Hacker News Reader

Comment by DmenshunlAnlsis | Hacker News Reader

The Tesla’s stopping distance of 152 feet from 60 mph was far worse than any contemporary car we’ve tested and about 7 feet longer than the stopping distance of a Ford F-150 full-sized pickup.

That’s worrying, and almost as worrying is that Tesla says their braking is much faster. I’m sure plenty of fans will make comments about shorting stock, but this is Consumer Reports, and they have a very good reputation to protect.

I’m not thrilled by this tidbit either.

The Model 3’s stiff ride, unsupportive rear seat and excessive wind noise at highway speeds also hurt its road-test score. In the compact luxury sedan class, most competitors deliver a more comfortable ride and rear seat.

View on HN

This part really intrigued me about the stopping distance:

The Tesla spokeswoman says the company has the ability to update its vehicles over the air. “Unlike other vehicles, Tesla is uniquely positioned to address more corner cases over time through over-the-air software updates, and it continually does so to improve factors such as stopping distance,” she says.

I'm curious the mechanism that makes that possible, or if it's just BS.

Why would you think it's BS? The Model S gets over-the-air updates all the time; if I recall correctly, they increased the ride height a while ago through an update after a car drove over a metal wotsit, causing a puncture in the battery compartiment and a fire.

They can probably update the settings for the brakes as well, as long as the underlying mechanicals support it.

There are plenty of reasons why you would default to a lower ride height. What would be the purpose of intentionally making your brakes worse than an f150?

I don't think anyone's suggesting it was intentional. Parameters for managing the brakes may just have been set badly, or the anti-locking algorithm written poorly. A possible explanation is they may have optimised it to improve handling while braking.

I think the implicit question is "can an OTA somehow improve braking performance - why isn't it optimal already?" not "do Tesla OTAs change other aspects of the car?"

Electric motors and generators are essentially the same thing run in different directions. Run electricity through a motor and it generates motion while putting load on the power. Move a generator and it creates resistance as it extracts electricity. This also answers both how they can change it and why it is not already optimal. Actively using the motors to break causes it to use energy instead of recovering it. Hybrid drivers have long known to break gradually when possible to maximize battery recharge. It is possible they had an electric neophyte as a test driver. Fair enough if they were supposed to test from the perspective of someone new to hybrids and electric cars trying a Tesla as their first.

Not sure if you read the article but what they were testing is pretty much emergency braking from 60mph to full stop.

So the goal is not regenerate the battery (and it doesn't matter if the driver was "an electric neophyte"), it's to brake hard and to repeat the process a couple of times to see if you get brake fade.

From the article: Our testers get a car up to 60 mph, then slam on the brakes until the car comes to a stop. They repeat this multiple times to ensure consistent results. Between each test, the vehicle is driven approximately a mile to cool the brakes and make sure they don’t overheat.

Whoops. My mistake for giving into the temptation to segue after my explanation of what I did know.

What does "gradually braking to optimize battery recharge" have to do with an emergency stop?

I never said I thought it was BS. I was curious what the physical mechanism to change the braking characteristics was. If there was no physical mechanism that I assume it's BS.

The car most definitely has software controlled anti-lock braking equipment, and likely has software controlled electric brake cylinders. Both of those things could plausibly be updated via an OTA update.

They improved acceleration via a software update[1] so it's plausible at least.

[1] https://www.engadget.com/2016/11/16/tesla-p100d-software-per...

Edit: why downvotes?

To me, it is disturbing. I get used to the way my vehicle handles, for it to suddenly change because of a software update could potentially be dangerous- but would almost certainly result in a trip to the mechanic for a checkup.

Maybe they have a system for warning owners about these sorts of updates?

Well, yes. You get prompted to update your car (you can choose not to). And then after the update is complete, you get a Whats Changed prompt the next time you're in your car. Essentially a patch notes for your vehicle

For something as major as changing the way a car breaks, do you think "patch notes" (if read at all) are able to properly convey the magnitude of the change?

How do you know that the change is driver-visible? The change might only be for braking hard enough that the anti-lock feature kicks in... which most drivers don't do very often.

Yes? My 75D had an entire second shaved off of its 0 - 60 time. I was aware of the change and was able to deal with it just fine...

I'm also not sure what the alternative you're imagining here is. Take it in to a shop just to have them install the update for you? I also don't see how this is any different than changing worn out brakes on an old ICE car. I had the brakes changed in my old '97 Accord, and the difference was noticeable when I pulled it out of the driveway. I think you need to give people a little more credit...

