Investor Engineering Discussions

[Buckminster]

Could Tesla go back to induction motors? These guys reckon they are above 95% efficiency particularly at high speeds:
Mahle's cheap, highly efficient new EV motor uses no magnets
Add a bit of carbon fibre (maybe not high tension wrap?) and Tesla could get to 96%+.

 

[mongo]

Could Tesla go back to induction motors? These guys reckon they are above 95% efficiency particularly at high speeds:
Mahle's cheap, highly efficient new EV motor uses no magnets
Add a bit of carbon fibre (maybe not high tension wrap?) and Tesla could get to 96%+.

Rotor losses and heating are still an issue. Along with low speed inefficiency and less precise control.

 

[Buckminster]

https://twitter.com/x/status/1657189367536820231

 

[JRP3]

Do not use Ohmmu Lithium 12V battery in older Teslas and don't try to charge the new 12/16V NMC battery out of the vehicle if it goes dead.

 

[MC3OZ]

['Recording an eye watering 345 mph, the Accel Plane is officially the fastest electric vehicle on the planet! Before it retires to the Science Museum, Imogen and Robert wanted to meet the team from Rolls Royce, Evolito, Yasa and Electroflight to find out how they've made electric flight take off! From axial flux motors, advanced cooling and packing staggering amounts of torque - this plane is choc-a-bloc with astounding technologies that are paving the way to mainstream electric aviation.']

Daimler has bought the company that makes axial flux motors and will be using them in Benz cars.

So for electric motors a lot of useful innovation is happening.,

Axial flux motors can be half the weight while delivering the same power output, and more torque.

 

[Doggydogworld]

['Recording an eye watering 345 mph, the Accel Plane is officially the fastest electric vehicle on the planet! Before it retires to the Science Museum, Imogen and Robert wanted to meet the team from Rolls Royce, Evolito, Yasa and Electroflight to find out how they've made electric flight take off! From axial flux motors, advanced cooling and packing staggering amounts of torque - this plane is choc-a-bloc with astounding technologies that are paving the way to mainstream electric aviation.']

Daimler has bought the company that makes axial flux motors and will be using them in Benz cars.

So for electric motors a lot of useful innovation is happening.,

Axial flux motors can be half the weight while delivering the same power output, and more torque.

Gotta go faster than 345 mph to be the fastest.

Here's How The Revolt Systems Streamliner Set The Electric Land Speed Record At Bonneville

The team ran 357 miles per hour to become the fastest EV in the world

jalopnik.com

 

[JRP3]

I'd like to see Tesla build remote charge cable eject into the vehicles. This video for EVJect shows a few scenarios where this would be quite useful. I expect to see more criminals stalking at charging sites as time goes on.

 

[dbeiler]

To whoever needs to hear this:

CAN and LIN are two different communication protocols. The terms CAN and LIN are not interchangeable. We use each term SPECIFICALLY to DISTINGUISH one from the other.

POE (power-over-ethernet) does NOT reduce the total wire conductor required for a circuit. We do not send power and data over the same wire simultaneously. We send them through the same CABLE which contains multiple, separately insulated wires in it.

A 48V wiring architecture does not affect the entire wiring harness. It only affects the sections of harness used to carry current to high current draw things like seat motors, door locks, fans, etc. Communication systems operate on 5 volts or less.

 

[mongo]

POE (power-over-ethernet) does NOT reduce the total wire conductor required for a circuit. We do not send power and data over the same wire simultaneously. We send them through the same CABLE which contains multiple, separately insulated wires in it.

Yes, it does, yes they do
10/100 Ethernet + power , 6 wires
1 pair TX
1 pair RX
48V
48V_RTN

PoE, 4 wires:
1 pair RX, center tapped +48V
1 pair TX, center tapped 48V_RTN

Automotive CAN, 3+ wires:
1 pair CAN (high speed)
1 12V (per load)
Chassis return

Automotive Ethernet, 2 wires in a remote node architecture:
1 pair TX/RX, center tapped +48V (shared)
Chassis ground return

 

[dbeiler]

Automotive Ethernet, 2 wires in a remote node architecture:
1 pair TX/RX, center tapped +48V (shared)
Chassis ground return

Interesting. I've never come across this in the industry. How does the module maintain a constant operating voltage if the wire voltage must be modulated to transfer signal? Through module capacitors?

 

[dbeiler]

Automotive CAN, 3+ wires:
1 pair CAN (high speed)
1 12V (per load)
Chassis return

Also, why is this listed? This is an example of data and power not being sent simultaneously on the same wire. Hence the need for the dedicated 12v wire.

 

[mongo]

Interesting. I've never come across this in the industry. How does the module maintain a constant operating voltage if the wire voltage must be modulated to transfer signal? Through module capacitors?

The data signaling on the pair is differential and the runs are not referenced. The power feeds are taken at the center taps of the transformers on both ends so it's constant. For PoE, one pair is biased at 48V and the other at 0.

Also, why is this listed? This is an example of data and power not being sent simultaneously on the same wire. Hence the need for the dedicated 12v wire.

