4680 cell design, chassis integration & factory discussion for investors

[GhostSkater]

How so? They were cheaper than the alternatives until the other OEMs cancelled all of their orders and made a glut of batteries available on the market. And they said that they think the 4680s will be cheaper than even that surplus by the end of this year.


All of the battery day goals were long-term goals. I think through 2026, so we still have plenty of time for them to refine the 4680s.

From what I have seen the Gen2 4680 already has energy density slightly greater than the current 2170 cells. (And I think to be fair we really should compare capacity after ~1-year. It seems like most Panasonic 2170s lose ~5% in the first year, while the Gen1 4680s lose almost nothing in the first year. We will have to see how the Gen2 4680s perform in this respect.)


Of course there are people that know. I think the issue is that Tesla is new to battery manufacturing, and they started with a "simple" formulation as they learn. (Concentrating on new physical manufacturing advancements.) Like they didn't put silicon in it to start with. (Though apparently they did for Gen2.) Is it the "Tesla silicon" they talked about at battery day? I don't think we know yet, but those details may come out now that cells have been harvested and sent off for analysis.

If Panasonic had stepped up and increased 2170 production in the US the 4680 likely wouldn't have been necessary in the first place. But Panasonic seems to be slow walking all of their expansions and new plants in the US. As far as we know there still aren't any other Tesla compliant cells that are qualified for the tax credits other than the Panasonic 18650s/2170s and the Tesla 4680s.

Other new battery manufacturers, like Northvolt, seem to be doing a much worse job of getting production up and running than Tesla is.

Thanks for putting into words what I didn't have the patience to type yet again since all that has been discussed many times in other topics

@AlanSubie4Life nothing against, just a disagree with your take

 

[AlanSubie4Life]

Thanks for putting into words what I didn't have the patience to type yet again since all that has been discussed many times in other topics

@AlanSubie4Life nothing against, just a disagree with your take

Fair enough. I want 4680 to be a success but it's a cell I would tell everyone to avoid at the moment. I would never want that cell in any vehicle I owned. Hopefully Gen3 will be awesome.

 

[AlanSubie4Life]

Of course there are people that know.

I meant outside of Tesla.

while the Gen1 4680s lose almost nothing in the first year.

Can you link to a few examples? That's news to me. Need several datapoints from the energy screen in several vehicles with information on their usage, location, charge habits, etc. .

Here's an example at 65.8kWh for a 2023 Model Y AWD 4680. So 4.7% in less than two years (assuming 69kWh starting - I don't have an SMT capture of AWD 4680 brand new but I think 69kWh is close to correct). Seems like fairly typical capacity loss.

Battery health tracking SS

We live in Central Florida, so that may have helped. 16 miles each way to/from work with very few stop lights and good roads. Yes, had 265 miles at 100% when new.

teslamotorsclub.com

I think through 2026, so we still have plenty of time for them to refine the 4680s

Hopefully you're right. Remember they've got to be competitive with and better than what Porsche etc are using at that point otherwise those abysmal 10% challenge results are going to really start to stick out.

 

[Sweed269]

If you take the same number of cells, it doesn't matter if you arrange them for 400v or 800v. They both charge at the same C rate. Maybe in the future we will need more than 250kw but right now almost no cars can sustain this level.
The bigger the cell, the harder it is to cool. The 4680 will never be better than a 2170, but may match its performance.

 

[AlanSubie4Life]

If you take the same number of cells, it doesn't matter if you arrange them for 400v or 800v. They both charge at the same C rate

Yeah it's just to keep the current down so that you're not current limited by the charger. Has nothing to do with the cells as long as they have a C rate that requires current high enough to current limit the charger in a 400v wide config. Higher C cells are more likely to do this of course - so the higher the C rate the more likely you are to benefit from 800V.

Maybe in the future we will need more than 250kw but right now almost no cars can sustain this level.

Yeah that's a problem. Would be good to push that to 40-50% SOC for an 80kWh pack (so 3C to 40-50% - I do think there's little need for charging faster than those rates to that SOC). But even that might not be enough for CT.

The 4680 will never be better than a 2170

This seems like a small flaw in the plan. I sure hope that is not correct.

 

[MP3Mike]

Hopefully you're right. Remember they've got to be competitive with and better than what Porsche etc are using at that point otherwise those abysmal 10% challenge results are going to really start to stick out.

Options are great, and tests like the 10% challenge are good to help people find a vehicle that will work for their needs. But not every option has to be great for road tripping. A lot of people pretty much never go on road trips, and would be happy with a less expensive car that has slower peak charging. I am one of those people. I am currently at 86% AC charging, and the only reason I have DC charged at all was because it was free or cheaper than my home charging. (1000 miles of Supercharging that came with the purchase, and when EA had their big free vend event because of their "network upgrade".)

As far as Gen2 4680 performance, we don't know where it actually is yet, as there is supposed to be an update coming soon that increases charging speed on the Cybertruck. But it seems like it is already pretty good, as it basically kept up with the Chevy Silverado that has pretty good charging performance and double the battery capacity on the coast-to-coast trip.

