Going with 48V DC systems seems to me, as a telecom guy, to be the way to go, partly because the early switching gear from WeCo, others settled in on 48V as pretty much the standard, motor-generator sets in the basements cranked it out to ‘float’ the big wet cells, interconnected by 6 or 8 inch 1/4" or 1/2" thick stacked busbars. Many, many Amps during ‘busy hour’… Once solid state gear came along, less current needed, rectifiers replaced the old motor-generators… But, tons and tons of copper used, whether busbars or cabling…
So EVs need a lot of copper to carry their currents…
I really should sell off my copper stashes…
Note, for telecom’s -48V DC use, it was actually running '52V unloaded, additional gensets or rectifiers came up online as needed to hold the load, I recall a 35,000 Amp busy hour load back wheat here were multiple switches, still in the electromagnetic days… Basement stayed pretty toasty with all that work going on… Backup AC power were a couple locomotive diesels in the basement, and later an additional turbine up on the roof, its advantage was that it could take up the full load in about 8 seconds…
I suspect there won’t be - though I don’t know if technical issues are the reason why. Rather, I think it’s the business model that may drive it.
Given where the major players are, it is increasingly likely that the initial phase of adoption of AV transportation will be Level 4 fleets operating as robo-taxis. These will be largely commercial operators running TaaS operations in urban centers, as Cruise and Waymo are currently doing. They’re already running driverless taxis now in a few cities. They’re strictly geo-fenced, but you’ll see them expanding their service area incrementally - both within those cities, and by adding new cities. In the coming years we’ll see those operations get really established, and they’ll rack up the miles of actual Level 4 autonomy to convince regulators that these things are safe.
As that happens, though, these companies are going to be living and breathing with governmental partners. At first it will be the regulators - state licensing and permitting to allow driverless cars. But I think an important element of the business model of these companies will also be to partner up with government to provide transportation services in place of coverage transit routes.
All of that is easier if you’re green. Uber broke the political power of cab companies in most cities. Transit unions see autonomy coming down the track (as it were), and they’re going to be pushing back hard against the idea of using AV’s as a substitute (or even a subsidized complement) to existing transit service for low-income folks. So at the end of the day, being emission free is going to be an important selling point.
Let’s do some math. And thanks to wecoguy, we’ve got the formula we need (and which I can never seem to remember correctly). Watts / volts = amps.
4000 watts (the high end of the range you quoted) divided by 12 volts is 333 amps. I’ll admit that’s a lot for a car. Most cars have alternators that put out around 150 - 180 amps. Some - including mine - have options for higher amperage alternators. In my case, the highest option is 230 amps. That still leaves us short, especially considering the car needs a fair number of those amps for it’s own use.
But my car - really a commercial van adapted to passenger use - has provision for an optional second alternator. From the factory. Granted, when you choose that option, you get two 230 amp alternators. But that gets us 460 amps total, which is what we need. And a quick search of the internet shows alternators that can output 350 or more amps on their own.
Why do these large alternators exist (and in overall designs already engineered for current automobiles)? Audio systems. Big, honkin’, ear-drum-bustin’ sound systems need a lot of power. Strangely enough, power in the same ballpark as this. (Or at least at the lower end of the range mentioned for automation.)
I’ll grant that there is some engineering benefit to stepping up the voltage for this kind of power need to 24 volts rather than 12 volts. That cuts the amps in half. And amps are the thing that melts electric wiring. So moving 4 kW of electric power around a vehicle is easier done at 24 volts than 12 volts. But even at that, 24 volts is just two ordinary car batteries connected in series. Not a difficult problem at all. And there’s some argument to keeping that electric system separate from the native vehicle system. Heck, most of today’s fully electric vehicles also tote around a boring 12 volt lead-acid car battery to run all of the ordinary car electronics (lights, windows, seats, HVAC fans). So a 12 volt system for the basic vehicle plus a higher voltage system for the automation isn’t at all unreasonable. (And some of those high output alternators were 24 volt*** alternators.)
So my non-engineer engineering solution for ICE cars is to use a second, high output alternator for the vehicle automation, along with a separate battery (or two). That should provide adequate power to run all of the systems a level 5 car would need.
As to the added drain on the vehicles motive power source - well that’s going to be the same whether the vehicle is powered by an electric motor or a bunch of dead dinosaurs. 4kW of electric power is about 5.5 horsepower. Barely noticeable in most ICE cars today. And still exactly the same power draw as would happen to an EV’s main battery (less a bit of losses in converting the power from spinning stuff to electric stuff).
My bottom line: If an automation system will work on an electric vehicle, it will work on an ICE vehicle.
