Cybertruck to use 48V instead of 12V?

[BigAl]

Why do I feel like I’m drowning in knowledge… someone throw me the rope of stupidity so I can get out of all this well informed nonsense ?

Sponsored

 

[JBee]

Heat Pump: Based on all the tear-downs of other vehicles, I feel extremely confident the heat pump will be powered by high voltage (400VDC - 1200VDC, wherever we land on that). I very much doubt they'll power the heat pump with low voltage (0-48VDC).

Hey haven't seen you around for a while.

I can agree that the heatpump could be at pack voltage, but it could also be at 48V because of safety/maintenance convenience. Some tests show that heat pump only consumes under 1kW whilst in use, which is only 20A or so at 48V. Even at max output, it's probably still only around 3-5kW, so similar to large split A/C, and will within 48V capability.

BTW the current DC-DC PCS converter in the M3 only currently uses 12V or 16V. I haven't found the heat-pump voltage with a quick search, but it might already use the higher 400V pack voltage in the current models. I don't know if that's a viable option for a 1000V pack version though.

But obviously there is an argument to have the heat pump at pack voltage to avoid having a large buck converter to get the required kWs at 48V. All the other 48V would consume much less energy than this, but might still have significant peak power load. Meaning that it might be unavoidable to have a large capacity pack to 48V converter for all the other devices (power steering and compressor come to mind), at which point it might NOT be substantially more expensive to add the heatpump load at 48V to the converter, and run them all from 48V, instead of an extra 1000V converter/motor etc.

At this point I'd like to mention that "if" we see a 1000v pack instead of 400V, there's a chance the 48V system might also run the 110/230V inverter for bed power. At 400V I'd expect them to tap the 400V and go straight to 110/230V inverter setup, but not necessarily from 1000V pack. They might just throw in a Powerwall inverter that runs off 48V instead to keep the part count down. Being bi-directional in the powerwall, it might also form the basis for the onboard charger in the CT, if the 48V converter is also by-directional. Not many parts, and all of them doing more than one thing which is the best way to reduce part count and cost.

Essentially the 48V system might become a bit of an internal bus to have all the non-drivetrain consumers in a mid tier power level setup. But without specific costings of the variations involved it is hard to tell, even though there are enough viable options.

Another even more powerful way to do this would be to use the motor controllers themselves as a charger/power inverter by switching the motors out of circuit. There are some that are developing this already, the cost savings are large, plus it offers high vehicle power level capacity V2G/V2G/V2V/V2X.

Of all these layouts, I think something using the pack voltage motor controllers and for charging (V2X), and the HVAC/230V output as well, and use 48V for all the other high power accessories, and 12V for control and the rest, would make most sense to me, with the least amount of converter capacity and steps, and resulting power losses.

The computer argument confuses me. Yes, computers use many different voltages internally. But the main source of the power is one voltage. Every computer in a car is powered by the 12V system. There isn't a separate voltage system for the 3.3V needs, the 1.5V needs, the 5V needs etc. Each computer has a power supply that creates the needed volts. Telsa just has to design their computers with power supplies that are powered by 48VDC instead of 12VDC. It's not rocket science. And even if it was, they have that resource available. I believe Tesla makes all the computers in their cars. They would arguably be the easiest thing for Tesla to change to 48VDC.

I don't have a problem with having a power supply (PSU) to supply the motherboard voltages, and then further motherboard onboard voltage regulators for the processors etc. The whole computer thing came up because some claimed 12V was NOT necessary at all and you could do it with 48V. The reality is you need to have much lower than 48V to run any electronics so 48V at these power levels has no benefit to reduce wiring whatsoever, given the proximity of the computing to the PCS in the fsame firewall, and might as well just come straight from the existing PCS at the voltage needed for the motherboard directly.

Simply they are already doing what you say from the pack, but don't need to do so via yet another converter from 48V, so they can say "everything runs from the 48V bus" like some think here.

Ethernet can do everything CAN can do, and do it faster, and with MUCH higher bandwidth. Ethernet does not have to use RJ-45, and will not in a car. (Other than a probable connection for a diagnostic device or Laptop). The CAN benefits like ring topology and redundancy also exist in Ethernet protocols. In the industrial world it's known as Device Level Ring (DLR). The industrial world has been moving away from CAN based protocols and toward Ethernet based protocols for quite some time. I couldn't find the quote, but I'm confident Elon at some point implied CAN will be replaced with Ethernet eventually.

https://www.hms-networks.com/news-a...rings-ethernet-ip-using-device-level-ring-dlr

https://www.electronicdesign.com/ma...whats-the-difference-between-can-and-ethernet

I'm not claiming that Ethernet doesn't have the ability as such, rather that it is not common architecture in low bandwidth automotive application MCU's, which in turn is why there aren't that many MCU's that support it. There simply no reason to have high bandwidth, and the associated cost in hardware, and sensitivity to interference, if the connection only needs a very low bandwidth anyway that can be done with CAN as well.

