Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Speculation that the Model X could be more modular.

This site may earn commission on affiliate links.
JohnSnowNW

JohnSnowNW

Active Member
Feb 13, 2015
2,731
3,190
Minnesota
#1
The Model S is already modular in many ways, but one way it is not is with regard to the main wiring harness. Well, that's my understanding anyway, and is one of the reasons that adding auto-pilot to pre-auto-pilot vehicles is prohibitively expensive. This lack of modulation in the "mains" of the vehicle are one of the reasons we are considering leasing the X. I would prefer not to purchase one of the early models, only to find out 2-3 years down the road that we can't add on some awesome new feature because the vehicle isn't "upgrade-able." Of course, this is simply one of the prices of early adoption, but what if it didn't have to be...

Perhaps one of the features that is being kept under-wraps is greater modulation. So, if new sensors, or a new advancement, that would be difficult to implement in a traditional wiring harness design would no longer be as great of an issue. This would mean that you wouldn't see such drastic depreciation when new features are released/added to newer models.

I'm not that familiar with modular systems, and the ways it may be difficult to implement. However, if you consider that these cars are more similar to rolling computers, then I think it makes sense to make them as modular as possible. Because we all know what happens to the value/usability of our phones after only a couple years. And one of the reasons Google has been working on a modular phone design.

Anyway, just a musing I had, and am interested to hear thoughts on the idea. If this has already been speculated, I apologize, but a quick search of the forum didn't reveal this specific suggestion.
 
#2
JohnSnowNW said:
...
Perhaps one of the features that is being kept under-wraps is greater modulation. So, if new sensors, or a new advancement, that would be difficult to implement in a traditional wiring harness design would no longer be as great of an issue. This would mean that you wouldn't see such drastic depreciation when new features are released/added to newer models.
...
Click to expand...

"greater modulation" is not a tangible feature that Tesla would be advertising. The implementation of the internals of the vehicle are only of interest to a very small minority of the population. Examples of features that Tesla would advertise are: Self-driving (auto-pilot), Self-parking, tow hitch, ​17" display, etc.
 
#3
vandacca said:
"greater modulation" is not a tangible feature that Tesla would be advertising. The implementation of the internals of the vehicle are only of interest to a very small minority of the population. Examples of features that Tesla would advertise are: Self-driving (auto-pilot), Self-parking, tow hitch, ​17" display, etc.
Click to expand...

I think it's "tangible" in the sense that it may interest people who may have concerns about spending money on a vehicle that could potentially be obsolete within a few years. Well, not obsolete in the sense that it will still be a functioning car...but auto-pilot vs non-auto pilot.

As I mentioned, the Google modular phone comes to mind. Its entire selling point is modulation.

Granted, I could be wrong about people caring whether the vehicle can be retrofitted in the future.
 
Last edited:
#4
This could be an opportunity for the digital revolution...

A traditional wiring harness is a bundle of lots of individual wires each carrying a voltage or resistance to either make something happen at the distributed ends or convey information back to the central computer.

Most current cars use a mixture of traditional wiring and CANBus digital networking. Under CANBus, much of the communication becomes digital computer data instead of resistances or voltage levels - in some cases including command signals from the central computers to the devices at the perimeter (I've seen cars that appear to use a CANBus message to direct the rear light cluster to flash the turn signal.)

In principle, you could build a car that's almost completely digital - that uses a set of common parallel 12V power lines to everything in a region of the car and a second digital bus (possibly with CANBus protocols, possibly not,) to carry all of the data in both directions. This car could possibly use the same set of four wires (even two wires, if you use communication on power line protocols) to hook up all of the pieces in a door or in the entire rear of the car - but that has tradeoffs...

I'm suggesting it as a second bus because you wouldn't want a minor problem with one of a thousand devices - the left mirror fold, say - to kill the bus and keep the car from operating. CANBus can be killed by something shorting the two wires it goes over together, most others have a similar simple failure possibility. A single module that fails in a fashion that disables the bus will disable every device on that bus segment.

The 12V parallel power lines would mean distributed fusing (most likely internal to each module) if you don't want a single module problem to take a large piece of the car down - and a single wiring problem with the parallel lines will still kill a larger piece of the car.

