10 ms·

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

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/photo/2 https://twitter.com/DriveTeslaca/status/1730305183572177107/...

TheLoafOfBread 3y ago
Audi A6 48V mild hybrid from 2018 - https://www.greencarcongress.com/2018/02/20180228-audi.html https://www.greencarcongress.com/2018/02/20180228-audi.html
- liminalsunset 3y ago
  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
whelp_24 3y ago
why do steer by wire at all? Is it cheaper somehow?
- m463 3y ago
  Hmmm.. I would guess decoupling would allow variable ratio steering based on speed or preferences.
  - lttlrck 3y ago
    Electric power assist already allows that. It's probably cheaper, lighter and maybe safer to have no steering column.
    - m463 3y ago
      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_steering https://en.wikipedia.org/wiki/Steering_ratio#Variable-ratio_...
    - yreg 3y ago
      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.
    - giobox 3y ago
      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.
geoelectric 3y ago
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.
- oittaa 3y ago
  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.
theluketaylor 3y ago
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.
- solidr53 3y ago
  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.
  - bdamm 3y ago
    Wait, what? They're converting their motor controller into a 400V-800V DC transformer?
    - p1mrx 3y ago
      yes: https://www.youtube.com/watch?v=xm4VDfdKUm0&t=719s https://www.youtube.com/watch?v=xm4VDfdKUm0&t=719s
  - theluketaylor 3y ago
    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.
- stetrain 3y ago
  Cybertruck does the pack splitting thing like the GM Ultium vehicles: https://twitter.com/baglino/status/1730391023799386480 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.
kllrnohj 3y ago
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-by-wire-yoke-review/ https://www.motortrend.com/reviews/2023-lexus-rz450e-steer-b...
- alistairSH 3y ago
  What happens if the engine shuts off (and the alternator stops spinning)? Does the steering just run off the 12v system?
  - vostrocity 3y ago
    That Lexus is an EV.
  - jimt1234 3y ago
    Engineering Explained did a really good video about this: https://www.youtube.com/watch?v=agMrewRJTow https://www.youtube.com/watch?v=agMrewRJTow
    - dkjaudyeqooe 3y ago
      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.
      - HPsquared 3y ago
        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.
demondemidi 3y ago
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?
- boznz 3y ago
  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.
  - boznz 3y ago
    Just a thought but if its 48V and its ethernet I wonder if it is using power over ethernet?
    - grecy 3y ago
      I'm sure we will learn stuff like this when Munro tears one apart
  - jaggederest 3y ago
    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)
    - kalleboo 3y ago
      Although if you wanted to, I don't see why you couldn't build a CAN switch?
      - jaggederest 3y ago
        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.
bilsbie 3y ago
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.
- HeyLaughingBoy 3y ago
  > Drive by wire kind of scares me Physical linkages break too.
  - Yodel0914 3y ago
    Yeah, but not from a bad firmware update
    - geerlingguy 3y ago
      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?
    - kube-system 3y ago
      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.
      - Yodel0914 3y ago
        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.
KennyBlanken 3y ago
48v AC and steering pumps are commonplace in hybrids and EVs.
upbeat_general 3y ago
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?
- agloe_dreams 3y ago
  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.
  - Projectiboga 3y ago
    I hope they learned from the Titan submersible to always go at least one step up from the cheapest game controller.
    - gopher_space 3y ago
      Their whole problem was custom hardware. I don't know why people were so hot on them reinventing tested commodity electronics.
      - rcxdude 3y ago
        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.
        gopher_space 3y ago
        Seriously? Why would you do that?
        3y ago
darkclouds 3y ago
[dead]
arcticbull 3y ago
> ... 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.
- liminalsunset 3y ago
  > 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.
  - 3y ago
ajwilson 3y ago
48V power electronics are quite commonplace in industrial applications.

TheLoafOfBread 3y ago

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

liminalsunset 3y ago
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

whelp_24 3y ago

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

m463 3y ago
Hmmm.. I would guess decoupling would allow variable ratio steering based on speed or preferences.
- lttlrck 3y ago
  Electric power assist already allows that. It's probably cheaper, lighter and maybe safer to have no steering column.
  - m463 3y ago
    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_steering https://en.wikipedia.org/wiki/Steering_ratio#Variable-ratio_...
  - yreg 3y ago
    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.
  - giobox 3y ago
    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.

geoelectric 3y ago

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.

oittaa 3y ago
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.

theluketaylor 3y ago

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.

solidr53 3y ago
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.
- bdamm 3y ago
  Wait, what? They're converting their motor controller into a 400V-800V DC transformer?
  - p1mrx 3y ago
    yes: https://www.youtube.com/watch?v=xm4VDfdKUm0&t=719s https://www.youtube.com/watch?v=xm4VDfdKUm0&t=719s
- theluketaylor 3y ago
  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.
stetrain 3y ago
Cybertruck does the pack splitting thing like the GM Ultium vehicles: https://twitter.com/baglino/status/1730391023799386480 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.

kllrnohj 3y ago

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-by-wire-yoke-review/ https://www.motortrend.com/reviews/2023-lexus-rz450e-steer-b...

alistairSH 3y ago
What happens if the engine shuts off (and the alternator stops spinning)? Does the steering just run off the 12v system?
- vostrocity 3y ago
  That Lexus is an EV.
- jimt1234 3y ago
  Engineering Explained did a really good video about this: https://www.youtube.com/watch?v=agMrewRJTow https://www.youtube.com/watch?v=agMrewRJTow
  - dkjaudyeqooe 3y ago
    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.
    - HPsquared 3y ago
      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.

demondemidi 3y ago

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?

boznz 3y ago
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.
- boznz 3y ago
  Just a thought but if its 48V and its ethernet I wonder if it is using power over ethernet?
  - grecy 3y ago
    I'm sure we will learn stuff like this when Munro tears one apart
- jaggederest 3y ago
  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)
  - kalleboo 3y ago
    Although if you wanted to, I don't see why you couldn't build a CAN switch?
    - jaggederest 3y ago
      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.

bilsbie 3y ago

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.

HeyLaughingBoy 3y ago
> Drive by wire kind of scares me Physical linkages break too.
- Yodel0914 3y ago
  Yeah, but not from a bad firmware update
  - geerlingguy 3y ago
    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?
  - kube-system 3y ago
    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.
    - Yodel0914 3y ago
      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.

KennyBlanken 3y ago

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

upbeat_general 3y ago

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?

agloe_dreams 3y ago
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.
- Projectiboga 3y ago
  I hope they learned from the Titan submersible to always go at least one step up from the cheapest game controller.
  - gopher_space 3y ago
    Their whole problem was custom hardware. I don't know why people were so hot on them reinventing tested commodity electronics.
    - rcxdude 3y ago
      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.
      - gopher_space 3y ago
        Seriously? Why would you do that?
      - 3y ago

darkclouds 3y ago

[dead]

arcticbull 3y ago

> ... 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.

liminalsunset 3y ago
> 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.
- 3y ago

ajwilson 3y ago

48V power electronics are quite commonplace in industrial applications.