My point here is that the Ethernet does not need to be 48V to work as a comms interface, "just because" there are "existing" 48V (and a 24V) PoE standards.

That doesn't mean that's why they are using 48V, or are actually powering devices from said PoE.

Simply, it would make more sense if Tesla created it's own PoE standard, which in turn could use even less cables than the common 8 core LAN cables and provide higher levels of power as required to make 48V more useful. You could do the same as ethernet over power systems using just two cables for example, that are also the two cables that supply the power as well. That's a quarter of the wiring and connectors straight out.

Further, using 48V is NOT the only way to reduce cable mass. The "bus system" proposed by Tesla actually does more to reduce wiring than a switch to 48V, but obviously switching to 48V for some high powered devices helps as well as the bus.

But here everyone seems to think it's "ONLY" because of the switch to 48V, whereas it's not.

Cable architecture and in particular wire length, diameter and number all reduce cable use, not just a switch to a higher voltage. The change is simply more than just upping the potential to 48V.

This is the proposed bus from the Tesla patent:

 

[firsttruck]

I watched about 10minutes and didn't see aanything about 48V?

Although it was interesting on a personal level about Cory's fast food roots I didn't find anything scrolling through.

Sorry the cueing did not work in my original post. I think it is fixed now.

video cued to Cory Steuben's comments about Master Plan Part 3 & 48V. (timestamp 50:02)
[ Aug 29 edit : fix the cue point ]

 

[Crissa]

I remember a while ago you bought yourself a new Mac notebook.

Why does that have a 11.46V battery then?

You do know that CPU/GPU/MCU have to be sub 3V because of the chip wafer they use would be destroyed otherwise?
It would just melt into a blob at 48V.

The point is 48v doesn't help for the PC parts like you claim for multiple reasons.

So... you know that 11.46 nominal isn't 12 volts, right? It's 2 x the 5v. You can just halve it to get to a 5v nominal, which you can't do with automotive 12v.

And so you accept that all the voltage has to be stepped down... Then make the argument that if it weren't stepped down to 3v it would melt at 48v. Well, it would also melt at 12v.

If you have to transmit it through wires, higher voltage is better.

You are being so obtuse at being wrong.

-Crissa

 

[JBee]

So... you know that 11.46 nominal isn't 12 volts, right? It's 2 x the 5v. You can just halve it to get to a 5v nominal, which you can't do with automotive 12v.

And so you accept that all the voltage has to be stepped down... Then make the argument that if it weren't stepped down to 3v it would melt at 48v. Well, it would also melt at 12v.

If you have to transmit it through wires, higher voltage is better.

You are being so obtuse at being wrong.

-Crissa

You're a joker, but not worthy of rebuttal. You're sprouting nonsense again.
Enjoy your holiday... I know I did.

 

[CyberGus]

 

[firsttruck]

Musk: 12 volts, which is a pretty absurd number, really it's wrong for EVERYTHING.

 

[CyberGus]

Musk: 12 volts, which is a pretty absurd number, really it's wrong for EVERYTHING.

If 12 was good enough for Jesus, it's good enough for me

 

[JBee]

Well now you know:

So they have a distributed controller network, with each controller connecting to local endpoints, and then they use Ethernet to connect the individual controllers using what looks like a four core busbar cable:

From the picture it looks like the cable itself can be cut in such a way that it fits into the receiving end of the connecting device without the use of an extra connector. The cable end itself appears to be the connector. It would seam self evident that the receiving device then clamps down on the cable to seal it and hold it in place. The patent also say that all cables can carry current from source to load ends, implying that there are no dedicated comms cables, rather it would seem each cable can carry power and comms. Similar to a Ethernet over power device, the comms is modulated over the same cable that also provides power.

It also says it can handle both low and high voltages. It possibly might be able to do this at the same time seeing it has two pairs, which would offer some flexibility when routing.

Either way it's good to see a redesigned bus system to reduce the cable harness complexity, along with being more accommodating for the various necessary voltages.

 

[Baldey]

Musk: 12 volts, which is a pretty absurd number, really it's wrong for EVERYTHING.

The British imperial system really does suck. As does time. So does dividing by two, f that.

 

[Baldey]

No I'm not saying that all.

I'm saying the "bus architecture", that Tesla is proposing with their new wiring system, does more than changing it from 12V to 48V, to reduce cabling.

