Investor Engineering Discussions

[MC3OZ]

It is very interesting the Jordan manages to extract many clues from Battery Day suggesting Tesla will probably make LFP in 4680 format.

At the time of Battery Day LFP was definitely the preferred option for the "entry level" chemistry and I still can't understand Tesla making 3 TWh of cells without making some cells an "entry level" chemistry.

A Jordan also hit on a good point that a 4680 based LFP pack might be easier to heat in cold weather. cold being the major limitation of LFP. But LFP cold weather performance would not be an issue for energy storage, the battery could be in a well insulated enclosure, with a good heating system.

Here are some quotes from the Q4 2021 earnings call on Manganese:-

Yeah. To be clear, we do think that old stationary storage, Powerwall and Megapack, will be -- will transition to an iron-based system, basically a non-nickel system. Manganese is also, you know, could be part of the future, but primarily iron. It just comes out iron-nickel.

Talking in the context of grid storage above.

And like I said, with some -- manganese is like a wild card. There's also not a lot of manganese. And I should say like we did short-change the energy business last year, and that vehicle took priority over the energy side.

Again about energy storage.

Yeah, we don't use 4680 at all for the iron upgrade cells.

Lots of commentary around this part of the call talking about using different form factors for iron cells.


I thought I heard something in the call about Iron and Manganese being the options to scale to TWh of cell production, bit looking at the transcript that isn't there. It was in some Twitter discussion....

Here is Limiting Factor's tweet in response:-

https://twitter.com/x/status/1486794171906240514

I can't find Elon's original tweet, but he didn't respond to the question.

Overall trying to deduce what Tesla is planning from a forensic analysis of statements by Elon and Drew is problematic.

The one-off Twitter post on Manganese was after earnings call? Elon posts a lot of stuff on Twitter, it isn't always significant.

Bottom line - I still have no idea what Tesla plans are in terms of chemistries and form factors.

But assuming LMNO works, I would use LMNO for Powerwall and "entry-mid level vehicles" and LFP for Powerpack and Megapack and possibly entry level vehicles..
The advantages LMNO seems to have over LFP are higher energy density and better performance at 5C charge/discharge rate, with about the same cycle life and cost.

The disadvantages for LMNO seem to be that it is new and relatively unproven and Manganese is not as abundant as iron, also not a thermally stable as LFP but more thermally stable than NCA.
Bottom Line here - we don't know if LMNO is proven and ready for commercial production, but it might be the Manganese chemistry mentioned in Battery Day. Or in the group of potential Manganese chemistries.

 

[mongo]

All good points. Has anyone actually found the entirety of the patent? I looked but couldn't find it (free access of course). From the blurbs that I've seen I don't see any infringement as Ford is describing a process that is different than Tesla.

As this is engineering, we can further discuss in that thread and then bring the conclusions back here for business impact.

Patent (images don't work (on my phonecat least))
United States Patent: 11247693

 

[Discoducky]

Patent (images don't work (on my phonecat least))
United States Patent: 11247693

Thank you! So my read, I'm not a patent lawyer, but I've worked with many on filing my patents, so I'd say that it looks like the heart of the patent is "coded wireless activation" via key-fob and they want to adjust windows, sunroof, seat positions, heat/cool and be able to save those.

Hmm, doesn't that sound like prior art? YEP! That is simply a setting profile. Saying that this all for a pet, IMO, does not constitute novel. If not novel, then it is not patent-able.

The RFID detection is cool, so that could be novel. But, I'd bet someone already has a patent that covers an application for detection of an RFID tag inside a vehicle.

 

[mongo]

Thank you! So my read, I'm not a patent lawyer, but I've worked with many on filing my patents, so I'd say that it looks like the heart of the patent is "coded wireless activation" via key-fob and they want to adjust windows, sunroof, seat positions, heat/cool and be able to save those.

Hmm, doesn't that sound like prior art? YEP! That is simply a setting profile. Saying that this all for a pet, IMO, does not constitute novel. If not novel, then it is not patent-able.

The RFID detection is cool, so that could be novel. But, I'd bet someone already has a patent that covers an application for detection of an RFID tag inside a vehicle.

yah
RFID inside a car is how (dead-battery) keyless start works.
Paw button is remote start (practically the same climate profile too).
I will say, novel is applied as a coarse filter for a patent and doesn't neccesarily mean it would be able to withstand a lawsuit.

 

[Doggydogworld]

It is very interesting the Jordan manages to extract many clues from Battery Day suggesting Tesla will probably make LFP in 4680 format.

Then Elon shut that idea down cold in the earnings call, lol. So Jordan added a preamble mansplaining why Elon didn't really mean what he said!

At the time of Battery Day LFP was definitely the preferred option for the "entry level" chemistry and I still can't understand Tesla making 3 TWh of cells without making some cells an "entry level" chemistry.

3 TWh was aspirational. Like a million Robotaxis.

A Jordan also hit on a good point that a 4680 based LFP pack might be easier to heat in cold weather.

