News Tesla Model S with 10 TFLOPs Could Double as Gaming Rig

bigdragon

Distinguished
Oct 19, 2011
1,107
547
20,160
All that computing power is likely locked up in the Autopilot system and related visualizations. I doubt that you'll actually be able to play PS4 or PS5 quality games on the car's computer. Some people might enjoy that feature though...especially while waiting for a charge during a trip.

Personally, I hate the new steering wheel and the removal of the steering column stalks. Park/Drive/Neutral/Reverse selection and windshield wiper control will be very annoying and distracting. The windshield wiper controls on the 3 (and Y) are already very annoying. Looks like they'll be even worse on the new S. Otherwise, the new interior looks good and like Tesla is establishing a consistent design theme.
 

P.Amini

Reputable
Jan 20, 2021
21
24
4,515
As an engineer I love technology and all that fascinating stuff but that car lover inside me loves those pure mechanical gasoline cars. I hate to admit, but I hate electric cars and even all these electronic stuff in our cars...
 
  • Like
Reactions: Phaaze88

InvalidError

Titan
Moderator
I hate to admit, but I hate electric cars and even all these electronic stuff in our cars...
I don't mind cars going electric as long as 7-15kW L2 chargers become ubiquitous.

What does bother me is how tightly integrated and proprietary everything is. With a conventional car, you can retrofit almost any part to any car as long as you have the manufacturing capability to make appropriate modifications with grinder errors relatively undone by welding and vice-versa. If you want to transplant a Tesla drive train in something else though, you need to reuse the main computer, BCM and motor because they are locked to each other, much like how Apple disables features on its products when it detects that sensors, screens and other components got swapped even if they are part swaps from another otherwise identical device - even if you are able to source parts yourself, you cannot make them work. At least not without taking them apart and transplanting some chips.
 

jasonf2

Distinguished
I don't mind cars going electric as long as 7-15kW L2 chargers become ubiquitous.

What does bother me is how tightly integrated and proprietary everything is. With a conventional car, you can retrofit almost any part to any car as long as you have the manufacturing capability to make appropriate modifications with grinder errors relatively undone by welding and vice-versa. If you want to transplant a Tesla drive train in something else though, you need to reuse the main computer, BCM and motor because they are locked to each other, much like how Apple disables features on its products when it detects that sensors, screens and other components got swapped even if they are part swaps from another otherwise identical device - even if you are able to source parts yourself, you cannot make them work. At least not without taking them apart and transplanting some chips.
What you are describing is mainly because of the manufacturers in the game right now. With GM now committed to a 2035 timeline and all of the other majors starting to roll out products it is only a matter of time before the components start to OEM stabilize. Tesla has a technology lead because off the shelf really isn't "good enough" yet. When a significant output from the other manufactures start rolling out shared components will fuel the aftermarket game and it probably won't be much different from todays ICE machines. The exception being because of the torque characteristics and mechanical simplicity of the electric motor an all electric has the potential to be a significantly more elegant drivetrain. That being said upgrading a single performance component as a tuner is not likely to be an option because there won't be a lot to change.
 

InvalidError

Titan
Moderator
Tesla has a technology lead because off the shelf really isn't "good enough" yet.
Technology "not being good enough yet" does not prevent you from standardizing the inputs/outputs so that companies can still improve their individual implementation of the black box. A motor-drive unit has basically only four essential inputs: DC power, drive and regen power limits depending on battery condition and set-point for RPM or torque up to whatever the power limit will allow or design limits, whichever comes first, while non-mechanical outputs are just telemetry to monitor its state. You don't need to discover the most efficient electric motor design theoretically possible and technically feasible to decide on a standard interface for that. Same goes for coolant pumps and routing valves, you don't need to decide on the best pump and valve design ever before cooking up a standard for how to power and control them regardless of the physical implementation.
 