It's entirely possible, I'm not sure if it is worth it overall. You reduce wire count and enable a lot more upgrading in exchange for needing unique network-enabled versions of all the pieces (with more computer hardware in them) and potentially worse failure modes (but also potentially more BIT - if the network modules can be bus powered, then they can report a blown fuse in their module over the bus as well as other failures.) More cost at the component level, simpler and cheaper assembly with cheaper wiring and easier troubleshooting/repair.
Walter
 
#5
Saghost said:
This could be an opportunity for the digital revolution...

A traditional wiring harness is a bundle of lots of individual wires each carrying a voltage or resistance to either make something happen at the distributed ends or convey information back to the central computer.

Most current cars use a mixture of traditional wiring and CANBus digital networking. Under CANBus, much of the communication becomes digital computer data instead of resistances or voltage levels - in some cases including command signals from the central computers to the devices at the perimeter (I've seen cars that appear to use a CANBus message to direct the rear light cluster to flash the turn signal.)

In principle, you could build a car that's almost completely digital - that uses a set of common parallel 12V power lines to everything in a region of the car and a second digital bus (possibly with CANBus protocols, possibly not,) to carry all of the data in both directions. This car could possibly use the same set of four wires (even two wires, if you use communication on power line protocols) to hook up all of the pieces in a door or in the entire rear of the car - but that has tradeoffs...

I'm suggesting it as a second bus because you wouldn't want a minor problem with one of a thousand devices - the left mirror fold, say - to kill the bus and keep the car from operating. CANBus can be killed by something shorting the two wires it goes over together, most others have a similar simple failure possibility. A single module that fails in a fashion that disables the bus will disable every device on that bus segment.

The 12V parallel power lines would mean distributed fusing (most likely internal to each module) if you don't want a single module problem to take a large piece of the car down - and a single wiring problem with the parallel lines will still kill a larger piece of the car.

It's entirely possible, I'm not sure if it is worth it overall. You reduce wire count and enable a lot more upgrading in exchange for needing unique network-enabled versions of all the pieces (with more computer hardware in them) and potentially worse failure modes (but also potentially more BIT - if the network modules can be bus powered, then they can report a blown fuse in their module over the bus as well as other failures.) More cost at the component level, simpler and cheaper assembly with cheaper wiring and easier troubleshooting/repair.
Walter
Click to expand...

This is good analysis and you are listing the main pros and cons. I guess one way to think of it is how it's being done in other applications that we could liken to a car i.e. the need for very quick/instanteneous control and a high degree of reliability.
Aircraft: as far as I know commercial airliners still have a huge bundle of wires to control many different aspects of the controls and peripherals individually.
Fighter jets: I don't know enough about how they are designed, probably more computerized set-up?
Medical equipment such as intensive care respirators etc: a high level of IC-control nowadays, very little hardwiring.
Space craft: Not sure, I would be thrilled to learn more about how SpaceX are planning on designing their Dragon capsule with regards to these things.
 
Last edited:
#6
Johan said:
This is good analysis and you are listen the main pros and cons. I guess one way to think of it is how it's being done in other applications that we could liken to a car i.e. the need for very quick/instanteneous control and a high degree of reliability.
Aircraft: as far as I know commercial airliners still have a huge bundle of wires to control many different aspects of the controls and peripherals individually.
Fighter jets: I don't know enough about how they are designed, probably more computerized set-up?
Medical equipment such as intensive care respirators etc: a high level of IC-control nowadays, very little hardwiring.
Space craft: Not sure, I would be thrilled to learn more about how SpaceX are planning on designing their Dragon capsule with regards to these things.
Click to expand...

The only one of these I can speak with any authority on is military aircraft. The programs I'm familiar with are still using a hybrid approach - two or three digital busses (one for primary flight controls and critical avionics, one for other avionics, sometimes a third for other mission equipment) and a bunch of individually routed wiring to all of the sensors, devices, and antennae. In some cases all of those individually routed bits go to a digital interface unit that converts the voltage/resistance into a digital form and puts it onto on of the digital networks.

But those programs also started decades ago, or are upgrades from programs that did. I'm not sure how a recent clean sheet like JSF works.
Walter
 
#7
We will likely see a gradual move towards complete digitalization more or less, but it will take time. With telecommunications this has taken decades to happen. I think some forms of telecom even today, that are highly dependent on security and reliability, still use analog hardwired signaling.
 
