the steer by wire works in cooperation with a mechanical system. Other cars have had backup mechanical steering systems that engage if the steer by wire fails.

What's new about the Cybertruck is it _only_ has a steer by wire system with no mechanical linkages at all. But this is being misreported as the first steer-by-wire system which doesn't cooperate with a mechanical linkage- Nissan has done that for quite a while in some models.

Sounds like a terrible idea to me, honestly. Even Boeing aircraft still have a mechanical linkage. And Cybertruck won’t be a fighter jet no matter how much Elon proclaims it so, it’s just worlds away in terms of maintenance and safety engineering compared to what happens on an aircraft. Not saying it will be common but some sucker will eventually be killed by this.

> Even Boeing aircraft still have a mechanical linkage.

Airbus has been FBW for nearly 40 years now. And recent Boeings are FBW as well, the mechanical linkages are only between the yokes, something which is not relevant on a car.

When i hear fly-by-wire i think of wire rope going from the flaps the cockpit

I think aeroplanes are not a good point of comparison.

For one, mechanical steering is still very complicated, with many meters (miles?) of hydraulic hoses, actuators, pressure control systems etc.

Two, aerospace is healthily paranoid about QA and failure modes. I'm sure a steer-by-wirw can be extremely safe when done with that mindset. But Im not sure if that's the car industry's mindset, let alone Tesla's.

Finally, fly-by-wire offers enormous safety benefits in aeroplanes, because of what it enables. But I have no idea what steer by wire enables for a car.

This answer seems to generate its own proof.

>For one, mechanical steering is still very complicated

Yes, and digital steering is by nature even more complicated.

>aerospace is healthily paranoid about QA and failure modes. I'm sure a steer-by-wirw [sic] can be extremely safe when done with that mindset

What? So if we have the right 'mindset' then we can ignore any complexity and adapt anything we want. Why even let users drive cars now? Make their fleet 100% autonomous. Just do it with a mindset of QA and safety!

> Not saying it will be common but some sucker will eventually be killed by this.

People are killed by vehicles every day, and some of those are vehicle failures as well. Saying someone will be killed by it isn't interesting. The question instead is will it fail more often than existing designs, which is something only time will tell. But we've had throttle by wire for a long, long time now, and unintended acceleration issues caused by the system failing are largely unheard of. It's completely possible to engineer a robust electrical system just as it's completely possible to have design faults in the mechanical systems that cause premature wear & failure.

Yikes! Even assuming it all works correctly and safely, who actually wants a steering wheel like that? What's the upside?

The only possible advantage I can think of is that when you start the car up again after having reverse parked, it'll be more obvious what way the wheels are currently pointing, so you can straighten up. But that's such a trivial problem.

>>What's the upside?

A much easier to manouver car, because you can get true variable speed ratio steering.

It's explained really well here:

https://www.youtube.com/watch?v=agMrewRJTow

Tbh I still think it's a bit of a silly idea. But it's not completely without its merits(the one done by Tesla was though - that was completely entirely pointless).

I just meant the yoke -- variable ratio steering would still work with a wheel, right?

I'm skeptical but I can see the potential benefits of drive-by-wire, definitely. The yoke, nope (except maybe for racing drivers).

Oh I guess it's just that once you elimited the need to ever turn the steering device more than 90 degrees, it doesn't need to be a wheel anymore - same reason why bicycles don't have steering wheels but handlebars.

The yoke is to make visibility of the dashboard better. You can deal with rectangular screens easily then instead of needing to force things into a curve. Not necessarily a good enough reason, but it's not entirely frivolous.

Forgive my ignorance, but is "variable speed ratio steering" not what the Citroen SM was doing 50 years ago? https://en.wikipedia.org/wiki/DIRAVI

No, that just seems to be variable speed power assisted steering - so at high speeds it gives you little assistance, at low speeds it gives you loads.