Except it's not. Tesla says 4680 structural pack thermal control will not happen through the sidewalls like the 18650 and 2170 packs (i.e. no more snake tubes). Cylindrical gives you no advantage in bottom surface area, so there's no heating/cooling advantage. Of course there was that one 4680 pack mock-up in Berlin that still had snake tubes, but that's hopefully just artistic error.

 

[MC3OZ]

Then Elon shut that idea down cold in the earnings call, lol. So Jordan added a preamble mansplaining why Elon didn't really mean what he said!

3 TWh was aspirational. Like a million Robotaxis.

Except it's not. Tesla says 4680 structural pack thermal control will not happen through the sidewalls like the 18650 and 2170 packs (i.e. no more snake tubes). Cylindrical gives you no advantage in bottom surface area, so there's no heating/cooling advantage. Of course there was that one 4680 pack mock-up in Berlin that still had snake tubes, but that's hopefully just artistic error.

Elon is confident about 3TWh, like he is with a million Robotaxis.

When did Tesla say 4680 cooling would not happen though sidewalls?

The pack in Berlin was just a mock-up.

My posts are based mainly on comments Elon and Drew have made.

It is true iron chemistry featuring in battery day might have had nothing to do with 4680s.

But that suggests additional questions-
1. Is LFP part of the 3TWh?
2. Is Tesla making LFP in some other form factor.

My best guess is that Tesla intends to make 3TWh of 4680 cells.

 

[JRP3]

Of course there was that one 4680 pack mock-up in Berlin that still had snake tubes, but that's hopefully just artistic error.

That's a lot of effort to showcase their pack structure and make an error. Would have been much easier to make it without the tubes and just infer bottom cooling if that were the case.

 

[Doggydogworld]

That's a lot of effort to showcase their pack structure and make an error. Would have been much easier to make it without the tubes and just infer bottom cooling if that were the case.

"Join the Cell Team" page doesn't show any cooling tubes. This clip came from Battery Day.

Tesla showed another slide with "cooling channels" below the cells.

Finally, the problem with fat cylinders is getting the heat through all the layers to the sidewalls. The tabless design directly moves heat from each layer down to the bottom of the cell via the thermally conductive copper. Other companies, e.g. Rivian, say this thermal path is more efficient even without the tabless design.

Jordan's structural pack video spells out the bottom plate cooling approach in more detail.

 

[mongo]

"Join the Cell Team" page doesn't show any cooling tubes. This clip came from Battery Day.

View attachment 771257

Tesla showed another slide with "cooling channels" below the cells.
View attachment 771272

Finally, the problem with fat cylinders is getting the heat through all the layers to the sidewalls. The tabless design directly moves heat from each layer down to the bottom of the cell via the thermally conductive copper. Other companies, e.g. Rivian, say this thermal path is more efficient even without the tabless design.

Jordan's structural pack video spells out the bottom plate cooling approach in more detail.

Is there anything to support calling the lower voids are cooling channels? Before image doesn't show cooling either.
How does Rivian create a thermal path if the cell isn't tabless?

 

[Doggydogworld]

Is there anything to support calling the lower voids are cooling channels? Before image doesn't show cooling either.
How does Rivian create a thermal path if the cell isn't tabless?

Outside engineers assume they're cooling channels. No one has thought of a better reason to put channels there, and since the other pics show no snake tubes the bottom plate is kind of the last remaining option. The "before image" shows the cells lined up in pairs. The larger spaces beside each pair are for the snake tubes.

I don't know how Rivian creates a thermal path. It's probably not as good as tabless, but still better than side cooling. Lucid also uses axial (bottom plate) cooling for their 2170s, but only in a single layer (Rivian puts cells both above and below the cooling plate). Lucid has a lot of battery experience from Formula E, and says their testing shows axial is better for 2170. I don't know how Tesla missed that boat. Maybe they measured it for the much skinnier 18650 and just assumed the results would hold for 2170?

 

[mongo]

Outside engineers assume they're cooling channels. No one has thought of a better reason to put channels there, and since the other pics show no snake tubes the bottom plate is kind of the last remaining option. The "before image" shows the cells lined up in pairs. The larger spaces beside each pair are for the snake tubes.

I don't know how Rivian creates a thermal path. It's probably not as good as tabless, but still better than side cooling. Lucid also uses axial (bottom plate) cooling for their 2170s, but only in a single layer (Rivian puts cells both above and below the cooling plate). Lucid has a lot of battery experience from Formula E, and says their testing shows axial is better for 2170. I don't know how Tesla missed that boat. Maybe they measured it for the much skinnier 18650 and just assumed the results would hold for 2170?

Gotcha.
Seems like bottom cooling would lose a lot of energy to the environment. Also more at risk to bottom damage. Side cooling with an insulated case seems most efficient (assuming the environment is the opposite temperature delta you want) and is.what the Berlin crossection showed (though I'm still puzzled by how the BMS connects to all groups). The voids would then be structure/ insulation/ protection.
I supposed they could do a triple wall extrusion with crush zone, then channels? (Not what picture shows).