BTW my computer comments were in regards to Crissa saying that there is no 12V in computers, when in fact there is. I just used it to highlight that wire "resistance" is also determined by wire length. Meaning a shorter bus type system has less wires, even if it still is 12V.

Well, it kinda was what you were saying:

All of the semiconductors used for processing, be that MCU/CPU/GPU are all under 3V and typical off the shelf PC hardware is all 12V/5V. There's nothing that is 48V, and running a few hundred watts over 12V is easy and light enough that 48V won't make a difference for it.

The main items will be all the medium power consumption items, like HVAC compressor or accessories like winches etc. But I doubt we will see a 48V window or light, 12v works just fine for that as well.

But i guess your main point is:

Guys thanks for all the lectures, but I'm well versed in the difference a higher voltage makes on cable sizing.

My point was "not" if 48V was worth transitioning car electrical to, rather that even if you did 48V, not everything in the car could run off 48V, so some things wouldn't be connected to 48V at all.

For 48V to make a difference it would have to draw more than 10-20A, so would be a large appliance, say the size of the heat pump used for the HVAC.

But none have gotten rid of 12V, and neither will Tesla with the CT. It will still have 12V. It will just have 48V for the heavier energy consuming devices.

Which i think i disagree with... You seem to like to use a lot of words, which can confuse. Words bad. Im gonna try to use as few as possible to make my point. As bad as i am at it, ima try to be simple and concise here: ------------

Why not have a 48V bus that loops the car, and then have cheap buck converters on all of the low[er] voltage components? You could have your own automotive grade PoE: with a high speed network for camera data and maybe some signaling, and massive power lines capable of handling 10-20x or more current of standard PoE (30W max)

Or are you proposing having multiple loops for various voltages? That seems inefficient.. I can only think of lights running on 12v, here's a list off the top of my head of every thing else:

-gate actuator
-toneu cover actuator
-air compressor
-parking pawl
-electronic brakes
-rear wheel steering actuators
-STB redundancy systems
-power steering
-door actuators
-seat motors
-AC (Many Fans)
-high-end speakers
-screen
-main pack contactors
-differential power disconnect electromagnet coils
-fronk actuators
-BAW
-remote turret in the back
-PoE cameras
-many pumps
-Multiple computers (yes i know chips run on one volt, but their power delivery does not. Or at least i hope you don't send 100 amps through your outlets in Australia.. Would not put it past ya mate)

 

[TyPope]

This is one of those chicken and egg problems. The wires can be thinner saving copper and the transmission losses are lower with 48V than 12V.

The problem is there aren't many 48V automotive "things".

Well, Tesla can run a 48V architecture and use those converters where needed. Eventually, as time goes on and stock gets used up, Tesla can design more efficient motors (BAW, Windows, etc.) to eliminate the converters... the best part is no part and it you have to have a part, have a lighter part.

Thus, Tesla, by being so vertically integrated, can make the change to 48V architecture. Manufacturers will make parts for the new architecture because there are a ton of Teslas being made and it'll be cheaper for them to use less wire and a 48V architecture. Once they are making parts for Tesla, they can easily do the same for everyone else.

 

[Baldey]

Exactly. Being vertically integrated and designing what they need in-house is a huge advantage. Its like i said in another thread, legacy automakers are now little more than OEM part integrators. Not much innovation happens there, it can't. They've become complacent sad excuses for companies, that just keep throwing a new stamped and polished skin on the same pile of outdated crap. And it is frankly embarrassing that they once were the powerhouse of this economy. Engineers have wanted to transition for a while, but as always the sales people told them it was impossible. Well, enjoy catching up now.. I'm sure uncle sam will save you


The whole automotive industry kinda gives off that stench.. The resistance to change. You see it in facebook comments. Just keep pumping that oil, and meeting the status quo. Get better at burning that oil, but never consider anything better. Bean counters and pencil pushers, all of them. I guess it would take an idiot like Elon to shake things up, and i love him for it.

 

[firsttruck]

This is one of those chicken and egg problems. The wires can be thinner saving copper and the transmission losses are lower with 48V than 12V.

The problem is there aren't many 48V automotive "things".

Well, Tesla can run a 48V architecture and use those converters where needed. Eventually, as time goes on and stock gets used up, Tesla can design more efficient motors (BAW, Windows, etc.) to eliminate the converters... the best part is no part and it you have to have a part, have a lighter part.

Thus, Tesla, by being so vertically integrated, can make the change to 48V architecture. Manufacturers will make parts for the new architecture because there are a ton of Teslas being made and it'll be cheaper for them to use less wire and a 48V architecture. Once they are making parts for Tesla, they can easily do the same for everyone else.

