Anyone got HW3 retrofit yet?

[Knightshade]

plus its only a matter of time before tesla offers battery upgrades once the cars are out of warranty. Why would u pay 15k for a 300 mile battery when their new cars have 500 mile range. Makes no sense..

Until such time as Tesla has more batteries than demand for new cars (and powerwalls) it'd make no sense for them to offer that though- right now the company continues to be battery-restricted on production of new vehicles.

Maybe in a few years when they've got terafactories it'll be a different story.

 

[TT97]

Until such time as Tesla has more batteries than demand for new cars (and powerwalls) it'd make no sense for them to offer that though- right now the company continues to be battery-restricted on production of new vehicles.

Maybe in a few years when they've got terafactories it'll be a different story.

One thing Tesla could do if they offer battery upgrades is reuse the old batteries for their powerwalls - as the powerwalls are much less demanding on the batteries than the cars (and do not have the premium on space or weight as cars), even a battery with 70% retention capacity can continue living a long and happy life at the farm upstate (inside a powerwall).

 

[MarcG]

One thing Tesla could do if they offer battery upgrades is reuse the old batteries for their powerwalls - as the powerwalls are much less demanding on the batteries than the cars (and do not have the premium on space or weight as cars), even a battery with 70% retention capacity can continue living a long and happy life at the farm upstate (inside a powerwall).

They would have to redesign the Powerwall to be able to use 18650 cells, which is what Model S & X use for their battery packs.
Current Powerwalls use 2170 cells, those found in Model 3 & Y.

 

[TT97]

They would have to redesign the Powerwall to be able to use 18650 cells, which is what Model S & X use for their battery packs.
Current Powerwalls use 2170 cells, those found in Model 3 & Y.

I was actually thinking batteries from the 3&Y (and I am assuming future cars as well). There will be multiple times more 2170s in circulation than 18650. They would probably need to redesign the powerwall regardless if using older batteries as it would require more batteries to match the same capacity of new batteries (of course, a box is much easier to redesign than a car - you can use the stretch tool in Paint).

 

[israndy]

I don't really see enough people buying new battery packs for their Teslas going forward. It's not too common as it stands. Mostly damaged packs are replaced under warranty or after an accident. Taking the pack part to repair it looks next to impossible with all the glue and such. So that means taking the individual non-damaged modules and figuring out how to build a power wall from them. I imagine it would more likely be something an individual RVer would do to get a nice 24volt battery for their vehicle than something coming down an assembly line at Tesla New York.

I did read about JTs new business, Redwood Recycling or somesuch, that would recycle the electronics from a Tesla at the end of life. I can see a machine designed to specifically take-apart the battery packs in dead cars, like Apple has that machine to dismantle old iPhones. I think Redwood is trying to get the components out to melt down and start again. Supposed to be cheaper than mining for the materials. Cannot see a mass operation trying to reuse the individual batteries, so much time wasted trying to build a matching pack from the remnants of the old cars.

I think the people wasting time doing that will be the hobbyist, with a ton of plastic bins to put batteries in as they get test results back. Selling some cells that are too high or too low capacity to match the bulk to tested cells. The price of used cells is going to go thru the floor with new pack pricing going below $100/kwh

 

[NHK X]

Just got my cost estimate for FSD retrofit for my 2017 MCU1/HW2.5 X. No MCU2 upgrade for me at this time, just FSD computer solo. All my other attempts were rejected, first time to get the appointment to stick with a cost estimate. Of note I booked for the service center, and they actually switched it to mobile (seems to be their preference since they gave me no option to keep at service center).

I hope it goes through, if it does guess their late may estimate on their website is true. If anyone else has a similar build, might consider booking for late May or early June and see what happens.

 

[nuts]

At this rate i would have to pay them 10-15k just to get out because of the above + the loss from my investment into the car.

I seriously have some crazy level of respect for the early adopters....god bless you guys for getting this company off the ground.

Thank you for sharing. Your situation is exactly why it makes no sense to pay Tesla to not deliver software.

By the time FSD is "feature complete" and approved by regulators is probably still a few years out.

 

[Knightshade]

Thank you for sharing. Your situation is exactly why it makes no sense to pay Tesla to not deliver software.

By the time FSD is "feature complete" and approved by regulators is probably still a few years out.

The "approved by regulators" bit is BS.

There's no regulations holding back L2 systems. If they had something that could safely do intersections TODAY they could release it TODAY and just require the driver to keep paying attention while it does it.

They're not releasing it because they don't have it. Regulation don't enter into it.


Even if they wanted to release L3 or higher, that's already legal in a number of states as well (though certainly not all of em, but it's not like the car doesn't know what state it's in)

 

[nuts]

Even if they wanted to release L3 or higher, that's already legal in a number of states as well (though certainly not all of em, but it's not like the car doesn't know what state it's in)

i dont think anyone has worked out legality about whos at fault, what happens if there's a malfunction where you die, where you kill someone, or if the car goes haywire and mows down 20 people on a sidewalk while you are napping.

