Investor Engineering Discussions

[Gigapress]

My comments about single crystal cathodes relate to this patent.

Single crystal is one important part of the 100 year battery.

Limiting Factor researched a lot of great stuff like this before battery day.

When it wasn't included in battery day, it has been largely forgotten about.

It's very likely that Tesla didn't show all their cards during Battery Day. They didn't even talk much about advances in cathode production whereas they spent a few slides covering anode production.

A couple years ago in a presentation that was briefly posted on YouTube video before being taken down, Dr. Dahn had said that defects in cathode crystal structure was actually a big cause of battery degradation because the cracks multiply over time as the battery is cycled, repetitively straining and relaxing the cathode material, gradually imprisoning more Li atoms while increasing internal electrical resistivity.

Drew and Elon did discuss the importance of the cathode material retaining its structural integrity, but the solution presented was to use novel coatings and dopants.
Pretty suspicious that Tesla didn't mention anything about monocrystalline cathodes as a solution to the same problem, despite the fact that Tesla has a patent for it and their main academic research partner had been hyping up the importance. Notably, their slide with a microscopic image of their cathode appears to show a multicrystalline structure. Maybe that means single-crystal is coming in the future but not early versions of 4680s.

 

[JRP3]

TL;DW - due to sunk costs, minor 800V benefits aren't worth it for Tesla. Of course if Tesla were 800V and the others 400V it'd prove they were "5 years ahead", lol.

If charging speed is limited at the cell level by C rate what significant advantage do you imagine is provided by running at 800V?

 

[MC3OZ]

It's very likely that Tesla didn't show all their cards during Battery Day. They didn't even talk much about advances in cathode production whereas they spent a few slides covering anode production.

A couple years ago in a presentation that was briefly posted on YouTube video before being taken down, Dr. Dahn had said that defects in cathode crystal structure was actually a big cause of battery degradation because the cracks multiply over time as the battery is cycled, repetitively straining and relaxing the cathode material, gradually imprisoning more Li atoms while increasing internal electrical resistivity.

Drew and Elon did discuss the importance of the cathode material retaining its structural integrity, but the solution presented was to use novel coatings and dopants.
Pretty suspicious that Tesla didn't mention anything about monocrystalline cathodes as a solution to the same problem, despite the fact that Tesla has a patent for it and their main academic research partner had been hyping up the importance. Notably, their slide with a microscopic image of their cathode appears to show a multicrystalline structure. Maybe that means single-crystal is coming in the future but not early versions of 4680s.

View attachment 810272

View attachment 810274

Some of the slides used on battery day were probably stock-photos lifted from other sources.

And for various reasons they might not reveal all that they intend to do, or the timing.

The million mile battery didn't feature becuase battery day was all about lowering costs and being able to scale rapidly.

It is a reasonable guess that a million mile battery might cost more and might be slower to produce. Unless Tesla gets serious about V2G, most cars don't need a million mile battery.

The one application where cost per cycle and long life is very important is grid storage.
If a battery costs 30% more, but does 5X-10X more cycles the enonomic advantage is odvious.

But scaling cell and vechicle production volumes is likely to be a higher priority. The million mile battery can being kicked further down the road would not surprise.

Jeff Dahn seems to be coming up with improved versions of the million mile battery all the time. They can also pick and choose which improvements to deploy.
Leaving out single crystal might reduce complexity and save money, but ultimately if they want to make the longest lasting battery possible, it probably has to be single crystal.

 

[Gigapress]

Some of the slides used on battery day were probably stock-photos lifted from other sources.

Good point, I didn’t think of that.

The million mile battery didn't feature becuase battery day was all about lowering costs and being able to scale rapidly.

It is a reasonable guess that a million mile battery might cost more and might be slower to produce. Unless Tesla gets serious about V2G, most cars don't need a million mile battery.

At 2019 Autonomy Day, Elon was asked to “could you just clarify how you get the battery packs to last a million miles?”

Elon: “The million-mile life is basically just about getting the cycle life of the pack—you need basically to—you know, you need basically you know on the order of—let’s say you’ve got—basic math—if you’ve got like a 250 mile range pack you know you need 4,000 cycles. So, very achievable. We already do that with our stationary storage…”

For context, he had also just before giving this answer discussed prioritizing standard range vehicles to deal with cell supply constraints. SR vehicles are using the iron phosphate cells now.

The answer to the million-mile battery and cell supply is 250-mile LFP packs which can already do 4k cycles and are perfect for the robotaxi model.

The extra stuff might slow down production if you want to push cell life longer or allow faster charging though. That will probably matter more for the nickel cells.

 

[MC3OZ]

Good point, I didn’t think of that.


At 2019 Autonomy Day, Elon was asked to “could you just clarify how you get the battery packs to last a million miles?”

Elon: “The million-mile life is basically just about getting the cycle life of the pack—you need basically to—you know, you need basically you know on the order of—let’s say you’ve got—basic math—if you’ve got like a 250 mile range pack you know you need 4,000 cycles. So, very achievable. We already do that with our stationary storage…”

For context, he had also just before giving this answer discussed prioritizing standard range vehicles to deal with cell supply constraints. SR vehicles are using the iron phosphate cells now.

The answer to the million-mile battery and cell supply is 250-mile LFP packs which can already do 4k cycles and are perfect for the robotaxi model.

The extra stuff might slow down production if you want to push cell life longer or allow faster charging though. That will probably matter more for the nickel cells.

That might have been Elon deflating some of the hype.

We know Jeff Dahn is still doing work on things like single crystal and Tesla is still patenting the results.

For some applications, 10,000 cycles might be better than 4,000 cycles, even with a price premium.

I was most interested in the fact that the 100 year battery was single crystal.


