Thoughts on TESLA

[ERD50]

My friend just got off the phone with Tesla. He was not very pleased.

They walked him through a 3 minute, "hold these two buttons while you press the brake, release, then don't move for 3 minutes", essentially a "Control_Alt_Delete" for a Tesla.

And if it does it again? "Now you know the procedure to reset it!"

Here's a description I found on the web:

Option 4 – Deep power down and reset – Reset of all systems

Sit in driver’s seat. In park with all doors closed.
Menu, service, power off.
Wait about 60 seconds for the interior lights to go out.
Wait 30 seconds more to be safe.
Press brake pedal.
Wait about 30 seconds whilst everything boots up (you may see odd error messages for a few seconds).
Done.

-ERD50

 

[bobandsherry]

Interesting perspective regarding battery range from a Tesla owner/fan who lives in cold climate:

For those waiting for the short-range '$35k version' here's my advice: I was waiting for that too but with the $7,500 rebate getting halved and then reduced again in mid-2019 it made way more sense to get the larger battery version now. I actually got the long-range battery (310 mi) and while I would have gotten the mid-range if it was available I don't think I would be happy with anything less than the 260mi mid-range model. I'm in snow-country and only charge my baby to 80% daily so I start with around 240 - 260mi, but that range estimate is for best-case scenario only. Going 80mph with my cabin at 72F and seat warmers on, that range is cut drastically, basically I get maybe 140mi of spirited driving.

With recommended charging rate and cold weather, you quickly lose a lot of the range Tesla gives you. Granted, it's still a LOT of miles of driving before I get anything close to range anxiety but I'm very glad to have the extra bump in battery capacity.

 

[bobandsherry]

Yes. As he was telling me this, I'm thinking of all the interference that powerful motor and control system must be generating. Switching high currents and high voltages. They obviously have the proper shielding and controls in place, but that often relies on very good grounds. And we know that grounding in the harsh environment of a vehicle (here with salt on the roads) can get iffy after a while.

-ERD50

Welp, when ya read stories like this one where guys Tesla S caught on fire twice in one day, has you wondering if proper shielding is in place.

Tesla Model S catches fire in California parking lot and reignites hours later at a tow yard

Another article with more information: Tesla Model S Catches Fire Twice in One Day

Found this comment from Tesla interesting:

Battery fires can take up to 24 hours to extinguish. Consider allowing the battery to burn while protecting exposures.

After all fire and smoke has visibly subsided, a thermal imaging camera can be used to actively measure the temperature of the high voltage battery and monitor the trend of heating or cooling. There must not be fire, smoke, or heating present in the high voltage battery for at least one hour before the vehicle can be released to second responders (such as law enforcement, vehicle transporters, etc.). The battery must be completely cooled before releasing the vehicle to second responders or otherwise leaving the incident. Always advise second responders that there is a risk of battery re-ignition.

Due to potential re-ignition, a Model S that has been involved in a submersion, fire, or a collision that has compromised the high voltage battery should be stored in an open area at least 50 ft (15 m) from any exposure

 

[NW-Bound]

My friend just got off the phone with Tesla. He was not very pleased.

They walked him through a 3 minute, "hold these two buttons while you press the brake, release, then don't move for 3 minutes", essentially a "Control_Alt_Delete" for a Tesla.

And if it does it again? "Now you know the procedure to reset it!"

-ERD50

Is it the same procedure to use if it happens while you are cruising at 70 mph down Freeway 405 to San Diego?

I would not like it either. I would not care if 99,999 other Teslas do not have this problem. If my car is infested with a Gremlin like this, I would not feel safe driving it unless the cause is positively identified and fixed.

 

[eroscott]

Is it the same procedure to use if it happens while you are cruising at 70 mph down Freeway 405 to San Diego?

You can reset some rare/strange issues while driving. The User Interface/UI and even AutoPilot systems are separated from the basics of the drivetrain. I've reset my Tesla while driving and being in AutoPilot. The center console display just reboots and comes up in short timeframe. All driving features work as normal while this happens.