–Peter
*** Note - cars are really running closer to 14 volts than 12 volts. Old fashioned lead acid batteries are at 13.2 volts (give or take) with no load on them. It takes a voltage higher than that to charge the battery. A car alternator puts out about 14 volts so the battery will charge properly. So the whole electric system in an ICE car really runs pretty close to 14 volts with the engine on. Anything that operates without the engine running (security and remote entry systems) need to be OK with 13 volts or less on a routine basis. That high output 24 volt alternator was really a 28 volt output, which I simplified for discussion. It also gives a bit of a fudge factor to the numbers, as we’d need a few more amps at 12 volts than 14. So calling a car 12 volts is just fine for this discussion.
Most ICE cars don’t even have a plug. And many people who drive ICE vehicles don’t readily have a place to plug it in. Remember, one of the challenges of EVs is installing/finding a place to plug it in. And what you are saying here is part of the argument against ICE vehicles … because it is essentially saying that in order to make an ICE autonomous, you need some core changes to it, like adding a power supply with the AC to DC converters and all, and like perhaps changing from 12V to 48V (and more likely adding a side-by-side 48V system), and adding all sorts of other smaller things to support the new features.
Meanwhile all EVs have almost all of that built-in due to their very nature.
Please understand that I am NOT saying that ICE can’t be autonomous, because it can. Instead I am showing that some of the types of arguments against EVs apply in the argument against autonomous ICEs.
As far as the 4kW number, I am not sure exactly how accurate it is, but I do know that if I drive in my Tesla for an hour, and for 60 miles, that it uses a total of about 16.5kWh. I am not convinced that it is using 4kWh of that for sensors and processors. I suspect it is using far less than that.
Great post, on a number of levels. A friend in the airbag business told me the automakers have wanted to move to 48v architecture for at least two decades. Obviously, there are substantial barriers to doing that. Maybe self-driving will be the thing that pushes it over the edge.
Tough to beat. Our recent trip to Seattle proved the convenience to us, as we wandered about, needed a ride, a quick App request for an XL ride (I’m big n tall) and an SUV was at the curb in a few minutes, done deal, I tipped in cash, on to the next gallery, or whatever we needed…
This is a hold over, mostly because of common components from Tier 1 and 2 suppliers who still provide a very high quality low cost option for motors/switches/actuators and control gear at 12v are driving the bill of materials.
Moving up to 48v will continue to be limited by supplier base and will move in parallel with systems that expand from new models.
Don’t most computers, cameras, etc all work at 5v or even 3.3v?
Maybe some sensors run at 12v, but I doubt they run at 48v.
So whether the input is 12v or 48v they must have a regulator that converts to lower voltages.
Yep. There are occasionally some 12 volt bits as well. Your typical PC power supply takes the 120 volt AC from your wall and converts it to the lower voltage DC.
But the issue in a car is distribution. (That’s an issue for electric utilities as well, and for exactly the same reason.) I you are producing 300 amps at 12 volts under your car’s hood, but need that power in the back where the self driving computers are located, you need to run a big, fat cable the length of the car. That cable is expensive to buy and hard to route around car because it doesn’t bend easily.
If you double the voltage to 24 volts, you cut the amps in half to deliver the same amount of power. And that allows you to cut the size of the cable down significantly, which makes the cable less expensive and easier to bend to get around the car. Double again to 48 volts, and the amps halve again, and the wire gets smaller and cheaper and even easier to route.
At any rate, there will be converters to change the voltage from the distribution voltage to whatever the specific bit of computing equipment need for its internal use. So it won’t matter all that much if you are converting from 12 volts or 48 volts or any other voltage. Some form of converter will be needed. The basic distribution voltage is ultimately going to be influenced by both engineering and cost considerations. Converters between two common voltages are likely going to be less expensive that custom converters between two more arbitrary voltages.
I think car makers would normally route first to a power distribution panel (aka fuse box) and then to where it needs to go. I suppose they could have both a 48v and a 12v fuse box. The weight and complexity would increase, negating some of the gains going to 48v
We are considering 3 power and control ring circuits in our products. (Not cars, but a very close equivalent). The standard 12v, 48v for high power users and 300/600v for main propulsion equipment as installed. Our standard is to provide distribution (fuse) blocks for all 12v circuits, but any 48v or higher system has no such feature. Only inline fusable links with separate power interrupts as designed by the Tier 1 or Tier 2 system supplier.
Our approach has been to consider everything 12v as “conventional” and anything other as specialty. We started this with LED sub circuits at 5v. Even though the application and scope has changed, we adopted the same approach for all of these non-standard items.
Sure, but that’s not really the issue. I think we all agree that ICEs can be automated. What matters is whether there are significant advantages for automating an EV platform over an ICE one. I think it pretty obvious that there are. Response times for regulating an electronic drive train is much faster than for a mechanical one. Gathering electricity to run computers directly from a battery is simpler than getting that electricity from internal combustion. Those are big advantages.