 

[MP3Mike]

what the Berlin crossection showed (though I'm still puzzled by how the BMS connects to all groups). The voids would then be structure/ insulation/ protection.

Have you viewed the Ingenrix or Munro tear down videos of the Plaid battery pack? Cooling runs down one said of the pack, and BMS connections run down the opposite side.

 

[Cal1]

Sure wish someone would explain the differences (if any) between a LR and a Plaid battery. Lot of people claim they are the same but I've not seen anyone actually open up a 2021 LR and verify it is the same. I did find an article Here's Why The Battery Pack of Tesla Model S Plaid is An Electrification Masterpiece claiming Plaids are 18650's and LR are 2170's. But he's the only one and I'm not sure this is true.

Sandy verified 18650's in Plaid. Funny no one has come out and said exactly what gives Plaids more power than a LR. Elon said that the carbon fiber wrapping is needed to keep the windings from exploding at high rmps. But never said what magic gives the extra umph. Is it all software? I assume Plaids have more powerful inverters. At least that was the case in my old P85. Sandy's battery episode never said anything was Plaid specific.

 

[mongo]

Have you viewed the Ingenrix or Munro tear down videos of the Plaid battery pack? Cooling runs down one said of the pack, and BMS connections run down the opposite side.

Yeah, but Plaid is 5 modules. With one big block, there aren't enough terminals at the sides for all the groups. I count 24 cells per row by 40, but a 96s10p pack needs to be 20 wide with BMS on both side to connect. That setup would create high potential between adjacent groups due to the ordering . They could be 80s12p, but then the pack voltage is a lot lower.
Seems like the pack will some internal grouping/ discontinuities.

 

[MP3Mike]

Yeah, but Plaid is 5 modules. With one big block, there aren't enough terminals at the sides for all the groups. I count 24 cells per row by 40, but a 96s10p pack needs to be 20 wide with BMS on both side to connect. That setup would create high potential between adjacent groups due to the ordering . They could be 80s12p, but then the pack voltage is a lot lower.
Seems like the pack will some internal grouping/ discontinuities.

Well the Model 3/Y is only 4 modules, that are longer than the Plaids, and the BMS connection are only on the short side of the module just like in the Model S Plaid pack. I suspect that a lot is hidden under the blue intumescent "goo".

 

[JRP3]

Tesla showed another slide with "cooling channels" below the cells.

We don't know that those are cooling channels, frankly they seem excessively large to be such. I'm still inclined to favor the most recent mock up over older CGI.

 

[mongo]

Well the Model 3/Y is only 4 modules, that are longer than the Plaids, and the BMS connection are only on the short side of the module just like in the Model S Plaid pack. I suspect that a lot is hidden under the blue intumescent "goo".

The Plaid modules are 22s72p with the 22 being the short axis of the module/ long axis of the pack (110) , so the BMS can measure all groups. 4680 cells won't pack as conveniently.
3 and I think Y use a PCB with a seperate flat flex PCB to connect to the groups. Seems like that would mess up the structural pack aspect, but may be the way they need to go.

 

[Oil4AsphaultOnly]

Sure wish someone would explain the differences (if any) between a LR and a Plaid battery. Lot of people claim they are the same but I've not seen anyone actually open up a 2021 LR and verify it is the same. I did find an article Here's Why The Battery Pack of Tesla Model S Plaid is An Electrification Masterpiece claiming Plaids are 18650's and LR are 2170's. But he's the only one and I'm not sure this is true.

Sandy verified 18650's in Plaid. Funny no one has come out and said exactly what gives Plaids more power than a LR. Elon said that the carbon fiber wrapping is needed to keep the windings from exploding at high rmps. But never said what magic gives the extra umph. Is it all software? I assume Plaids have more powerful inverters. At least that was the case in my old P85. Sandy's battery episode never said anything was Plaid specific.

I don't recall if it's in the same video, but Sandy did mention that the u-channel cooling tubes in the plaid allows more consistent cooling/heating amongst the cells in the module. With a more consistent temperature, the voltage between the cells would be more consistent, so they can be more aggressive with the current (both to charge and discharge). The old cooling design wasn't able to keep all the cell temps as close together, so they had to throttle the charge current earlier to keep a larger temperature buffer between readings.

 

[Cal1]

Yes but I don't know if this is Plaid specific or on LR also. Still no idea what gives Plaid additional power except carbon wrapped rotors. He didn't say anything about a more powerful inverter or really any power adding options.

Maybe its more powerful because I believe it is!

 

[mongo]

Yes but I don't know if this is Plaid specific or on LR also. Still no idea what gives Plaid additional power except carbon wrapped rotors. He didn't say anything about a more powerful inverter or really any power adding options.

Maybe its more powerful because I believe it is!

The rear drive unit being two motors does wonders for both power output and power band. Each motor only needs about half the torque output, so much opportunity for gearing and windings.