#8
In general, hardware modularity comes at additional cost, complexity, and size (you need additional space for more connectors, module disconnect hardware like rails, etc.). You also limit future designs to the form factors of your previoius modules. The trend is the exact opposite in computers, with tighter integration and less access becoming commonplace. The benefits are better design and lower cost, the downside is less flexibility. It will be interesting to see which way Tesla goes, but I wouldn't bet on them forgoing upgrade profits in favor of updating old cars with new hardware.
 
#9
Johan said:
We will likely see a gradual move towards complete digitalization more or less, but it will take time. With telecommunications this has taken decades to happen. I think some forms of telecom even today, that are highly dependent on security and reliability, still use analog hardwired signaling.
Click to expand...

Seems reasonable. In principle, I love the idea of (nearly) every box having just two, two wire connectors you plug into it (one upstream to the computer, one downstream for the next module to plug into,) and a single set of wiring running from one to the next. I especially love the idea of the central computer being able to tell you exactly what is wrong and even where any wire breaks are (at least between which modules to look, one short span to check,) so there's no need for complicated troubleshooting.
Walter
 
#11
Stoneymonster said:
In general, hardware modularity comes at additional cost, complexity, and size (you need additional space for more connectors, module disconnect hardware like rails, etc.). You also limit future designs to the form factors of your previoius modules. The trend is the exact opposite in computers, with tighter integration and less access becoming commonplace. The benefits are better design and lower cost, the downside is less flexibility. It will be interesting to see which way Tesla goes, but I wouldn't bet on them forgoing upgrade profits in favor of updating old cars with new hardware.
Click to expand...

I don't know if that's fair. The trend I see in computers is towards tightly designed boxes with minimal internal upgradeability, true. But it's also a trend toward more externalization on a few standardized buses. More and more things are getting put onto eSATA, USB 3, Thunderbolt or even WiFi as external accessories to the computers that no longer have them built in or space to add things.

There's definitely an art to deciding which interfaces are worth standardizing and when it is better to break the standard for more efficiency or better performance. It's all Systems Engineering - but a lot of engineering is as much art as science, even today. :)
Walter
 
#12
I sure don't hope that when I turn the steering wheel or push the brakes in my future Tesla that the signal is sent via USB :)

By the way, before Tesla kind of closed the access through firmware upgrades and directly contacting people kindly asking them to stop hacking the car, it became clear that many things are routed through and internal network, over IP-protocol/ethernet. Heres what the connector looks like:
20140302_174732.jpg


Successful connection on the Model S internal Ethernet network

They got this far, granted it's only access to the 17" screen but anyway:

3.jpg
 
Last edited:
#13
Saghost said:
I don't know if that's fair. The trend I see in computers is towards tightly designed boxes with minimal internal upgradeability, true. But it's also a trend toward more externalization on a few standardized buses. More and more things are getting put onto eSATA, USB 3, Thunderbolt or even WiFi as external accessories to the computers that no longer have them built in or space to add things.

There's definitely an art to deciding which interfaces are worth standardizing and when it is better to break the standard for more efficiency or better performance. It's all Systems Engineering - but a lot of engineering is as much art as science, even today. :)
Walter
Click to expand...

Oh I agree. I was mostly referring to things like removeable batteries, RAM, storage going away. Batteries in particular is one area where going away from user serviceable parts means allowing more efficient use of internal space and thus longer battery life in smaller devices.
 
#14
Stoneymonster said:
Oh I agree. I was mostly referring to things like removeable batteries, RAM, storage going away. Batteries in particular is one area where going away from user serviceable parts means allowing more efficient use of internal space and thus longer battery life in smaller devices.
Click to expand...

Fair enough. In the context of this discussion, though, I think the increased emphasis on external boxes on these standardized interfaces is actually pretty similar to what the OP is suggesting for the car of the future - the "world of devices" Microsoft was talking about (hopefully not all running Windows!!!) where all the pieces have a little bit of computer to them and talk to each other over simple networking rather than using customized wiring and unique voltage/resistance based signaling of past cars.
Walter
 
#15
Saghost said:
Fair enough. In the context of this discussion, though, I think the increased emphasis on external boxes on these standardized interfaces is actually pretty similar to what the OP is suggesting for the car of the future - the "world of devices" Microsoft was talking about (hopefully not all running Windows!!!) where all the pieces have a little bit of computer to them and talk to each other over simple networking rather than using customized wiring and unique voltage/resistance based signaling of past cars.
Walter
Click to expand...