One can probably build a high-def big screen TV from vacuum tubes, but why bother when it is more efficient and easier to do so using electronics.
Once the cost and performance of EVs match that of ICEs there will be no reason to automate ICEs. Not when it is more efficient and easier to automate EVs. The major car companies recognize this, which is why most self-driving technology is being developed in EVs.
There may be advantages, but I submit they are not substantial enough to matter. There is a huge installed base and momentum in the ICE market, and while EV’s are the future there is lots of money to be made in the present.
In 2021 there were 97 models of cars other than the three Teslas which offered some form of self driving, most of them gas powered. Acura, Ford Bronco, Audi, Mercedes, Kia, Maserati and lots more. Some offer totally hand-free (OK not totally) with lane following, stop and go, and other amenities, some have more including GPS assist, some have less. But once you have the servos and solenoids in place all you’re looking at is the processor (and software, obviously).
Not all are up to Tesla’s capabilities, but by many reports some are already superior. Here’s a list of those enabled, note this is from 2021:
Taking the second question first, modern ICE vehicles are already electronically controlled. Pressing on the accelerator is no longer a mechanical linkage to the engine under the hood. It is merely an input device to a computer. A human pressing on the accelerator is no different from a computer controlling that input device - that input to the control process.
And let’s be clear here, all other inputs to the steering and braking and turn signals and every other basic vehicle control are completely independent of the kind of motor the vehicle uses. So there is no difference between remotely steering an EV and an ICE. I’ll grant that there is a minor difference in braking - an electric motor can be used as a brake - regenerative braking. But they still have ordinary friction brakes as well, and any automated driving system will need to be able to use those brakes. If they can use the friction brakes on an EV, they can use the friction brakes on an ICE. Further, the logic in using regenerative braking in conjunction with friction braking is already built into the EV. That logic is going to be available to any self driving system. The self driving package only needs to decide to brake and how much - the existing EV determines how much braking action comes from each braking system.
Does any performance difference matter? Again, I’d argue if there is any difference, it is immaterial. If you wanted to get peak performance out of either type of engine (ICE or EV), a computer will be able to obtain that performance more consistently than any human can. Professional auto racing of many different types (Formula 1, NASCAR, Indy car, NHRA) have all banned certain types of driver aids over the years. They do this because they want their sports to involve the skill of the drivers and not just have the skills in the pits be the sole determinant of who wins.
Is there a delay in the time between the driver pressing on the accelerator and the car moving faster? Yes. But that is a function of design. A drag racer has as little delay as possible. And what delay might be there is used to maintain traction at the tires. If you had full power to the tires instantly, they would just spin in place rather than accelerate the car. Higher horsepower ICEs already have this issue, and have had it for decades. Power can be delivered to the tires faster than the tires can turn that power into acceleration. Traction control is standard in cars today. While we mostly think of it for low traction situations (rain, snow, mud, for example), it’s still used on good pavement in good weather to keep tires from spinning. Even EVs need to limit their power output to keep from spinning their tires.
This whole “response time” thing is just a red herring once you actually understand cars.
There are three functions that need to be controlled for automated driving, braking, steering, and torque. Torque is what makes the wheels go around and is the major difference between EVs and ICEs.
EVs have the capacity for instant torque. Electricity creates an electric field that produces rotational force directly to the wheels. Very few moving parts required.
ICEs create torque by mixing a regulated amount of air with fuel to produce combustion.
A series of such explosions generates reciprocating motions which is translated to rotational movements by the crankshaft. This rotational movement of the crankshaft is translated to the wheels via a set of variable sized gears called a transmission. Lots of moving parts. And because combustion causes lots of heat, even more moving parts are required, but we won’t go into that.
This means that when sensors send a signal to the AI driver of a hazard ahead that can be avoided by accelerating to the left, the EV can respond quicker than the ICE because the change in torque occurs almost instantaneously.
Furthermore, Google the number of moving parts in an EV drivetrain compared to an ICE. You’ll get various estimates that typically show EVs have about an order of magnitude fewer moving parts. In principle, which do you believe represents the greater engineering challenge: regulating a machine with lots of moving parts or one with few such parts?
Self-driving cars are essentially computers with wheels. I’m sure it is possible to build a laptop powered by an internal combustion engine. But batteries are good enough that there is no point to design an ICE laptop. Currently the only reason to build a robotic ICE is for range or costs. Once batteries become good enough to mitigate those concerns no one will bother with robotic ICEs. The advantages of the simpler EV platform with much fewer parts and the inherent capacity to deal with high voltages/current are just too great.
That’s no doubt true, but those aren’t the limitations to FSD. The limitation is the tech can’t safely digest the information it is being presented with.
Regular ICE vehicles respond fast enough in virtually all driving situations. The point of failure is almost never the car itself.