jasonf2

Distinguished
Technology "not being good enough yet" does not prevent you from standardizing the inputs/outputs so that companies can still improve their individual implementation of the black box. A motor-drive unit has basically only four essential inputs: DC power, drive and regen power limits depending on battery condition and set-point for RPM or torque up to whatever the power limit will allow or design limits, whichever comes first, while non-mechanical outputs are just telemetry to monitor its state. You don't need to discover the most efficient electric motor design theoretically possible and technically feasible to decide on a standard interface for that. Same goes for coolant pumps and routing valves, you don't need to decide on the best pump and valve design ever before cooking up a standard for how to power and control them regardless of the physical implementation.
I agree on the greater premise. But when GM decides to put an O2 sensor in its newest engine design the O2 sensor design is so standardized that completely reinventing the wheel and building their own new O2 sensor is counter productive. So in most cases while the sensor may have some packaging differences to make it a "GM" sensor it is at its core the same sensor that Ford is using in its vehicles, probably produced in the same factory by the same OEM outsource manufacturer. While at its core the O2 sensor is considered off the shelf today that wasn't always the case. It is an ICE specific design that took quite a while to develop as a standard part and because of its limited application outside autos (sensing oxygen levels in a stream of very hot exhaust gas) it is pretty much useless for anything else.
"Off the shelf" for electric power trains really comes from the industrial side. Most of those designs are hardwired. So while I could technically go to an industrial supply house right now and purchase a battery bank, dc drive, dc motor, and mechanical transmission they would not work well for the application of slapping them in in the place of an ICE powertrain. Even if I developed a car around the powertrain I certainly wouldn't get anything near what tesla or GM is getting from their designs. That has lead to very proprietary systems being developed to make the powertrains efficient for the auto application. Things like water cooled mesh wound DC electric motors (while they do exist in very niche markets) are almost unheard of in industrial controls. But put a few million autos on the street and it becomes an off the shelf part. Most of the solid state industrial motor control tech is catered to an AC input outputted to a 3 phase AC motor. These designs, while critical for the industrial applications they are used for, are useless for the auto industry. DC motors (in the power range that autos use) are pretty non existent in the industrial scene as well. The electric car market is not volume developed enough bring the OEM manufactures and until they develop standardized components "off the shelf" electric car components don't exist yet.
To your "A motor-drive unit has basically only four essential inputs" comment. You have completely lost me there. The only power input is from the battery bank. You have some control parameters listed in there (limits, RPM set point) and torque. These things are all controlled by the drive, which even in a dc system is going to be a PWM design. Regenerative braking isn't really an input at all, but a byproduct to be dealt with when you use your drive motor to reduce speed (temporally making it a generator rather than a motor). Diesel electric trains and many industrial speed controller designs have to employ braking resistors to bleed off the excess dc bus voltage in the form of heat for decades. Elegant designs store as much of this energy as possible to be fed back into the system for reuse but at best it is a glorified battery charger that cannot operate at the same time that output power demand is active.
The torque band on an electric motor is pretty much flat which is why it kills ICE on a drag strip. Ultra low rpm and ultra high rpm falls off, but everything in the middle is even. So torque limits are either a max system limit (current limited), or a waveform modification to control "slip" in the motor by artificially strengthening/weakening the flux field (which can greatly improve efficiency). RPM is mostly controlled by pulse frequency leveled by motor inductance from the PWM (minus mechanical slip). Because of the slippage issue the most expensive drives involve an encoder feedback system coupled with input/output power bus monitoring on the drive to accurately model the internal flux field characteristics of the motor and adjust accordingly. Cheap drives, golf carts and power wheels simply use a voltage control model that again doesn't work well in the auto application. The efficiency gains are pretty big with the more advanced designs but there are so many tweaks that can be done in an auto application that it again lends to proprietary drive designs having a significant advantage. This is a big reason that EZGO isn't the premier electric car producer in the world today, but they make a good golf cart.
This doesn't even go into the battery bank challenges.
When all auto players are really at ramped production the industry will self standardize to minimize costs. Tesla, if it can survive, is probably the exception due to their vertical supply chain mindset. Even so I would not expect to be able to easily put a GM component into a Ford or vice versa. Then again try sticking a ICE GM V8 ECM on a Ford V8 today and see what happens.
 
Last edited:

InvalidError

Titan
Moderator
Most of the solid state industrial motor control tech is catered to an AC input outputted to a 3 phase AC motor. These designs, while critical for the industrial applications they are used for, are useless for the auto industry. DC motors (in the power range that autos use) are pretty non existent in the industrial scene as well.
There is no such thing as "DC motors" in modern electric cars, all the most efficient designs are either induction or reluctance. The single biggest difference between industrial motors and automotive ones is that EVs feed the electric motor's VFD straight off the battery DC bus instead of the DC output from a 3ph AC rectifier bridge.