They've thought about all these scenarios.

 

[eroscott]

My son is on a multi-state roadtrip as we speak headed to the in-laws. Drive is way more comfortable and relaxing as the car 'drives itself' on the highways (hands always on wheel of course in this level 2 system but way way less cognitive work). He is driving a Long Range (LR) AWD Model 3. Everything is logged in the Tesla as they have interfaces to all components in the car.

One of the things people don't realize is about it is the fuses.

Owner (who is an electrical engineer):
There indeed are zero fuses on the 12v side. All circuit protection is performed solid-state, which in practice means a transistor (MOSFET) is used to switch all the loads, and the body controllers monitor the current going though each of these and in the event of an overload, it just switches off the transistor. It's much faster and safer, and allows a more reliable and easy to diagnose car. It is more expensive, but this was needed to make the M3 fault tolerant to achieve full self driving with confidence.

Owner (who is an aero engineer):
Wow. That's possibly a bigger change to how a car works than anything else Tesla has done to date.

 

[braumeister]

Sorry this is OT, but I couldn't resist. Just got an email from my local utility:

Save up to $3,000 on a new Nissan LEAF.
For a limited time, Nissan is offering a $3,000* incentive to customers of Duke Energy. Offer valid through Dec. 31, 2018, on retail purchase of a new LEAF.
Bring a printout of this email or the PDF flyer along with a copy of your Duke Energy bill to your local Nissan dealer.

Adding the $3K rebate on to the $7,500 tax credit would be pretty attractive if one were in the market.

 

[aja8888]

Only an anecdote, not data - but the friend of mine who picked up his Tesla S on Wed has had some some problems.

Screen went gray, then black, heat went full on, radio was cutting in and out, all sorts of wonky things. Told him he was in park when he 'shifted' to reverse, but actually was in 'forward'. It seemed to reset itself after a minute.

-ERD50

My neighbor a few houses down, bought a new Corvette this summer. He paid near $100K for it and it has all the options. Nice car, I drove it around the neighborhood...impressive! It's a mess electronically though. His dash electronic screen goodies quit and some other not related suspension control functions stopped.

It's been in and out of the dealer several times and he just got it back and supposedly it's all fixed. He's not happy with the dealer's solutions and several attempts at fixing it.

Let's face it, all the newer generation cars and trucks are loaded with all kinds of computer controlled switches, motors, monitors, etc. and this creates an environment for lots of things to go haywire.

 

[aja8888]

I forgot to mention.....as I was waiting for DW to pick me up at the hospital discharge door this afternoon ( I had my SVT ablation procedure today), a Tesla Model X SUV pulled up to pick up another patient who was also being released.

I got a close up tour of that X and it is very, very impressive. The driver had to wait a bit so I got to see the fancy screen and the whole interior. Nice vehicle. A lot nicer than the Hyundai Santa Fe picking me up.

 

[NW-Bound]

You can reset some rare/strange issues while driving. The User Interface/UI and even AutoPilot systems are separated from the basics of the drivetrain. I've reset my Tesla while driving and being in AutoPilot. The center console display just reboots and comes up in short timeframe. All driving features work as normal while this happens.

They've thought about all these scenarios.

How often have you seen this phenomenon? What is the experience of other Tesla owners?

The separation of critical functions from non-essential ones has been a principle for design of aircraft systems, back long before I started working in this field in the late 70s.

This was done mainly for redundancy management. For example, an aircraft may have a triplex (3-channel) or quadruplex (4-channel) autopilot for automatic landing, but the autothrottle control may be of lesser redundancy. The rationale for this is that the throttle servo is usually designed to be rate-limited (and large turfofan engines have long spool-up time anyway), and any error should be more easily caught and corrected by the pilots. A malfunctioning autopilot can drop the nose of the aircraft when it is 100 ft off the ground before the pilots can react.