Your argument is that: therefore there won’t be self driving ICE cars. Yet there already are, so the argument falls.
Further, the differences are trivial in the grand scheme. “Self-driving” is not going to be the difference between someone buying an EV or an ICE since both already exist. The cost differential is deminimus. Regulators, and more importantly manufacturers who would bear liability for defective systems are already unconcerned about the differences. You are. Well, bully, but again, so what? It’s important to you I guess, but not to most anyone else.
It’s like you are arguing about whether OLED or QLEDs are the best TVs, and while there is a slight difference it’s immaterial. Both are TVs, both work fine, both are on the store shelves, and both have their relative advantages and disadvantages.
OMG!!! In spite of having rebuilt 4 or 5 engines in my lifetime (and not just replacing a part, but disassembling every nut and bolt and threaded fastener (apparently you can’t use the word “$crew” here - stupid TMF), inspecting and repairing or replacing every part, having the precise tolerance parts machined (I know what needs to be done, but lack the tools and skills for precise machine work) back to their proper tolerances, reassembling, and successfully running) I never knew how many parts there are in an engine or how they actually work!!
OMG!!! In spite of driving for 40+ years, I never noticed that the torque from the gasoline motor wasn’t delivered instantly! I can’t believe that all those times I’ve been in wrecks it was because I didn’t have instant torque available to me!! (Never mind the fact that none of them could have been prevented with instant torque and, in fact, 2 out of 3 were failures to properly brake - one by me and one by the guy who hit me while stopped at a stop light, with the third a truck merging into my lane trapping me between him and a barrier.)
Listen, I’m not trying to argue that EVs are inferior to ICEs. Frankly, they have lots of advantages. But none of those advantages relate to their ability to be controlled by automation. NONE. ZERO. NADA.
–Peter
PS - OK, perhaps the level of monitoring for failures is an advantage. When an ICE engine has a failure (and there are plenty of failure points to monitor - most of which are already monitored and will display a warning to the driver) it may need to stop to prevent additional damage or injuries to the occupants or bystanders. An EV has fewer failure points to monitor, reducing the load on any monitoring subsystem. But either way, the monitoring subsystem just needs to notify the main AI of a failure (and possibly a severity of the failure) for the main AI to be able to decide what to do. It doesn’t really affect the complexity of the main AI handling the driving task, just the complexity of the monitoring subsystem.
PPS - Motors produce both horsepower AND torque. They are somewhat related in a way that I never seem to recall well enough to explain to others, so I won’t try here. Focusing blindly on only one overlooks the need to manage and control the other.
PPPS - The situations where accelerating is the right choice to get out of trouble on the road are vanishingly small. Braking or turning (or a bit of both) is virtually always the better option. The most important reason this is true is that by braking you are taking kinetic (or is it potential?) energy out of the system. Wrecks at higher speeds have more kinetic/potential energy and can cause more damage to people and stuff. Wrecks at lower speeds have less kinetic/potential energy and therefore can cause less overall damage. A second reason is the interface between tires and road. That interface is how you control the car. Unless you have 4 wheel or all wheel drive (and before you say it: Yes, many - but not all - EVs do have all wheel drive), the power from the motor is only going to two wheels. But braking power is always applied to all 4 wheels. So every two wheel drive car can brake faster than it can accelerate. Always. ALWAYS!
Therefore, this whole “response time” thing is still a red herring.
Sarcasm notwithstanding, the OMG part of your post is your quaint belief that all this mechanical expertise matters when it comes to autonomous driving. Self-driving is an electronic and software issue, not a mechanical engineering one. Your ability to rebuild an ICE engine is about as relevant to autonomous driving as being able to shoe a horse. The fact that you consider rebuilding a gas car engine an accomplishment is why self-driving will be based on an electric drivetrain. Why bother with so many moving parts when there is a much simpler alternative that is much more suited for electronic control?
Faster response times provides a greater margin of error. It provides more time for AI to make decisions. Instant control of torque provides another way of controlling speed besides braking. Lots of advantages for an electric drivetrain. -
My argument is that there won’t be any self-driving ICE cars commercially produced because (for engineering reasons) these will not be competitive with self-driving EV cars. Companies that were developing self-driving on an ICE platform are all switching to an electric platform. Your examples of advanced cruise control are irrelevant because they aren’t close to what is required for full Level 4-5 self-driving. That’s like equating an abacus to a Mac Pro. As far as I can see, no company is planning an autonomous fully ICE vehicle for commercial use. Hybrids are the closest you get.
Which is why this argument is mostly moot. The money being spent in the real world on autonomous driving is almost exclusively with an electric platform. The people with actual expertise on the issue have committed to electric vehicles. The latest is Uber’s launch with Motional for robotaxi service in Las Vegas using the all-electric Ioniq. Uber Launches Robotaxis! - CleanTechnica