With this it literally changes how much your wheel turns for the same amount of steering wheel/yoke movement. So at high speeds turning the yoke by 90 degrees will turn the wheels maybe 10 degrees so you don't kill yourself, but at low speeds the exact same movement will move the wheels 45 degrees so the car is a lot more manoeuvrable.

Ahhh, got you, thanks. I'd misunderstood the Citroen system as being somewhat like your second paragraph.

Transformationally easier packaging requirements.

Can you clarify? You just mean by going from a round wheel to a rectangular yoke, right?

(I assume you don't mean removing the whole steering column. That's definitely a big deal, but it's independent of the shape of the steering control)

I was talking about the steering column - looks like I missed what OP was focusing on.

I am curious what cybertruck will charge like at existing superchargers that max out at 400V.

GM's clever solution is to present the battery pack as 2 400V packs. It does pose some very tricky pack balancing requirements though.

Other 800V cars charge quite poorly on 400V stations, usually with a hard cap at much lower than the potential max charge due to inverter limits.

E-GMP (EV6, Ioniq5) actually charges quite fast on 400V as they run it through the motor/inverter which already pumps out more than 100kW

I think no one else is doing that at the moment.

Wait, what? They're converting their motor controller into a 400V-800V DC transformer?

yes: https://www.youtube.com/watch?v=xm4VDfdKUm0&t=719s

The E-GMP solution is also a clever one. I believe the inverter is limited to about 150 kW sustained (I think it can peak a little higher to the motors). 150 kW isn't bad but it's also a fair ways away from the 230 kW E-GMP will do on 800V.

Cybertruck does the pack splitting thing like the GM Ultium vehicles:

https://twitter.com/baglino/status/1730391023799386480

It will be interesting to see Tesla's rollout of 800V stations, so far they don't have any within their own network.

IIUC, Ultium is 400V in operation and only switches to 800V while charging on 800V units.

This sounds a little different. It operates at 800V and switches to 400V for charging.

I think that will simplify things and hopefully avoid Ultium's weird glitches during charging.

Presumably it can charge either at 400V or 800V depending on the charger it's plugged into. But I think you're right, Ultium is still a 400V drive architecture versus this being an 800V drive architecture.

True, I left out that part. :)

TBH, I'm really looking forward to seeing the Tesla tested on 800V CCS chargers via the adapter. It seems like it is at least possible that it might charge faster on those than on the current 400V superchargers.

Of course, that would change if Tesla starts enabling 800V support on V3/V4.

In the Top Gear interview they confirmed the truck can do 350kW charging on a V4 charger, 15-85% in 18-20 minutes.

https://www.youtube.com/watch?v=uefydJUbRhc @ 34min

Yeah, I saw that, but tbh, I'm taking it with a grain of salt at this point. We've already seen one screenshot of the charge rate at ~70% and it wasn't great. It was showing ~10 minutes remaining to get to 80% too. That was on a slower charger, but the charge rate was less than 80 KW.

Considering how bad the 4680 Model Y was, I'm not optimistic.

> ... but I suppose one major difference is that at 48V, transformer-based DC-DC conversion is usually a better choice for digital electronics compared with a simple buck, which historically seems to be a significant jump in cost and faff to design.

There's no such thing as 'simple' transformer-based DC-DC conversion. If you want to use just a transformer, you're looking at AC. If you then want to convert it to DC after, you're going to need a rectifier plus a buck converter. The CyberTruck is 48VDC meaning if you want to do any kind of voltage conversion you're going to use a switching architecture of some sort.

> ... Perhaps PCB planar transformers will help here, but this may be one reason we don't see 50V EPR USB C devices yet (disappointed that the Cybertruck didn't decide to be the first 240W USB-C device, still stuck at 65W)

EPR is 48V.

Switching regulators are a pretty well known quantity now - and super efficient. I don't think there's much faff.