The separation of critical and non-essential functions is usually a physical one, meaning they are implemented in different subsystems or boxes in the aircraft. And this was done back in the days of analog autopilots for jetliners, long before any computer was used.

It looks like Tesla is doing the right thing to separate the critical functions for driving from the bells and whistles. It is likely that the car may even have different CPUs for different functions. However, it appears that this is done so that bugs in the display or non-essential functions will not cause the car to crash, and not to protect against hardware failures as is done with aircraft.

This points out a difference between car designs and aircraft designs. Can you imagine an aircraft maker telling its airline customers or the FAA that you should just do a "Ctrl-Alt-Del" when the cockpit display goes crazy, or indicators start flashing like Christmas decoration lights?

The lousy quality of software for PC has conditioned the public to the point that they accept buggy products as a fact of life. This is sad.

 

[NW-Bound]

My neighbor a few houses down, bought a new Corvette this summer. He paid near $100K for it and it has all the options. Nice car, I drove it around the neighborhood...impressive! It's a mess electronically though. His dash electronic screen goodies quit and some other not related suspension control functions stopped.

It's been in and out of the dealer several times and he just got it back and supposedly it's all fixed. He's not happy with the dealer's solutions and several attempts at fixing it.

Let's face it, all the newer generation cars and trucks are loaded with all kinds of computer controlled switches, motors, monitors, etc. and this creates an environment for lots of things to go haywire.

There are problems caused by hardware reliability, and there are problems caused by software. It appears your neighbor's Corvette suffers from hardware failure, while the Tesla may have software problems which can be reset.

And then, there are intermittent hardware problems that can cause the CPU to go crazy, which then heals itself when rebooted. ERD50's suggestion of EMI (electromagnetic interference) for example can explain the intermittent CPU crash too. Intermittent problems are very tough to track down whether in hardware or software, as anyone can imagine.

 

[Chuckanut]

And then, there are intermittent hardware problems that can cause the CPU to go crazy, which then heals itself when rebooted. ERD50's suggestion of EMI (electromagnetic interference) for example can explain the intermittent CPU crash too. Intermittent problems are very tough to track down whether in hardware or software, as anyone can imagine.

Cosmic rays can flip a bit from time to time. Or so I am told.

 

[NW-Bound]

Surely they can. But that is more a problem for spacecraft outside of the atmosphere, which attenuates much of the rays for terrestrial computers.

I wonder if any here still remembers the design of the original IBM, whose "radical design" used dynamic memory instead of static memory used in the earlier CP/M machines. Much was discussed in literature then about the state of a dynamic memory bit being defined by a much smaller electric charge than a static memory bit. This would make a bit easier to be knocked by a cosmic ray. Dynamic memory was also new, and its reliability was not taken for granted.

To compensate for this potential lack of robustness, IBM gave each byte a 9th bit to serve as the parity bit. A bit that failed, stuck to 0 or to 1 or flipped by a cosmic ray, would cause a parity failure, which generated an interrupt to the CPU and stopped it.

Then, at some point in the PC evolution which I forgot, the parity bit was dropped, presumably when it was found that the fear was overblown.

 

[aja8888]

There are problems caused by hardware reliability, and there are problems caused by software. It appears your neighbor's Corvette suffers from hardware failure, while the Tesla may have software problems which can be reset.

Neighbor's Corvette problems started erratically and the dealer tried a software fix. That didn't work very well and now they have replaced a module, but had to pull most of the dash.

 

[eroscott]

The separation of critical functions from non-essential ones has been a principle for design of aircraft systems, back long before I started working in this field in the late 70s. ...

Probably in the space (X) industry too <wink>.

Re: separation
The Teslas have various Bus Systems – CAN (Controller Area Network), LIN (Local Interconnect Network) and 100 Mbps Ethernet via a 6 port switch.
There is like 7ish CANs, 5ish LINs and 4ish Ethernets. ie. like subsystem separation and some gateways between them perhaps. Sometime none if security is a ultimate concern ... think braking system, drivetrain, autopilot, etc.