I was listing the wire counts of combined vs non-combined methodologies for comparison. The 6 wire Ethernet and separate power was the base case for that protocol. CAN plus 12V was baseline for automotive.

 

[dbeiler]

I was listing the wire counts of combined vs non-combined methodologies for comparison.

Oh, I see that now. Pre center-tap total wire count vs post center-tap count. Was not previously aware of center-tap existence. Too late to edit my original post, unfortunately.

The center-tap method doesn't exactly seem to be free lunch though. That reduced wiring would require a slight increase in complexity and parts count inside each module. Also, would likely give up some fault detection and isolation. At some point the reduction in wiring may not even be justifiable.

 

[mongo]

Oh, I see that now. Pre center-tap total wire count vs post center-tap count. Was not previously aware of center-tap existence. Too late to edit my original post, unfortunately.

The center-tap method doesn't exactly seem to be free lunch though. That reduced wiring would require a slight increase in complexity and parts count inside each module. Also, would likely give up some fault detection and isolation. At some point the reduction in wiring may not even be justifiable.

Complexity of adding the center tap connection to the existing transformer is near zero. The twisted pairs are isolated to begin with.

 

[dbeiler]

I was referring more specifically to isolation from the operational circuit after a fault detection. Minimizing shutdown of otherwise functional areas of the circuit and maximizing system redundancy as compared to CAN bus.

 

[mongo]

I was referring more specifically to isolation from the operational circuit after a fault detection. Minimizing shutdown of otherwise functional areas of the circuit and maximizing system redundancy as compared to CAN bus.

LIN: loss of wire = loss of function
Fault tolerant low speed (125 kb) CAN: loss of wire = still works
High speed CAN: loss of wire = loss of function
Ethernet: loss of wire = loss of function
Power: loss of wire = loss of function

Tesla has had patent apps showing a ring topology where it can bypass a missing link via directional packet routing.

 

[dbeiler]

I looked over the Model 3 and Y diagrams to see where they might potentially replace CAN with ethernet. Quite disappointed to discover they have very low utilization of CAN or any modern communications to begin with. No wonder they're making a big deal about minimizing wiring. They have the highest number of analog wires I've seen on a modern car. The master window switch has analog wiring running all the way to a passenger compartment mounted controller. A separate wire for each individual switch in the master switch panel! Instead of using multiple separate door mounted and rear mounted CAN and LIN modules as is the norm for euro vehicles nowadays, they instead run all vehicle wiring all the way up to three controllers under the dash. My God, why?! Need power to the left tail light? Let's run a wire all the way up to the dash. Need power to the right brake light? Let's run a wire all the way up to the dash. Need to know hatch power strut position? Let's run a wire all the way up to the dash. Let's just run a ton of wiring from front to back instead of integrating a rear module.

Coincidentally, a design with dumb/analog wiring architecture must have been an absolute Godsend during the chip shortage. Can't be held back by a door module shortage if your design uses a handful of analog wires instead of a door module.

On the bright side, if they've been making money using 20th century wiring systems, they should be able to make even more using 21st century wiring.

 

[Oil4AsphaultOnly]

I looked over the Model 3 and Y diagrams to see where they might potentially replace CAN with ethernet. Quite disappointed to discover they have very low utilization of CAN or any modern communications to begin with. No wonder they're making a big deal about minimizing wiring. They have the highest number of analog wires I've seen on a modern car. The master window switch has analog wiring running all the way to a passenger compartment mounted controller. A separate wire for each individual switch in the master switch panel! Instead of using multiple separate door mounted and rear mounted CAN and LIN modules as is the norm for euro vehicles nowadays, they instead run all vehicle wiring all the way up to three controllers under the dash. My God, why?! Need power to the left tail light? Let's run a wire all the way up to the dash. Need power to the right brake light? Let's run a wire all the way up to the dash. Need to know hatch power strut position? Let's run a wire all the way up to the dash. Let's just run a ton of wiring from front to back instead of integrating a rear module.

Coincidentally, a design with dumb/analog wiring architecture must have been an absolute Godsend during the chip shortage. Can't be held back by a door module shortage if your design uses a handful of analog wires instead of a door module.

On the bright side, if they've been making money using 20th century wiring systems, they should be able to make even more using 21st century wiring.

As you've noted, they're less dependent on suppliers with this arrangement. IIRC, they actually saw it as a positive to need fewer chips. So although it goes against "the norm", it seems to have served them well.

 

[bkp_duke]

Analog? Tesla's already utilize a lot of ethernet in their systems. It's been the godsend of those that have wanted to root into the car (i.e. salvage cars and aftermarket support).

 

[dbeiler]

As you've noted, they're less dependent on suppliers with this arrangement.

Yeah, certainly explains their interest in designing and possibly producing modules in house going forward. Especially since currently the main controllers are made in Tiawan and the Tiawan/China situation is what it is. Not that Elon's latest "Tiawan integration" comments have been helping any.

Analog? Tesla's already utilize a lot of ethernet in their systems. It's been the godsend of those that have wanted to root into the car (i.e. salvage cars and aftermarket support).

Where? To connect the underdash controllers?