> There's no such thing as 'simple' transformer-based DC-DC conversion

Pretty much all DC-DC converters nowadays are switching - and those chopping - and - rectifying steps you mention would be standard for any DC-DC converter whether or not it uses a transformer. In this particular case, a transformer is necessary in addition to those components, as opposed to a simple fixed inductor.

> EPR is 48V

Yes, I got confused with the maximum rating on the most popular EPR capable tester (Power-Z KM003)

> Switching regulators are a pretty well known quantity now

I'd agree with this, insofar as they're very common and we know how to design them effectively. With that said, I think it can be observed that topologies used with higher voltages generally do come with increased cost and complexity. First of all, the transformer is going to be a custom part, as opposed to something with jelly bean substitutes like inductors. Second of all, the kinds of components typically used with higher voltages like 48V are less integrated than ones for lower voltages, likely due to the kind of process technology you need to use to handle it. It's possible to get "smart" drive chips for applications like Intel processor VRMs that integrate all of the switching, gate drive, and control logic, but with higher voltage you typically have to source discrete gate drivers, controllers, feedback, etc, and I'd think it does add complexity.

> In this particular case, a transformer is necessary in addition to those components, as opposed to a simple fixed inductor.

Interesting, why do you say that? Maybe I misunderstood. Do you have a link?

Why did you call out the CAN bus for water penetration? I thought it was an electrical specification and logical protocol, does it specify wires too?

Correct neither CAN or Ethernet say anything about IP ratings, but I guess having an ethernet switch as a distribution rather than everything on a chain restricts the failure to one device.

This is the same kind of failure you used to see in 10base5 and 10base2 networks, where one computer having a loose terminator could cause the entire LAN to be offline. With 10baseT (and I assume automotive ethernet) the bus is contained in the hub/switch/router instead of needing to be terminated (as in CAN bus, from my brief research anyway)

Although if you wanted to, I don't see why you couldn't build a CAN switch?

Definitely nothing stopping it, according to the wiki star architectures are supported, but I don't think many of the parts suppliers build them that way, would be my guess.

I think the question would be why would you want to? At the point you're doing that, ethernet already exists including plenty of industrial specced hardware.

Presumably so you can reuse all of your existing CAN car parts. Even the lights are on CAN now, it's going to be a lot of stuff to transition to Ethernet.

And you have thieves stealing cars by hacking into the easily-accessible car light CAN bus - a CAN network switch could also act as a firewall to fix those thefts without re-engineering every single part.

It makes sense for Tesla since they're more vertically integrated but other car manufacturers seem more "parts bin" and having to migrate your whole car over to a new architecture will limit the parts you can use massively until the whole industry switches.

A few reasons: Ethernet electrical is more expensive and ethernet drivers are bigger code size then CAN, and more complex (buggier, you have a lot more buffers and way more state to manage with ethernet versus CAN). Plus CANFD can accommodate different frequencies, so cheaper hardware can participate on the same bus. This may change over time as larger cores are used (Renesas comes to mind) and bigger ethernet stacks that are already scrutinized can be shoehorned in.

Just a thought but if its 48V and its ethernet I wonder if it is using power over ethernet?

I'm sure we will learn stuff like this when Munro tears one apart

In this particular case, it's because the CAN bus shares two wires across all devices, with everything connected in parallel. Water ingress or catastrophic failure to any of the wires, connectors, or terminations/devices can result in an open or short circuit, or at least enough load to cause the signal drivers to be incapable of driving the right levels.

Although this particular example is from what some would consider a controversial video, this is an example of how water ingress can result in the disablement of vehicle electronics: https://youtu.be/Gio3DSTkQH4?feature=shared&t=181

Audi A6 48V mild hybrid from 2018 - https://www.greencarcongress.com/2018/02/20180228-audi.html

This one is a little bit different since there are actually still two (well three systems here: 48V, 12V, and the ICE).

These systems have existed for quite some time, since it was a lot easier to use high voltage rated components for all of the 48V high power stuff (e-turbochargers, e-hybrid-compressor-whatevers, other motor driven stuff) but keep just a single downconverter to 12V for everything else.