How often have you seen this phenomenon? What is the experience of other Tesla owners?

I don't hear about it too much but on Tesla forums sometimes you hear about problems more than not as is the nature of car forums.

I do this pretty infrequently explicitly. I've done it if my bluetooth connection is having problems (if you have ever researched / developers on the bluetooth stack you know it is wonky). For bluetooth, perhaps 1 every 4 months. I don't hesitate to do it if I have phone connection issues because I use the phone voice for BT and my Tesla gets my Google Calendar via the phones Tesla app. If I have addresses in my daily calendar activities then when I get in my car a list of my days events pop up and I just tap the one I'm going to and it puts it in the cars navigation. I like having/knowing the ETA or getting rerouted automatically if traffic.

I have also occasion rebooted if I was goofing around on new webpages in the car which hung or doing some speed testing and I wanted to start 'fresh'.

Implicitly I think a reboot happens when I get Over-The-Air/OTA updates regularly. If I get a notification on my phone when I'm in the house, I just go to the Tesla phone app and tell it to install the OTA update (note it is downloaded and verified *before* I would get the notification).

My current cars OTA updates (note that last entry is from my previous car so the time-between is inaccurate).

Note many of these OTA updates contain new features or tweaks to existing features (think music system, HVAC (mode to leave your pet in or mom), route planner, falcon-wing door (early on), etc). Some are software bugs. Some are related to the AutoPilot system or HD map usage.

 

[NW-Bound]

Probably in the space (X) industry too <wink>.
...

I did not work on any manned space program, so cannot speak there. On programs that involve only unmanned missiles, the consideration for safety was not the same as for commercial aircraft. For launch safety, there's a destruct button for them to send command to blow up the missile if it looks like it is about to go off-course and fall on population center.

A lot of testing is done in the design of missiles of course, and it's because the cost is so high and not for human life consideration. Also, flight time for missiles is short, so people generally do not worry about failure rates due to aging, compared to jetliners that fly for decades. And unmanned missiles do not have the "stuff" that manned vehicles do. For example, all the display functions are on the ground equipment.

I also work on military aircraft. They all have redundant equipment on board. However, the testing is not to the same level done for FAA certification. The truth is that military aircraft gets shot at anyway, and they carry a crew of 2, who can also eject. On the other hand, for jetliners, the FAA requires FMEA (Failure Mode and Effect Analysis) and FTA (Fault Tree Analysis) done to show that the chance of a crash in a flight is 1 in a billion. The risk of a military aircraft getting shot at by the enemy is a lot higher than that.

What does the NHTSA require? I do not know, as I have never worked on autos.

 

[eroscott]

Probably in the space (X) industry too <wink>.

I did not work on any manned space program, so cannot speak there. ...

It was a SpaceX joke. Your comments about their needs make logical and common sense tho. FYI, there have been various references of some SpaceX knowledge bleeding over to Tesla engineering and development in several categories.

 

[NW-Bound]

I knew what you were inferring. Yes, I know about Space X.

But the truth is that more and more modern features in cars are treated as functions in a TV, a laptop, or similar consumer products. If the feature works, great. If it does not work, then reboot. Car makers, not just Tesla, know that the braking and steering functions cannot be treated in a cavalier manner as the Bluetooth function, or the various fancy display, and this is good.

I just hope that they do not carry this attitude into the hand-off autopilot function. Well, if they did, they would not be in this business long anyway, so the problem is self-limiting. Witness the Uber fatal accident, for example.

PS. By the way, Waymo is still doing a lot of testing in my hometown. I have not seen one without a driver on board though.

 

[oneill225]

The proper safety standard for fully autonomous driving software is not perfection or 1 in a billion accident, it is simply safer than letting Joe Schmoe get behind the wheel. The vast majority of car accidents are caused by inattentive humans, not mechanical failures (or software). Tesla and the other autonomous driving developers have likely already met this standard, but convincing drivers to trust the technology will take a few more years.