The only power input is from the battery bank.
I wrote "inputs" not "power inputs" - control inputs are still inputs and a VFD with regen braking needs to know how much power it has available, how much power it can regen, how much torque is requested, the target RPM, maximum RPM, etc. For motor-drives to get standardized, ALL essential inputs need to be standardized.
 

jasonf2

Distinguished
There is no such thing as "DC motors" in modern electric cars, all the most efficient designs are either induction or reluctance. The single biggest difference between industrial motors and automotive ones is that EVs feed the electric motor's VFD straight off the battery DC bus instead of the DC output from a 3ph AC rectifier bridge.


I wrote "inputs" not "power inputs" - control inputs are still inputs and a VFD with regen braking needs to know how much power it has available, how much power it can regen, how much torque is requested, the target RPM, maximum RPM, etc. For motor-drives to get standardized, ALL essential inputs need to be standardized.
AC motors are stock build around utilizing the 50/60 hz waveform. DC motors, while a misnomer, are a class of motors designed around DC based speed control and are not compatible with AC Systems. Go to Automation Direct and see if you can find a speed controlled DC motor. Induction or reluctance is technically compatible with either system if the controller is properly designed. Calling it an AC or DC motor is like saying an engine runs on gas or diesel. Induction or reluctance is more like a cam configuration. BTW you forgot permanent magnet in the list.
 

InvalidError

Titan
Moderator
AC motors are stock build around utilizing the 50/60 hz waveform. DC motors, while a misnomer, are a class of motors designed around DC based speed control and are not compatible with AC Systems.
An induction or reluctance motor is fundamentally an AC device regardless of whether it is directly fed by AC or via a VFD and most traditional AC motors still used in modern production environments have been retrofitted with VFDs anyway because it increases startup torque and eliminates the need for PFC. In some cases, it also eliminates the need for a separate startup motor and clutch to get synchronous motors up to speed before applying power too since traditional synchronous motors have near-zero torque under stall conditions.

The main difference between traditional AC motors and AC motors used in EVs is that the operating frequency is in the kHz range and just like everything else with magnetics, higher frequencies enable much higher power densities.

As for PMSM, those are basically the same as a conventional synchronous motor and would work perfectly fine on AC if you spin them up to synchronous speed first since you don't have external excitation control to allow slip.
 

jasonf2

Distinguished
An induction or reluctance motor is fundamentally an AC device regardless of whether it is directly fed by AC or via a VFD and most traditional AC motors still used in modern production environments have been retrofitted with VFDs anyway because it increases startup torque and eliminates the need for PFC. In some cases, it also eliminates the need for a separate startup motor and clutch to get synchronous motors up to speed before applying power too since traditional synchronous motors have near-zero torque under stall conditions.

The main difference between traditional AC motors and AC motors used in EVs is that the operating frequency is in the kHz range and just like everything else with magnetics, higher frequencies enable much higher power densities.

As for PMSM, those are basically the same as a conventional synchronous motor and would work perfectly fine on AC if you spin them up to synchronous speed first since you don't have external excitation control to allow slip.
On that premise all motors are AC devices as they require the reversal of the internal magnetic fields in order to create the push pull effect between poles to create rotary motion. That is achieved by either wave form or mechanical/solid state switching in every electric motor. Again as I said before "DC" motor is a misnomer.
I have a significant amount of design and application experience (over a decade) with VFDs on 3 phase and your assumptions are incorrect. VFDs do very little for the torque of an ac wound motor, in fact in low rpm bands (0-20% rated RPM) they have very poor torque and from a design application you try to avoid that band. In low RPM high torque situations this is why you see "DC" drives/motors which are custom wound and driven to keep the torque band wider (servos and stepper applications). In applications where inertial loads are present the torque requirement can be drastically reduced though because VFDs can ramp to speed rather than be in an instant on situation. This smooths the inertial load over time.

The near perfect power factor and ramp pretty much eliminate peak inrush current and can clean up startup voltage sag issues as well. In large industrial applications where the utility monetarily penalizes for poor power factor they do eliminate the need for PFC cap banks because the VFD intrinsically creates near power factor unity. If there were no utility penalty PFC wouldn't exist anyways though. The primary reasons though that VFDs have been integrated are flexibility and electrical savings. In applications where VFDs are being applied there is either a need to vary the speed of the motor for the application or a soft start need. Electrical savings can be realized through this because most real world applications are not optimally efficient at a single RPM and variable mechanical gear reduction is cost prohibitive. By optimizing with a VFD electrical use or a more elegant mechanical solution can often be found.