 

[GhostSkater]

Fair enough. I want 4680 to be a success but it's a cell I would tell everyone to avoid at the moment. I would never want that cell in any vehicle I owned. Hopefully Gen3 will be awesome.

Agreed, waiting to see what the new charge curve improvement will bring, specially now that we know they added a bit of Silicon, if my math is correct, Cybertruck will do flat 250 kw up to 50% with that, then it's a matter of how big the peak will be and how long it keeps it once higher power Superchargers are available

But honestly, while low fast charging times are nice, specially in a marketing campaign, being the top of the game on it isn't that much important anymore, obviously you can't totally suck at it like Toyota and Bolt, but 30 mins 10-80% seems completely fine, this coming from someone that drove a Tesla for the first time a few months ago and did a 4000 km roadtrip and its was one with the LG cells, not once I waited for the car to charge

Tesla is much more about volume, as evident by having some packs that charge way better than others

 

[AlanSubie4Life]

But honestly, while low fast charging times are nice, specially in a marketing campaign, being the top of the game on it isn't that much important anymore, obviously you can't totally suck at it like Toyota and Bolt, but 30 mins 10-80% seems completely fine, this coming from someone that drove a Tesla for the first time a few months ago and did a 4000 km roadtrip and its was one with the LG cells, not once I waited for the car to charge

I’ve taken many road trips in my Tesla, doing as much as 1100 miles in a day. I have waited on the car many times. I would not want slower charging. I have taken my ICE vehicle on several occasions on long road trips, and one of the factors was the wait time and hassle of charging. And I don’t even have that great a range in that ICE vehicle - only 350-400 miles on a tank.

 

[MP3Mike]

I’ve taken many road trips in my Tesla, doing as much as 1100 miles in a day. I have waited on the car many times. I would not want slower charging. I have taken my ICE vehicle on several occasions on long road trips, and one of the factors was the wait time and hassle of charging. And I don’t even have that great a range in that ICE vehicle - only 350-400 miles on a tank.

Tesla does seem to lead in actual trip performance. If you look at Bjorn's 1000km challenge, Tesla takes 5 of the top 10 spots. (And three of the top 10 spots went to Nio with battery swapping.)

Hopefully you're right. Remember they've got to be competitive with and better than what Porsche etc are using at that point otherwise those abysmal 10% challenge results are going to really start to stick out.

Efficiency really matters too... It is not great even if you charge fast if you waste the energy with poor efficiency. The best Porsche he tested is spot #18. (He hasn't tested their brand new ones yet.)

 

[AlanSubie4Life]

Efficiency really matters too... It is not great even if you charge fast if you waste the energy with poor efficiency. The best Porsche he tested is spot #18. (He hasn't tested their brand new ones yet.)

Yep. 10% challenge conducted correctly is the way to go.

Tesla does seem to lead in actual trip performance

I would think so. Needs improvement! Need to stay on top. We’ll see with the latest Porsche.

Relevant to this thread, 4680 is not helping things.

 

[AlanSubie4Life]

Here's an example at 65.8kWh for a 2023 Model Y AWD 4680. So 4.7% in less than two years (assuming 69kWh starting -

More like starting at 68kWh it seems, so 3% for a car a little over a year old kept at 50%.

This is more or less in line with other cells. I would not be surprised at small differences in capacity loss (2170C is better than 2170L) and it may be slightly better than other cells but it is inconsequential for what we were discussing. Which was the very low density of the 4680. (68kWh rather than 69kWh also makes it less dense…)

But it seems like it is already pretty good, as it basically kept up with the Chevy Silverado that has pretty good charging performance and double the battery capacity on the coast-to-coast trip.

I really did not think it was that impressive. Basically someone dead set on winning and running the CT on vapors managed to keep up with a Silverado using an inferior charging network.

And how would it have done compared to Model S LR? (They should do the same.). Or Model 3 Highland if that is better (probably not, but might be, don’t pay attention).

Hopefully the update will be really good and then everyone will be talking about how great the Cybertruck is.

Improvement is needed on the 4680 regardless.

 

[kpanda17]

Maybe. Talk is cheap, we'll see how it turns out. It's cheap for a reason. Look at the charging curves!!!! That is not what you want in a road trip vehicle. Put that into a Model 3 Highland and it'll turn it into a complete disaster of a road-trip car. The step down from 2170C to 2170L was bad enough. They're going the wrong way! Put high quality cells in, and go to 800V (completely unrelated to the cells) to keep the currents down. I'm not sure we're ever going to get v4 Superchargers at this rate. It's bizarre people think it is fast enough. I have a super duper fast charging 2018 Model 3, and it's quite workable for a road trip car (in spite of its apparent high consumption Performance tuning when compared side by side with AWD) but it could most definitely be faster, since I do wait on the car on occasion. Need to be able to do longer distances, with faster charging at stops.

It's really infuriating and frustrating that things have been backsliding since 2020, the golden age.