 

[NW-Bound]

The proper safety standard for fully autonomous driving software is not perfection or 1 in a billion accident, it is simply safer than letting Joe Schmoe get behind the wheel. The vast majority of car accidents are caused by inattentive humans, not mechanical failures (or software). Tesla and the other autonomous driving developers have likely already met this standard, but convincing drivers to trust the technology will take a few more years.

?

Much is claimed about safety of the Tesla AP, but it is not autonomous!

It may be argued that the combination of the Tesla AP plus a human driver to supervise it is safer than a human driver alone, but the Tesla AP by itself is not safe.

Waymo technology is far superior and safer. Yet, they are still not ready to release it. Still not safe!

 

[NW-Bound]

By the way, autopilot systems in jetliners all have built-in self-test and monitoring. They will shut down when they detect failures for which they cannot compensate to continue.

The theoretical 1-in-a-billion chance of crashing is computed with the human pilot on board to take over in the case that the AP cannot continue.

Well, human pilots do fail too, and that's why the 1-in-a-billion number is not met in real life.

My point has been that humans are not perfect, but nor are machines. Trusting machines blindly before we know how to build them is not good.

PS. Autopilots on jetliners can fail. We do not know how to build something so reliable that their failure rate is as good as 1 in a billion flights. But aircraft makers know how to build them such that the chance of them having an undetected failure is extremely small. When the hardware fails and knows that it fails and disengages, that's when the human pilots step in.

And so, airplanes still have pilots. Two of them too.

 

[Midpack]

I thought this was interesting, a complete teardown and analysis by Munro. There are several videos and as far as I can tell after watching them, they really like the Model 3 overall, way ahead of any competitor in terms of drivetrain/power/systems technically. However, they were highly critical of the more basic parts count of the body and manufacturability (in the vid below). Established carmakers will get that right, but they have their work cut out for them in terms of the drivetrain/power "guts."

FWIW

 

[oneill225]

Much is claimed about safety of the Tesla AP, but it is not autonomous!
It may be argued that the combination of the Tesla AP plus a human driver to supervise it is safer than a human driver alone, but the Tesla AP by itself is not safe.
Waymo technology is far superior and safer. Yet, they are still not ready to release it. Still not safe!

You miss my point. The question is not whether autonomous driving systems (full or partial) are "safe", the question is whether they are "safer" than allowing humans to drive vehicles. If 20,000 accidents are caused by drunks and texting teens each year, then the safety standard for allowing autonomous driving will be anything short of that number.

As I said, the tech for fully autonomous driving is likely there, but it will not be fully implemented until we get more comfortable with turning over full control of our cars; that will take several years, but it is coming.

Liability is also a deterrent to fast adoption, but just as with the airlines, liability can be limited by law in order to preserve the greater good that autonomous driving will provide (less damage and death than human driving).

The airlines will eventually replace pilots in the cockpit, as well. At first, they will still be monitored by pilots on the ground (like drone flying), but even that back-up will go away as the technology and passenger comfort level renders them unnecessary.

 

[NW-Bound]

No, I did not miss your point.

But what you said below is plainly wrong.

... Tesla and the other autonomous driving developers have likely already met this standard...

I don't know how safe Waymo technology is because they have not released it to the public, but Tesla owners have to override or correct the AP several times on a trip. How is that autonomy?

PS. I am not saying they will not get there. I am saying that what they have is not it. I do not have to own a Tesla to know. Just looking at youtube videos of drivers having to correct the AP frequently tells me how it works.

 

[oneill225]

No, I did not miss your point.
But what you said below is plainly wrong.
I don't know how safe Waymo technology is because they have not released it to the public, but Tesla owners have to override or correct the AP several times on a trip. How is that autonomy?

I agree that we do not know and cannot know until these systems are legalized for full autonomous driving and companies like Tesla and Waymo roll-out the capability. Can you at least concede that it is coming and we are close (within a decade) of seeing full autonomous driving becoming a reality?