Also, the Whats Changed prompt literally takes up the entire 17in touch screen. It's sort of hard to miss.

I'm talking about the general populace not only actually reading a prompt but understanding it. Based on how people interact with computers/phones, I do not have a lot of faith in their abilities on this end.

I didn't propose an alternative but there's a huge difference in awareness between a physical in-person "update" vs an OTA update.

For something as serious as changing brake handling, I'm sure they would have some sort of prominent display/cannot continue until you agree prompt but there's always going to be people who do not read things in their entirety. For something that can be destructive as a car, I don't find that acceptable.

Is there an option to revert?

Good, that makes sense! It would be preferable if they gave you the notes before you installed the update (maybe they do provide a link for it?) Thanks for confirming that, anyway.

They've already been doing this, back when dual-motor cars first rolled out they had an update that cut the 0-60 from 4.6s to 3.9s.

Def changed the way the car drove, can't say that I minded in that case though.

I can see that for many other areas of handling, but breaking has always seemed the most variable to me. Different road conditions, tire conditions, if the breaks themselves are wet, etc.

I suspect you would get used to it fairly quickly.

Right, but I like to at least pretend to know what the road conditions/ tire conditions/ brake conditions are. Maybe I'm not as clever as I think I am, though :) I try to be a very conscientious and careful driver.

My understanding is that car brakes can generally exert more braking force than the tire's grip on the road. A long stopping distance may represent a problem with the anti-lock braking system (which is likely software controlled, and can be updated over the air).

Interestingly, Waymo's self-driving car cannot be updated remotely.[1] As a security measure, Waymo deliberately did not install a data connection to allow that. Probably a good decision.

[1] http://www.documentcloud.org/documents/4458000-Waymo-Driverl...

That's probably also to do with Waymo's cars being developed primarily for ride hailing fleets. With a private ownership model, requiring owners of variable technological faculty to manually apply updates would be less practical.

That was more a PR move than a real security measure.

Signing updates and checking signatures is established technology to certify that updates come from the right place and are not tampered with.

That and the updates are likely >10 GB, so you wouldn't want to download them over mobile data anyway.

It could, but it could also be a function of the car's weight (batteries are heavy) and the tires it comes equipped with from the factory. Poor tires will yield poor braking performance due to lack of grip.

But it can't be poor tires, because the Model 3's brakes were proper in the first test (~130 feet stopping distance). The same car with the same tires then had different braking behavior in every test after the first test.

That's just odd.

Most vehicles detect a sharp press to the brakes, and add extra assistive braking force. Thats from the observation that most humans do not apply enough braking force in an accident (potentially because they have never applied that much force before, and don't know how much is 'enough').

It's possible that the tesla software either isn't applying the extra assistive force, either due to incorrect tuning or a bug.

I can almost guarantee the problem is caused by the brake pads glazing over. The working surface of a brake pad will become work hardened, which drastically lowers their friction coefficient, once pushed past their designed temperature limit.

The Model 3 (3,800lbs for the long range model) is a fat pig of a car compared to other, conventionally powered cars in its class (3,300lbs for a BMW 3 Series) but this is countered by Tesla fitting them with relatively enormous brakes (14" rotors up front, 13.5" in the rear, larger rotors have more heat absorbing capacity for the pads).

If I had to guess the root cause for this problem, I'd say it is likely down to Tesla's focus on minimizing aerodynamic drag causing a lack of airflow within the wheelwell and across the brakes out the wheel itself. Either that or they chose an absolutely terrible brake pad compound but that just doesn't seem likely in this modern era. Either way, the brakes are getting so hot after a single stop that CR's cooling procedure (a reasonable one I might add) is inadequate to keep the brake pads from drastically overheating. This also explains why braking performance was still compromosed after letting the car cool overnight. The work hardened surface will perform poorly until it is literally ground off from further heavy use of the brakes.

> I can almost guarantee the problem is caused by the brake pads glazing over.

That's actually a very reasonable explanation. I can't say I'm a car expert, but your logic at least sounds correct.

Can't brake pad material also adhere to the surface of the rotor as well, requiring that they be turned to remove the material? I think that's what people mean when they talk about the rotor being "glazed," a related problem to what you're talking about. This is all a potentially expensive repair.

Usually, when pad material adheres to the rotor, people call that "warped rotors" because it feels like the rotor is warped when braking. It's not usually uniform, I think.

As for expense, brakes are simple and the parts generally cheap, at least for mass-market cars. You can get a full set of pads and rotors for less than $100 from Autozone.