I don't know what you are playing with but startup motors on electric motors are non existent on any normal sized motor. Even at 300hp (which is massive for an electric motor) starters (large magnetically energized electric switches) are still used with no spin up assist. The big ones will sometimes have aux cooling fans (100hp plus) but that is not relevant here. At a certain point I could see that there would be an advantage to using a smaller motor to spin to rpm on a clutch, but only because peak inrush could get large enough that the utility couldn't handle it and sag out the transformer they give you. That is well out of the range of any normal application though. A VFD would help in that application.

The comment on the khz range is also pretty misleading. Carrier frequencies of AC drives will vary between .75 khz and 16-32khz depending on the drive. The overall simulated waveform targets are between 0-60hz (on 60 hz induction motors) because that is what the motor is designed for at a typical 1200-1800 rpm depending on pole configuration. I am not familiar with the pole configurations of the EV motors (as most are proprietary at this point) but as Tesla motors are rated for an output in the 18000 RPM range I would expect at least 10x waveform targets . Probably more if they increased the number of poles to increase torque. Regardless the PWM pulse frequency (carrier) and the simulated waveform targets are not relative to each other. In fact lower carrier frequencies have a number of advantages such as lower motor temps and reduced line ringing (destructive to motor insulation). The primary reason higher carrier frequency designs are used are reduced size of drive components and acoustic reduction from the motor. High frequency carriers require very short distances from the drive to the motor (Probably why tesla integrated it into the package).

Your line on higher frequency enables higher power densities statement is incorrect in this application. As carrier frequency increases the inductive (and on a smaller part capacitive) reactance issues cause extra impedance and extra heat issues in the motor. Also the impedance differences between the drive, output wiring and motor cause a line reflection that creates a standing wave (line ringing) which gets much worse with increased carrier frequency. This doesn't really effect efficiency but can created line voltages high enough to corona the wire, degrade insulation (especially in lower inductance motors) and cause RFI in other systems. So while you can give me a high school physics equation on waveform energy density the optimal sweet spots in the real world for VFD applications have limits that are pretty well documented.

When you compare a water cooled Tesla setup and an standard air cooled industrial induction motor they are really not in the same category. As far as power output the tesla outputs over 350 hp in a complete drivetrain that weighs less than 200 lbs. A 300 hp industrial motor itself, without transmission, will weigh 2700 lbs and cost $13000-$14000 (plus the LTL freight) alone without a drive which is custom and probably costs at least half as much as the motor. As stated at the beginning of this thing the current "off the shelf" parts available are not yet at a state of "good enough" for general application in the auto industry. In that same vein a tesla drivetrain would not drop into an industrial application and survive either, but I could at least pick it up.
 
Last edited:

jasonf2

Distinguished
You cannot make a 100lbs 200HP motor using 60Hz input simply because direct 60Hz motor requires enough winding inductance to prevent it from self-destructing and those end up weighing 1000+lbs.
It is also a function of cooling and torque distribution. The shaft and bearings on a 18000 RPM motor at 300 hp are able to be considerably smaller than the equivalent at 1800 RPM (which are pretty massive). It also helps shrink the gearbox significantly too. Liquid cooling makes a huge difference too as alot of the mass is simply heat sink especially on a TENV motor. I would imagine Tesla's motor design took some imaginative engineering to cool the rotor at sustained operation even with liquid cooling. Regardless looking it over online it is absolutely amazing that something smaller than an old Geo metro three cylinder drive train puts out so much power and doesn't blow itself up.
 

USAFRet

Titan
Moderator
Regardless looking it over online it is absolutely amazing that something smaller than an old Geo metro three cylinder drive train puts out so much power and doesn't blow itself up.
30 years of refinement.

The same 30 years that goes from 130MB to 1TB (10,000 x capacity and 10x the speed) , in the same size package, for the same "$100".

(and I used to have that Geo...lol)
 

InvalidError

Titan
Moderator
It is also a function of cooling and torque distribution.
And the cooling itself is a function of losses which are much lower in magnetic cores using advanced ceramics and alloys engineered for high frequencies than laminated iron cores, which enables them to be at least an order of magnitude smaller and lighter for the same power.