The problem is that most digital electronics expect something like 1-5V to operate, including most controllers, processors, sensors etc, and typically this conversion is done at the module level in a distributed way. Going from 12V to 1V isn't really a problem (or 24V ish for that matter, which is what they use in notebooks), and 20V and 30V class semiconductors are easy to find and have desirable performance characteristics.

Going from 48V at every module to 12V as an intermediate bus, usually requires a transformer because otherwise you'd be switching the current on for a very short time per cycle (duty cycle proportionate to voltage ratio), and this has implications for efficiency and component sizing. So I'd think this is quite a huge change compared to the old mild hybrid systems

I'm hoping the 800V architecture at least leads to Tesla supercharger stations my Ioniq 5 can actually use after NACS takes over.

The current 400V architecture would charge even slower than the raw math would imply, because it requires the car to handle the voltage difference internally. Going through that path bottlenecks the charge rate at something Chevy-Boltish like 50kW that's simply not very viable for road trips.

I would guess that at least all the version 4 charges will support 800V in the near future. Maybe even the version 3 ones can be upgraded.

why do steer by wire at all? Is it cheaper somehow?

Hmmm.. I would guess decoupling would allow variable ratio steering based on speed or preferences.

Electric power assist already allows that. It's probably cheaper, lighter and maybe safer to have no steering column.

I meant decoupled variable ratio, not mechanical variable ratio[1], not power assist.

sort of like turning mouse acceleration on and off.

[1] https://en.wikipedia.org/wiki/Steering_ratio#Variable-ratio_...

No matter what we think about FSD, Tesla's end goal are robotaxis – eventually probably with no steering wheels. Steer by wire is a stepping stone in that direction.

They almost always only vary the amount of power assistance, not the ratio. The ratio is fixed on virtually all electrically power assisted steering racks ever shipped. There is still a fixed mechanical linkage - decoupling this is what allows the ratio to be dynamically adjusted. To date only a handful of Lexus production cars have had dynamic variable ratio and decoupled steering via drive by wire.

Electric power steering is essentially the same in use as traditional hydraulic systems - they just replace the hydraulic assistance with electric motors.

The big advantage of decoupling and steer by wire is dynamic variable ratio, as well as packaging - no steering column to worry about making space for.

Nah, there are variable ratio mechanical systems. On a rack and pinion setup, for instance, you vary the spacing of the teeth on the rack. ZF's been doing this for decades.

Generally speaking almost all mechanical steering systems are not dynamic though, you are missing the point entirely. It can't change the spacing of the teeth on the fly - steer by wire effectively does. The number of turns lock to lock on a mechanical system is almost always the same, regardless of speed.

With steer by wire, you can easily implement say one turn lock to lock at low speed for easier parking, increasing to 2-3 turns at highway speeds for stability. Which is essentially exactly what the recent Lexus model with such a system does. At low speed it can turn its wheels from one extreme to the other in just 150 degrees of steering wheel rotation - no hand over hand turning etc. 150 degrees is essentially unthinkable on a mechanical rack and pinion system, as it would make the car incredibly twitchy on a highway.

> https://www.motortrend.com/reviews/2023-lexus-rz450e-steer-b...

  It can't change the spacing of the teeth on the fly.

Geez, Elon really ought to start a goalpost moving company. But hey I'll play: yes, yes it can. That's precisely what BMW's active steering is, yes it's mechanical, yes it varies the ratio depending on speed, and yes it's been around for a couple decades now.

BMW active steer doesn't get remotely close to 150 degrees lock to lock when maneuvering (minimum rotation reached is ~720 degrees/2 turns), or offer the infinite and instant range of adjustability you can get with a drive by wire system. But sure, it's just the same. I couldn't care less whether the system is on a Tesla or not, its just abundantly clear there are some design and packaging advantages when you lose the mechanical link in anything with a steering wheel.