 

[chillaban]

The "approved by regulators" bit is BS.

There's no regulations holding back L2 systems. If they had something that could safely do intersections TODAY they could release it TODAY and just require the driver to keep paying attention while it does it.

They're not releasing it because they don't have it. Regulation don't enter into it.


Even if they wanted to release L3 or higher, that's already legal in a number of states as well (though certainly not all of em, but it's not like the car doesn't know what state it's in)

If you're talking about it in the US, the NHTSA technically does have arbitrary and total authority to deem any vehicle fit or unfit, unfit either meaning it cannot be sold to consumers or means it requires a mandatory recall to correct it.

You're right in that there's no specific regulations to L2 and L3 systems but there doesn't need to be. There is plenty of authority given to regulatory bodies that they can just issue an order as the mechanism to enforce rules they want to enforce about such systems.

Even for a boring old ACC/LDW system you are required to submit a statement of the mechanism of action to the NHTSA.

 

[SSonnentag]

As of yesterday afternoon my 12/2017 Model S 100D has both the HW3 and MCU2 upgrades.

It's like a whole new car. Autopilot is SOOOOOO much smoother on HW3 than it was on HW2.5. I'm running the same version 2020.12.11.5 firmware I was when I took the car in last Friday. Total cost was $2500 out the door. It really is more confidence inspiring to be able to see the road markers, cones, stop signs and road markings. I wasn't expecting to be able to see the reflectors along the road on the display. Did I mention how much smoother AP is now? It's a night and day difference. With the old computer the car acted like it was getting more information than it could keep up with, so the car was always playing catch-up. This made steering weave more, braking more abrupt and so forth.

The new MCU2 screen has a warmer color temperature than my old MCU1 screen. I prefer the look of the cooler one, but it's not a big deal. Having multimedia options when parked is going to be so nice!

If anyone is on the fence about upgrading to MCU2, don't hesitate. Just find a way to get over the loss of your AM/FM/SiriusXM if you use those. The benefits of everything else being smooth, fast and feature-rich is well worth it.

 

[Knightshade]

If you're talking about it in the US, the NHTSA technically does have arbitrary and total authority to deem any vehicle fit or unfit, unfit either meaning it cannot be sold to consumers or means it requires a mandatory recall to correct it.

You're right in that there's no specific regulations to L2 and L3 systems but there doesn't need to be. There is plenty of authority given to regulatory bodies that they can just issue an order as the mechanism to enforce rules they want to enforce about such systems.

You're confusing two different things here.


Yes, they could issue a NEW rule BANNING a given system. But that's not at all the topic.

Some people think they need to give PERMISSION to ALLOW a given L2 system.

They don't.

If Tesla wished to roll out 100% FSD does everything originally promised, but does it as an L2 system, TODAY, they could. Nothing legally stops this- and they don't require any 'regulatory approval' in the US to do it.


"Pending regulatory approval" is a red herring excuse here.


Now, Canada and the EU are another story (which is why they tend to get late, or crippled, versions of the same features the US already has)

i dont think anyone has worked out legality about whos at fault, what happens if there's a malfunction where you die, where you kill someone, or if the car goes haywire and mows down 20 people on a sidewalk while you are napping.

Not for L2 systems- no.

The driver is responsible.


For L3 or higher- nobody sells one yet- but so far at least a few companies who PLAN to eventually sell one have said they (the car company) would be responsible.... (Volvo said this years ago for example, and this liability is one of the reported reasons Audi decided to give up on L3 in the A8)

 

[chillaban]

You're confusing two different things here.


Yes, they could issue a NEW rule BANNING a given system. But that's not at all the topic.

Some people think they need to give PERMISSION to ALLOW a given L2 system.

They don't.

If Tesla wished to roll out 100% FSD does everything originally promised, but does it as an L2 system, TODAY, they could. Nothing legally stops this- and they don't require any 'regulatory approval' in the US to do it.


"Pending regulatory approval" is a red herring excuse here.


Now, Canada and the EU are another story (which is why they tend to get late, or crippled, versions of the same features the US already has)

I don't think the difference is as great as you expect it to be. The way that a company prevents an expensive recall and potential enforcement action from the federal government is basically a filing to the NHTSA that amounts to asking permission, and the NHTSA generally provides some guidance and feedback that is their unofficial word that they don't intend on taking any action if this is released.

I used to work at an auto supplier and we prepared such paperwork ALL THE TIME for adaptive cruise control systems, before they hit the market. Even little details like changing the stop-and-go timeout from 3 seconds to 5 seconds were things that automakers ran past the NHTSA.

You can also just not ask for permission and push out features, that is very true, but in practice an automaker has to be incredibly dumb to consider doing so because it opens them to far more risk of being subjected to recalls and other regulatory actions and compromising their goodwill with regulatory bodies. And unfortunately the regulatory model in the US does make it so that if you're not on the NHTSA's "good side", you're going to have a hell of a harder time shipping anything in the future.