We're seeing this already in the home automation world. I just this week bought a product that can control light switches, dimmers, etc. and can recive info from a wide range of sensors, all on the 433.92MHz wireless bandwidth. The list of manufacturers complying to this standard is growing every day. I think wireless is a long time away though in a car, at least for important functions. It makes little sense since you'll have to run electrical wiring anyway to any peripherals and lights.

On a tangent, in my house I'm using a homeplug system to run ethernet over the powerlines and I can routinely and easily get speeds of several MBit/second over my LAN without the need for any network cabling. So the technology is mature.
 
#16
Johan said:
We're seeing this already in the home automation world. I just this week bought a product that can control light switches, dimmers, etc. and can recive info from a wide range of sensors, all on the 433.92MHz wireless bandwidth. The list of manufacturers complying to this standard is growing every day. I think wireless is a long time away though in a car, at least for important functions. It makes little sense since you'll have to run electrical wiring anyway to any peripherals and lights.

On a tangent, in my house I'm using a homeplug system to run ethernet over the powerlines and I can routinely and easily get speeds of several MBit/second over my LAN without the need for any network cabling. So the technology is mature.
Click to expand...

Yeah, I'm not thinking wireless is that likely because you still need power, as you said.

I think if I were doing it I'd make a bunch of two wire, communicate on power daisychains. Each box would have a plug for a standard wiring size, then a fuse sized to match that wiring covering both the module and the second plug it has for the downstream device (ideally, there'd be just enough electronics upstream of the fuse to know when the fuse pops and pass that message upstream when polled.)

This imposes a limit on the combined use of everything on the chain, but there's no reason that the high draw in cabin things like heated seats can't be on a higher power version of the bus. The wiring is heavier than any one device needs, but likely lighter than wiring each device individually because it has much less overall length (and much lighter than an analog wiring system.)

Each chain would have a box near the heart of the car that sits on the CANBus (or TCP/IP bus) and connects that chain to the rest of the car, as well as hooking to the 12V battery to power than chain.
Walter
 
#18
Johan said:
Space craft: Not sure, I would be thrilled to learn more about how SpaceX are planning on designing their Dragon capsule with regards to these things.
Click to expand...

Ive only been given a cursory look at dragon so I don't know its configuration, but id guess its probably a mix of can and traditional harnessing.

In the commercial world, heritage is traditionally one of if not the most important factor in system design. As such, the big/expensive spacecraft can have many hundreds of pounds of DC harnessing running to a thousand or more connectors (mostly d-sub) in bundles upwards of 3" in diameter (usually 26-28ish gauge wires). That's not to say CAN isn't on the horizon...its just a matter of convincing the very conservative customer and insurance base.

The smaller/newer/less expensive spacecraft that are popping out of the startup side of the industry are starting to mix can and old school.
 
#19
Interesting thread, however unfortunately my understanding of comments from Tesla is that they are moving away from retrofittable modularity, which they were pretty good about originally (when the fleet was small and it was easier) not towards it.

As for car buses, I haven't researched which ones Model S (let alone expectations for Model X) uses, but there are more than the old dual-copper CAN, like the older optical MOST, Flexray, whatnot...
 
You must log in or register to reply here.

Similar threads

S
  • Article
X Owner test drives new Model X Dual-Motor - Jan2024
Replies
23
Views
4K
Model X
JAG
J
C
Considering Model X purchase
Replies
84
Views
3K
Model X
JPFLX
J
A
2022 Model X Plaid Rear Driver Side Suspension issue 26K miles
Replies
1
Views
636
Model X
ATLsBest
A
J
What Would Compel You To Upgrade to a Future Model X?
Replies
63
Views
2K
Model X
thyname
thyname
E
Undecided...
Replies
10
Views
1K
Model X
Unclepaul
U
Top
Back
Blog
New
Forum list
Marketplace
Menu