The vast, vast majority of rack and pinion systems ever made, both old and modern, don't adjust number of turns lock to lock at all, whereas the ability to essentially infinitely adjust it is inherent in all drive by wire steering systems.

As far as I know there is no technological problem for BMW system to have whatever desired rate of adjustability and speed of reaction. It uses planetary gearset with adjuster motor connected to one of the gears - a bot similar to how Toyota Prius's transmission works. So it's more a question of how manufacturer decided it should feel for the user.

Before we change the goalposts again it's worth noting that Tesla (and to a lesser extent Lexus) make vehicles where the primary focus is on the gadgets and BMW's focus is on the driver interface.

  BMW active steer doesn't get remotely close to 150 degrees lock to
  lock when maneuvering

Well, no. Super fast steering ratios are not an advantage, they're a necessary evil with a square steering control. There's nothing inherent in the BMW design that would prevent such a fast ratio beyond the reputation for emulating analog interfaces with quite a bit of fidelity. It speaks more to the intended use (gadget vs car) than to the underlying technology itself.

  its just abundantly clear there are some design and packaging
  advantages when you lose the mechanical link in anything with
  a steering wheel.

Every time I see this I wonder if the commentor has seen a hydraulic power steering pump. They're tiny as hydraulic and pneumatic drives are often quite a bit smaller than comparable electric ones (e.g. cordless vs pneumatic power tools). The only clear advantage that electric steering offers is one of efficiency (as a result of not driving a hydraulic pump all of the time). The rest is largely a series of trade offs, especially if you look at recirculating ball / worm gear vs rack and pinion drives.

> Every time I see this I wonder if the commentor has seen a hydraulic power steering pump. They're tiny as hydraulic and pneumatic drives are often quite a bit smaller than comparable electric ones (e.g. cordless vs pneumatic power tools).

The pump or motor size is totally irrelevant to this discussion and to the packaging advantages.

There is no steering column with steer by wire. If you have seen a mechanical steering column, as found in all hydraulic and electric power steering systems, it requires a significant amount of space to route from cabin to the front wheels, and you are limited in the paths and angles it can take to reach the rack. This is completely gone in steer by wire - that's the packaging advantage - not the pump, not the motors.

Taken to the extreme, you can do designs that are completely impossible with mechanical steering columns - see the Canoo design for example, which uses steer by wire to create a cabin that would literally be impossible with a steering column:

https://www.canoo.com/canoo/

You could even design interiors that can switch from left hand to right hand drive, with a sliding steering wheel. It also reduces the cost of shipping left and right hand drive variants, as you no longer need to engineer space on both the left and right sides of the car to fit the column for different markets.

But no need to take only my word for it; ZF, as in your earlier examples, also make the exact same arguments about packaging (and also about infinite adjustability, fwiw):

> "SbW's (Steer by Wire) additional advantages include gaining space in general and new design options in the cockpit... There is now more legroom because there is no longer a steering structure running through it."

> "In terms of steering experiences – from sporty-direct to comfortable-relaxed – every setting is possible in the same car. The system can also translate the same steering wheel movement variably depending on the driving situation, i.e. for parking and at low city speeds, the wheels angle (very) strongly for the change of direction, while they angle only slightly at high highway speeds."

https://www.zf.com/mobile/en/technologies/vehicle_motion_con...

> "we can make the steering ratio completely variable depending on the driving situation – unlike the fixed steering ratios used for a mechanical connection between the steering wheel and the front wheels"

> "Installing steer-by-wire steering systems reduces the assembly effort needed to build today's right-and left-hand drive versions of the same passenger car model."

https://www.zf.com/mobile/en/technologies/vehicle_motion_con...

  If you have seen a mechanical steering column

I have in fact had to remove the column on one of my cars, mostly to deal with the attached electronics. It's a relatively small device with a couple of u-joints.

  see the Canoo design for example, which uses steer by wire to create a
  cabin that would literally be impossible with a steering column

See the first three generations of Volkswagen vans.