 

[Knightshade]

I don't think the difference is as great as you expect it to be. The way that a company prevents an expensive recall and potential enforcement action from the federal government is basically a filing to the NHTSA that amounts to asking permission, and the NHTSA generally provides some guidance and feedback that is their unofficial word that they don't intend on taking any action if this is released.

Can you cite any examples of Tesla having submitted such filings in the past asking permission when adding features to the existing AP/EAP/FSD features?

To my knowledge they've not done this- but I'm always open to being wrong if there's evidence of it.

 

[chillaban]

Can you cite any examples of Tesla having submitted such filings in the past asking permission when adding features to the existing AP/EAP/FSD features?

To my knowledge they've not done this- but I'm always open to being wrong if there's evidence of it.

Such filings are considered confidential and trade secrets, as they detail specific algorithms and under-the-hood implementation details. I'm not at liberty to share any of the filings that I've worked on before but I'm not sure if you can FOIA the NHTSA for Tesla's filings.

But here's an example of a response from the NHTSA to BMW asking for their Smart Summon equivalent feature: https://www.nhtsa.gov/sites/nhtsa.dot.gov/files/bmw-response-01042016.pdf

As you can see, the questions are generally phrased as "Does my car based off my description still comply with the FMVSS?" (AKA is it still legally classified as a street legal car, defined by a document written 70 years ago that has no notion of ADAS features)

The NHTSA's Q&A response is in practice construed as their blessing.


(So no, I cannot tell you when and what letters Tesla has submitted to the NHTSA regarding Autopilot features, as that's not public record. NHTSA selectively publishes relevant answers as they see fit. But despite there being officially no regulatory process, there is in effect an unofficial way to seek NHTSA's blessing which is part of most automakers' process)

 

[MarcG]

By the time FSD is "feature complete" and approved by regulators is probably still a few years out.

The "approved by regulators" bit is BS.

There's no regulations holding back L2 systems. If they had something that could safely do intersections TODAY they could release it TODAY and just require the driver to keep paying attention while it does it.

They're not releasing it because they don't have it. Regulation don't enter into it.

I agree with both quotes above. I don't think @nuts was trying to say is that "feature complete" is predicated on regulatory approval.

I think "feature complete" is at least a year away, then I do believe Tesla plans on releasing it when they feel it's ready, but it will require the driver to pay attention (therefore be responsible to take over and ultimately liable should an accident happen). I also don't remember Tesla ever stating that they are waiting on regulatory approval to release all FSD features.

That being said, I feel like the regulatory bit is likely at least a year beyond the "feature complete" point, and possibly many years later, depending on jurisdiction.

I for one do look forward to the day FSD is feature complete, even if I have to pay attention and be responsible for the car's action. That's the price of progress. Not worth $100,000 yet, though

 

[chillaban]

To be clear, I think Tesla's main roadblock to FSD is that their software isn't ready yet. I agree the "pending regulatory approval" wording is misleading because it makes it sound like Tesla has everything ready but just isn't getting permission yet.

I'm just trying to point out that even though the NHTSA is officially hands-off and does not have an explicit authorization procedure, in practice companies do not want to be on the receiving end of NHTSA enforcement action and will seek permission from the NHTSA in a pretty unofficial process of sending letters asking if their vehicle or if a specific feature still complies with the FMVSS.

 

[nuts]

The biggest problem with Tesla charging this as an additional feature on top of a depreciating asset like a vehicle is that it doesnt transform the vehicle into an appreciating asset. As these cars get older the technology will be expensive to repair and ultimately like we've seen with this whole HW2.5 >> HW3.0 upgrade path, the tech will also become outdated and insufficient.

To then add a $100,000 (or whatever) price tag to the equation to a feature that is limited to use for one vehicle and tied as a feature that will be removed upon transfer of ownership, makes it a very ugly value proposition for anyone outside of commercial use.

On the other hand, if they do something reasonable (which is not in Tesla's DNA) like tie the FSD as a perpetual license that follows the owner to any new Tesla they purchase, then I can see how they can charge $20k for the feature. It would also add to customer loyalty, as more competitors enter the space in the next 10 years.

Keep in mind, they already are asking for ($3k + $7k) for FSD --- so $10k plus tax --- or $11k in CA for a feature that may not even be used in a car's lifespan that people are buying them for. For example, my 2018 M3 - I have no confidence that I would use this feature before i get rid of the vehicle.

 

[chillaban]

I don’t personally need it to appreciate the cost of my vehicle. As long as it gets me the features I want and is economically more viable than getting a new Tesla then I am open to retrofits for a fee.

 

[nuts]

I totally get it, but i'm just going by all the marketing and sales rhetoric that Elon has come out with for this HW3 / FSD stuff.