"Warped" rotors, yes, I've heard that term. Not something that comes up in conversation as much as it used to.

They are a lot cheaper today but the cost of rotors varies tremendously depending on the manufacturer and their volume. I gather they turn rotors a lot less than they used to because they're so darn inexpensive to replace, but I'm pretty sure it's still economical for some of the higher-end stuff.

I'm not sure how true this is. As the car's weight increases so too does the grip, as it's a function of down-force multiplied by the roughly constant coefficient of friction of the tires.

Conceivably, the car's weight may mean that aerodynamic down-force produces a relatively smaller impact on grip, and air resistance a relatively smaller deceleration. But I'm not sure that's enough to account for the difference.

Excellent racing tires have coefficient of friction > 1. That is, a 2-ton car can have more than 2-tons of force applied from the tires.

But most consumer tires have coefficient of friction of less than 1. ~.9 for higher end stuff, but it could be as low as ~.5 or so. Which mean 2-tons of car will lead to LESS than 2-tons of acceleration / deceleration / turning force.

In effect, "more weight is more advantage" is only true if you have a coefficient of friction greater than 1.

The last line of your post isn't clear.

Since F=MA, A=F/M. Assuming the braking force (F) is μ (coefficient of friction) * Mass * Gravity, thats F=μMG.

Combining those equations you get:

A = μMG / M

And the M's cancel, giving:

A = μG

Therefore the braking distance due to road friction alone (and ignoring aerodynamic downforce effects) is independent of the mass of the vehicle.

Isn't this neglecting the forward velocity and momentum?

According to theory, the width of a tire has zero effect on tire grip. Race cars and high-performance cars prove this is incorrect: there's a reason they have wide tires (which also cause poorer fuel economy through extra friction). Tires and their interaction with the road are a lot more complex than some simple physics equations.

Hmm, that makes sense.

Its like the whole "area" doesn't matter as well. I can't help but think that maybe the Newtonian friction model is overly simple sometimes. Surely weight and surface area would have an effect?

But perhaps not really. Or at least, not in a way that's very measurable.

I think it's definitely overly simple. If it were correct, high-performance cars wouldn't have wide tires. According to Newtonian friction, the contact patch is the same size regardless of the size of the tire. Obviously that's silly; a Lamborghini is not going to get sufficient grip out of economy-car tires, nor would an 18-wheeler. That model completely neglects the complexities of the construction of the tires; they're not like balloons, they're actually semi-rigid.

But braking is a solved problem. Apply the maximum force to the tires, until right before the tires begin to slip. Then, reduce the force to prevent slipping. Stay within the "static coefficient of friction", which is higher than "dynamic coefficient of friction".

Alternatively, there's also "let the human do it". (Aka: threshold braking), but most people prefer to have the computer do this task during an emergency.

This algorithm has been used in anti-lock brakes for decades at this point. I'm not sure why it would need an OTA update to improve. Once this algorithm is properly implemented, then the only thing left is to buy better tires and use better rubber to get better grip.

This CR report suggests that Tesla's braking behavior is suboptimal. There's no reason for a smaller, lighter Model 3 to have a worse braking distance than the freaking F-150 truck.

> But braking is a solved problem.

That's not to say Tesla haven't decided to re-solve it / implement it themselves anew. The rest of the auto industry has been doing ABS much longer than Tesla has existed.

It's also not a prescriptive algorithm. ABS systems don't stay within the static coefficient of friction; they repeatedly retest the boundary between static and dynamic friction (hence the judder). It's a bit like optimising TCP window size for those familiar with it. There are a bunch of parameters you can tune.

It also depends what your goal is.

Typical ABS systems typically perform far worse on gravel or sand than non-abs systems for example. It turns out on gravel locking the wheels is usually a better solution because then you start to plough deep into the gravel.

No ABS system could determine that beforehand, since it isn't to know if the gravel is deep (allowing ploughing deep into it) or shallow (where staying on the boundary of static/dynamic friction works better).

I understand Teslas are fairly heavy cars. That explains most of the problem I think.

The Model S is in fact about as heavy as a Ford F150.

However, if this braking result is consistent across all Model 3 vehicles, that's a bit worrisome. I'm also a little concerned with the other issues they had (ergonomics, wind noise, etc). From everyone I know that drives or has driven a Model 3, they've had amazing experiences as far as actual driving

"this is Consumer Reports, and they have a very good reputation to protect"

CR doesn't have a very good reputation though? I mean... they used to. But their reputation is middling at best now.

Who has better reviews for consumer goods?

Reply on news.ycombinator.com