There is literally air between the steering wheel and the rest of the car in the Canoo, as well as significantly more front leg room - its nothing like the VW, which has an almighty big steering column between the drivers legs. Unless of course ZF and the rest of the car industry are also wrong about it taking less room to package and permitting new designs?

If you have seen a column - congratulations, you will then know it needs room to route to the wheels, and is restricted in the paths it can take to get there quite significantly, vs of course plain ole wires that can snake any way you like. Those restrictions are exactly why your VW vans have a column between the drivers legs.

This is getting genuinely funny now, however I think I'm about done with this one. Enjoy the rest of your day.

… have you not seen the inside of a Volkswagen (or any other rear/mid engine) van/bus?

Sure have!

Again, one mighty big steering column, present on the T2, T3, "the first three generations of Volkswagen vans" etc etc.

> https://www.cnet.com/a/img/resize/309b66cd757b35f8a47ba3d385...

Are you just ignoring the ZF articles now they don't fit your arguments anymore? Again - it's not just me saying this!

The Canoo - no column, and a ton of front cabin space:

> https://i.pinimg.com/originals/96/28/33/96283349b392b1da2f0f...

https://www.zf.com/mobile/en/technologies/vehicle_motion_con...

and you could invert X :)

It means where you put the steering wheel doesn't have to have a firewall penetration and mechanical linkage through to the engines. That's a lot of mechanical joints which need to be continuous, and need greasing, sealing and have limits on relative angles between them that you don't need - as well as they have to have enough strength to transmit the forces involved.

By contrast, with steer-by-wire your system is essentially a black box you bolt on somewhere and then a more compact set of motors which can be coupled to the wheels at whatever is mechanically optimal.

The main reason not to do it is because your control systems need to be perfect. Conversely, above a certain vehicle mass, it's also not realistic to think that a human would be able to control the car anyway if the power steering failed, or that control system failures would still be devastating (i.e. if the power steering starts fighting you, you're going to lose).

Big rigs used to not have power steering, i think with adrenaline you'll be able to steer. I do concede that the power steering fighting you would be bad

I just want feedback from the road, wires don't give me that

You can absolutely get feedback from steer-by-wire. Ask any racing sim enthusiast. Those racing sim wheels can provide enough feedback to rip the wheel out of your hands.

That "feedback" doesn't meet my poorly framed hyperbole. I'm intimately familiar as a driver; both in sims and Real Fast Cars.

At what point does it go from feedback to lies? Some computer saying "yep, this is what's happening", if you're lucky.

I'll yield that has been the trend for eons. That said, I don't want more of it.

What lead to my first comment is the understanding that Tesla doing something where it's... different wires.

Bus vs Ethernet or whatever. It sounds even more muted/numb.

I don't want a real car to feel like a sim. It's getting harder to tell the difference for the wrong reasons.

It's not that the simulators are closing the gap, but that the cars are becoming simulators. Lowering standards is definitely one way to meet them.

Take your "we have feedback at home" and keep it

I'm half joking through most of this, by the way. The problem is there but I'm deliberately overstating it

It's important to keep in mind what the actual market for the vehicle is. It's not a sports car. Road feel is an anti-goal, it's undesirable in this market. Just like it's undesirable in the market for the Lexus - luxury SUVs. All you want here is some dynamic weighting for cornering loads and even then not that much of it.

Yea, that's fair. Antithetical, these 'boats' are likely well served by not feeling as much

I'd appreciate more feel in something basic like an Altima. Not even for remotely high performance situations

I can already hear the road, I might as well feel it /s

More importantly, how my tires are reacting to it. For something so underpowered they spin in the wet from a complete stop, something awful.

It's kind of nice to still be able to steer a car that is disabled and needs to be pushed around. Not always feasible depending on the disability, of course.

Yes, tow trucks are great, but sometimes it's handy to push a car out of the way while waiting for a tow.