Nobody forced the legacy auto to out source almost everything 50-60 years ago and have no way to innovate. Just being re-packagers was their choice.

Even after Tesla showed what was possible with today's tech the legacy autos mostly just sat on their hands and whined everything was too hard and customers would not really buy EVs.

 

[JBee]

Well, it kinda was what you were saying:


But i guess your main point is:


Which i think i disagree with... You seem to like to use a lot of words, which can confuse. Words bad. Im gonna try to use as few as possible to make my point. As bad as i am at it, ima try to be simple and concise here: ------------

Why not have a 48V bus that loops the car, and then have cheap buck converters on all of the low[er] voltage components? You could have your own automotive grade PoE: with a high speed network for camera data and maybe some signaling, and massive power lines capable of handling 10-20x or more current of standard PoE (30W max)

Or are you proposing having multiple loops for various voltages? That seems inefficient.. I can only think of lights running on 12v, here's a list off the top of my head of every thing else:

-gate actuator
-toneu cover actuator
-air compressor
-parking pawl
-electronic brakes
-rear wheel steering actuators
-STB redundancy systems
-power steering
-door actuators
-seat motors
-AC (Many Fans)
-high-end speakers
-screen
-main pack contactors
-differential power disconnect electromagnet coils
-fronk actuators
-BAW
-remote turret in the back
-PoE cameras
-many pumps
-Multiple computers (yes i know chips run on one volt, but their power delivery does not. Or at least i hope you don't send 100 amps through your outlets in Australia.. Would not put it past ya mate)

Thanks for the post and list.

There's actually a couple on there I have to add to mine....and made me think of Airbags as well

I think that patent video Cybergus posted confirmed enough of my assumptions so I could move on to another subject, but here goes a quick recap of my thought process so far.

BTW I'm interested in learning more about it, than just spewing a whole bunch of words, but I also often feel that many don't bother to explain in more detail what they mean to really engage, so I try to be better, which I agree at times can go the opposite way, especially if there is little to no feedback loop.

There might be some additional confusion if one perceives the argument to only be about changing to 48V from 12V as the thread title would suggest. In reality the "WHOLE" system architecture is changing, not just some of the voltages along the line, and these changes are more significant than the change to 48V by itself.

Anyhoo...back to the system architecture. Let me quote my response to Cybergus patent video first as a quick reminder:

So they have a distributed controller network, with each controller connecting to local endpoints, and then they use Ethernet to connect the individual controllers using what looks like a four core busbar cable:

It also says it can handle both low and high voltages. It possibly might be able to do this at the same time seeing it has two pairs, which would offer some flexibility when routing.

Now I can imagine from a voltage perspective this distributed setup would have a 48V bus voltage between the distributed controllers, which I believe are all the red nodules on the drawing above, but possibly only the blue line in the drawing. Then at each controller, you could have a few buck converters with enough capacity, and the right voltage to handle the endpoint loads. This way you can aggregate the buck converters into larger more cost effective units together with the controller, which then acts as a distribution hub for the endpoints, and powers and controls them. In some cases those controllers could also have built in motor controllers as well to drive the smaller things, like fans, windows or seat motors, at any voltage that makes sense for them.

This architecture does a few notable things:

1. It allows for a single high power cable to run as a bus at 48V along the full length of the vehicle. Which makes a worthwhile difference in cable dimensioning, in that it is long, and at 12V would need to be large to handle that amount of power. Probably in the 1-3kW range, unless they run the 110V inverter from it, in which case it would have to be 3x the size at that power level.
2. It acts as a spine, in that no "long" and "heavy" low voltage cables have to run back to a single central point, rather only back to the nearest bus and controller, from where the endpoint can be supplied with the appropriate voltage. This reduces losses on lower voltages under 48V by shortening the wire length for those, and at the same time aggregates the buck/motor controllers so that having a 48V bus is worth it by running at a higher load.
3. The controllers could be modular, in that you would possibly only need 1-3 different types, and could reuse them across the vehicle for different things. This would bring down the overall part count, without any loss of features.
4. It allows the gradual replacement of non-optimised OEM parts with dedicated Tesla parts, and still offers the flexibility to run endpoints that might need a specific voltage that is less than 48V.

Note these are all answers to comments I made against running "everything only" on 48V, and I think the above solution to divide and conquer makes for a better architecture overall, in that it allows all the things to run on the voltages it needs too, without forcing everything to change to 48V.

Overall I'm pretty happy with that, as well as spending some time finding out how they are doing it, and discussing the matter, with future projects of my own in mind. I learnt something.

Sponsored