Can you link to the analysis? It just seems that in real-world form factors, it's not there. That's what I am referring to. If I take the Model Y AWD and scale up the battery (~69kWh IIRC) by the Gen1 to Gen2 density improvement, it still ends up being lower capacity than the 80-81kWh 2170L Model Y LR AWD (and it's not close). How can that be explained? You can't look at cell density, of course.


All the more reason for them to be the best & safest cells in the world. Make Batteries Great Again.

When 4680s are shipping in every 85kWh Model Y LR AWD, we'll know we've arrived.

But don’t forget, as Elon said, 4680s will be cost competitive by years end
Is that vs LFP? If that’s the case, which I don’t think that is true, it would be a major industry disruption.
Think the plan is 4680 vs 2170/NCM cost competitive

Side note, hope Andrew Bs plan for gen 3, 3a, and beyond to a full dry, high energy density is progressing at Tesla without him
Tesla 4680s could become the worlds best and even Tesla supplying some OEMs with batteries
Would be a cool source of revenue

 

[AlanSubie4Life]

But don’t forget, as Elon said, 4680s will be cost competitive by years end

I think we know with reasonable certainly how that will turn out.

Let’s just see how good the charging update is. Maybe the cells are better than what we have seen so far? One tiny step at a time.

I think we can only say that Baglino was unable to deliver a competitive cell by 2024. Why, I have no idea.

It would be great to be the world’s best but others are not standing still.

 

[MP3Mike]

Yep. 10% challenge conducted correctly is the way to go.

The 10% challenge does have a significant flaw though. It doesn't test the repeatability of the performance. Some cars have temperature management issues, and only the first charge on a trip is really good. That is where Bjorn's 1000km challenge has an advantage. And Bjorn customizes the charging plan to each vehicle the best he can to get the best result, which I assume a normal drive would do as well.

 

[AlanSubie4Life]

Tesla does seem to lead in actual trip performance. If you look at Bjorn's 1000km challenge,

That is where Bjorn's 1000km challenge has an advantage. A

Note that the top Teslas in the test are 18650.

And there is a 2170L in 10th.

No 4680. (And that is not just because it has not been tested - that is evident from the other results!)

This was my point. Teslas with 18650 in particular have decent road trip performance, still lagging behind ICE by about 20-40 minutes in this test. 40 minutes for 2170 seems very roughly right for an 8-hour drive if everything goes perfectly.

And competitors are going to get better. They already are. We’ll see that soon in these tests.

Tesla needs to make the 4680 better than 2170, not equal to it. The goal should be to exceed 18650 charging performance by a factor of 1.5 or so. (Maybe slightly less, have to actually do math to quantify it….but in any case after that there would be little benefit to further improvements.)

It’s all very well to talk about 4680 v2 charge profile improvements - something I have held as possible from the very beginning - but we have not seen them yet. At this point I expect them, but I expect them to be underwhelming. We’ll see. If Cybertruck is pushing 250kW at 50% SOC, I’ll be surprised and wrong - there will then be some hope for Gen 3 if they can get density up.

Bjorn’s current results for reference:

 

[Sweed269]

The 4680 may match, but will never exceed the 2170s thermal or discharge rate with similar chemistry. It was designed to be cheaper to produce/ build packs and more reliable. It simply doesn't have the ability to shed heat from the core fast enough.
Straight from a Tesla engineer working on the 4680.

 

[MP3Mike]

Tesla needs to make the 4680 better than 2170, not equal to it.

No, it doesn't. The primary goals for the 4680 are to be easier/faster to manufacture and less expensive.

 

[AlanSubie4Life]

No, it doesn't. The primary goals for the 4680 are to be easier/faster to manufacture and less expensive.

I mean I guess as long as they still plan for a mix of cells that is fine. They can put the 4680s in the cheap commuter cars I guess.

Cybertruck though. Yikes. That is where is gets really bad. I guess they can go to 18650 there.

 

[MP3Mike]

I mean I guess as long as they still plan for a mix of cells that is fine.

Of course they are. That is why they have said that they will buy every compliant cell that suppliers can make. (And have shown slides with at least three different battery types being used depending on use case.)

Cybertruck though. Yikes. That is where is gets really bad. I guess they can go to 18650 there.

There isn't enough 18650 capacity available to use them for the Cybertruck, and I don't think anyone is planning to expand 18650 production.

I think it will be interesting to see how Rivian's proposed 4695 cells perform, but we are at least a couple years from seeing that. And who knows what the Tesla 4680s will be like by then. (That is about when Tesla predicted that they will have achieved everything they presented at Battery Day.) Or if Tesla will have moved on to something else by then.

 

[MP3Mike]

It seems that LG is only a couple months away from starting 4680 production:

S. Korean EV battery giants gear up for 46mm cylindrical battery competition

S. Korean EV battery giants gear up for 46mm cylindrical battery competition LG Energy Solution, Samsung SDI ready to launch full-scale mass production

www.chosun.com

It will be interesting to see where those cells go and how they perform compared to Tesla's 4680s.

But it seems like their US battery plant is still a couple years off...