It has rear wheel steering, so there would be a lot of linkage needed to do that mechanically. That said, GM has been doing rear wheel hydraulic steering for a couple decades and the front steering is still mechanical in those (at least historically, not sure if that has changed with their EV lineup).

Just about every car manufacture has done rear steer at some point I believe. Nissan had it done initially with a Hydraulic system then electric.

Probably. I know Dodge and Mitsubishi have done it too.

Allows you to vary the steering ratio dynamically (how many "turns" lock to lock). Means a lot less arm twiddling during lower speed maneuvers, and can gradually increase the ratio as speed increases. Traditional mechanically linked steering racks (electric or hydraulic) don't do this and are the same number of turns lock to lock regardless of speed - they can only dynamically vary the amount of power assist, not the ratio. It's been something car manufacturers have wanted to do for a while.

Another benefit in theory is packaging - you can free up more space for cabin, front trunk or motors etc when you don't have to route a physical steering column.

BMW had this on mechanically linked system for 20 years. It's called "Active Steering" in their marketing speak.

It's not really the same at all in practice though. With steer by wire you can get the lock to lock literally as little as you like - 150 degrees at low speed lock to lock in the Lexus system for example. In theory it could go from one extreme to the other in a single degree, its just software, although obviously no one sane would implement this. Active Steering on BMWs doesn't get remotely close to 150 degree low speed lock to lock.

It's a system that changes the ratio of the steering wheel rotation to the wheel angle dynamically while still mechanically linking the wheels to the steering wheel. It does not have the same parameters as Lexus system but many other things differ between BMWs and Lexuses too.

I’ve been driving a BMW with the variable ratio steering rack for over 10 years. It’s great. But it’s only variable ratio, not a dynamic ratio. Don’t confuse the two.

A variable ratio is just a regular rack and pinion setup, except the teeth on the rack are more narrowly spaced at the centre. It’s pre-defined, fixed and static. 30 degrees of steering input will always result in the same movement of the front wheels. It’s mechanically fixed.

The Cybertruck is a truly dynamic steering ratio. 30 degrees of steering input will result in a different outcome depending on othe variables such as road speed.

Active Steering is dynamic. It's not in every BMW though, you can test if your car has it by turning the steering wheel when stopped: it should only require 2 turns between locks with AS.

You also don't need a collapsible steering column, as there is no steering column.

Steer by wire is cheaper to produce, as it allows you to reduce costs for RHD markets (e.g. Australia, New Zealand, UK, Thailand, Indonesia, Japan, etc).

For example, Ford F-150 and RAM 1500 trucks from the US are exported to Australia where they need to undergo an expensive conversion process before they can be sold. This is why a base model F-150 is US$74000 in Australia.

Aren’t basically all modern cars steer by wire? I guess ethernet vs CAN is notable but isn’t that a pretty minor implementation detail?

Nope, you are thinking of EPS. This is a steering wheel that is not connected to the steering rack at all. It is like a racing wheel for a game console and the steering rack is like a car in the game. Just software.

I hope they learned from the Titan submersible to always go at least one step up from the cheapest game controller.

Their whole problem was custom hardware. I don't know why people were so hot on them reinventing tested commodity electronics.

The complaint wasn't that they used an off-the-shelf controller, it's the fact that it was a no-name aliexpress special (and wireless to boot). They didn't even bother to use a well-respected brand like the wired xbox 360 controller, which has been used in all kinds of similar applications (they also didn't have a spare). It was obvious even to a relative layman that this was emblamatic of a cowboy attitude, compared to the carbon fibre hull which is much harder to evaluate.

Seriously? Why would you do that?

> it was a no-name aliexpress special

Not according to this:

https://en.wikipedia.org/wiki/Titan_submersible_implosion

> A modified Logitech F710 wireless game controller was used to steer

Even then, it wasn't responsible for the implosion, right?

Many modern cars are throttle by wire. Orthogonally, many have electric power steering instead of hydraulic power steering, depending what is multiplying your force, but still with actual linkage from your arms / steering wheel, to the turning of the wheels.

I am not aware of many that are actually steer by wire.

Would there be a “safe mode” for steering where it has basic operation if it can’t get speed data?

Drive by wire kind of scares me.

> Drive by wire kind of scares me

Physical linkages break too.

Yeah, but not from a bad firmware update

Yeah... I've encountered thousands of software bugs in my lifetime. Including with critical car systems like the gas pedal on my Toyota.

I don't know anyone with a passenger car who's ever had the steering wheel linkage break to the point turning the steering wheel stopped affecting the wheels. I like the failsafe there. Hopefully at least the brake cylinder is still connected to the brake pedal mechanically?

But firmware doesn't rust and snap or burst, like I've experienced on other vehicles.

Cutting corners is cutting corners. Build cheap firmware or use materials with poor corrosion resistance and you get problems.

It's much harder (impossible, really) for a owner to mitigate against firmware bugs, though. Regularly servicing and inspection should prevent/detect the vast majority of major mechanical issues, not to mention than many are slow-onset.

A firmware bug just happens, and there's no way (as an owner) to detect it beforehand.

Software isn’t unpredictable. Issues that are known can be resolved. Unknown hardware issues are just as bad, and they do happen. Most safety recalls are mechanical in nature today.

Fly-by-wire systems in airplanes are way more complicated than cars and they have excellent safety records, which even improved as those systems became widespread.

The reason so much bad software exists in the world is because of the prevalence of people writing software without an accredited engineering education, licensing, or regulation. Software can be good if built to the engineering standards expected of any other engineering discipline.

Has this ever happened to you? Good god, i want less SW in a car, not more. W/o power steering i can at least control a vehicle, only the brake pressure lost is a serious problem.

Same with keyless starters, i remember this story about some French car experiencing a kind of highway run-off that couldn't be stopped due to this BS solution to a non-existing problem.

"But hey, that's not innovative enough"

Yes, I've had: - cruise get stuck on, car didn't have disable cruise on breaks - gas pedal stuck under seat mat - gas pedal siezed on due to cable issue - brake pedal failure to master cylinder - steering wheel slipping during turns - shift lever disconnect - breaks failed due to cracked rotor - breaks failed due to leaking master cylinder

shit breaks a lot when you'e been driving for 35+ years

Yup. Got a set of Saturn Vue gearshift linkage cables in the garage because one broke and left the car in a non-drivable state and it had to be towed home. I have the entire assemblies because I was able to get by with just replacing the broken part and not entire cable going through the firewall.

I could go through a list of critical mechanical systems that have broken on cars I've driven over the years, including my beloved Toyota pickup.

> i want less SW in a car, not more

I just want Good Engineering Practice. Good engineering encompasses a lot, including doing a DFMEA (Design Failure Modes Effects Analysis) that will advise whether the software system is safer than the hardware linkage or vice versa.

A lot of controls in airliners are via software and electrical signals.

As long as the same rigour that is applied to the aviation industry is used and that they don't use an MCAS, I think I'm ok with it.

Yep, Airbus went full fly by wire 40 years ago, and while the feedback and some of the integration choices have been debated, I can’t say I remember concerns about the reliability. Boeing has also switched to FBW on their recent planes. And military went FBW a while back, in fact since the F-117 FBW has generally become a requirement because they’re all aerodynamically unstable to improve manoeuvrability and integrate stealth geometry, so they pretty much can’t fly without the FBW system adjusting constantly.

Though I would also guess the mechanical linkages of an airliner are under a lot more stress than a car’s, so the gains from FBW would be a lot larger, even aside from not having to run mechanical linkages through the entire hull.

48v AC and steering pumps are commonplace in hybrids and EVs.

48V power electronics are quite commonplace in industrial applications.

Reply on news.ycombinator.com