Why Jeff Dahn is chasing such long duration batteries is interesting. Some of the learnings can be applied more generally. I also think energy storage is an important market and energy storage cells might diverge from vechicle cells.

 

[-=buzz=-]

If charging speed is limited at the cell level by C rate what significant advantage do you imagine is provided by running at 800V?

Well, in theory you could reduce all the wiring thicknesses down to 1/4 area and reduce their weight, or keep them thicker and benefit from lower resistive losses.

 

[JRP3]

Well, in theory you could reduce all the wiring thicknesses down to 1/4 area and reduce their weight, or keep them thicker and benefit from lower resistive losses.

Which is probably what Elon is talking about when he says it might save $100 per vehicle, but I said significant advantage. Also the cell wiring sizes cannot be reduced.

 

[JRP3]

That might have been Elon deflating some of the hype.

The whole million mile battery concept is overblown. It's simply the convergence of energy density, cycle life, and cost. LTO has been capable of a million mile battery for years but the energy density and cost make it a poor choice. The current Model 3 with LFP is likely capable of it now as mentioned, (250+ mile pack), as is a 400 mile NMC pack.

 

[mongo]

Which is probably what Elon is talking about when he says it might save $100 per vehicle, but I said significant advantage. Also the cell wiring sizes cannot be reduced.

Well yeah, especially since 4680s don't have cell wires (per Berlin pic).
All the rest of the HV wiring could be downsized though. Same rational for 48V LV system (though that has other benefits).

 

[JRP3]

Well yeah, especially since 4680s don't have cell wires

They do have current carrying connections though, functionally the same thing.

 

[-=buzz=-]

Which is probably what Elon is talking about when he says it might save $100 per vehicle, but I said significant advantage. Also the cell wiring sizes cannot be reduced.

I published my post a bit too early.

The other benefit is that you could build the power electronics ( inverter & motor) in a smaller package for the same output. In case of the inverter I heard you could even save some circuits, which in the current shortage could allow for a higher production number.

But you have to pay for this with more insulation, unless the existing wires are overinsulated.
And the power electronics are higher priced as you're getting into the 'exotic' voltage ranges. While 400V is a common industrial voltage with many components already available as a commodity...

So it's the best thing for a top-of-the-range vehicle, but not necessarily available in the mainstream quantities.

 

[Buckminster]

interesting update on roof:

Notes (new information denoted in orange)
Drew seems genuinely excited about where battery industry is headed

"It seems likely that peak oil is near now, not because we're gonna run out of oil, which didn't end up being correct, but because there's just a better alternative. That's eventually going to be true of gas too, not because we're gonna run out of gas but because we're going to decide 'Hey, maybe we should use gas for really useful things only gas can be used for instead of heating people's homes...and businesses and industrial processes.' "

NEW - Drew believes no energy storage technology has been fully leveraged yet, not only lithium ion batteries but also gravitational, compressed air, and other battery types, in part due to regulatory structures

Hundreds of TWh worth of batteries needed to electrify everything and 20 TWh total annual production needed to make transition fast enough. (As stated in my energy manifesto last week, I think we'll end up making more than this in the long run.)

Battery industry raw materials are the constraint on rate of scaling, not factories. Battery industry was previously buying the leftovers from the rest of the economy. Now we have an opportunity to design the mining and refining processes to be customized for batteries.

NEW - Academia and startups can help by researching:

1. Better mining/refining/recycling techniques
2. Sodium-ion batteries or other kinds of designs to hopefully be compelling for stationary storage without requiring lithium
3. Moonshot battery designs that eliminate all use of metal in cathodes (iron, nickel, manganese) and instead uses graphite or other materials
4. Better physics simulation tools to accelerate the rate of iteration for batteries, electrolyzes, or anode/cathode materials

Regulatory barriers, especially with mining, are putting undue burden on solving the existential problem of electrification.

NEW - A decade ago, Tesla was too pessimistic about Model S battery cell electrochemical degradation, but too optimistic about the other stuff like pack moisture sealing, battery management electronics, mechanical shock and vibration, and thermal cycling. Notably, these things "don't show up until you've been in the field for ten years" which does not bode well for lagacy auto.

NEWish - For Solar Roof, they had an "ivory tower" approach where they "spent a lot of time on the shingle and not enough time on the stuff around the shingle" like perimeter flashings and obstructions such as vents. The solar shingle is "awesome", "super easy to build" and "not expensive at all". Confident that these issues will be solved with time and effort.

Confirmation that battery swap loses to supercharging because it is more complicated, slower to install, more expensive on CapEx and OpEx, and takes up more room on total infrastructure.

Comments
Drew once again demonstrates excellent skills with technical communication and public speaking. As usual, he was friendly, polite and had good body language and was actively listening to questions and giving real thoughtful answers. He can get into technical details by memory in a regular conversation without prepared remarks. Clearly Drew is not a fakeypants politician like, for example, the chief engineer at Ford (refer to Munro Live Mach-E interview).

Drew's statements about electrification of everything, hydrogen electrolysis and using gas (aka methane and propane) for irreplaceable applications other than heating adds to the evidence that SpaceX or Tesla will be working on electricity hydrogen/methane factories as I predicted in these posts: Post 1 (chemical section) & Post 2.

Solar Roof will work great and make profits eventually, because solving a long list of tricky geometric and installation details is not a significant technical risk. It's mostly just a lot of work and development of software tools to speed up tile selection and layout. It's good to hear that the hard part is not the problem.


Drew Baglino.

Elon is doing a lot and is the public face, but the team is stacked DEEP with great people who are in some ways better suited for the Technoking role. Tesla would do very well with Drew Baglino running the show. With Elon's time split between Tesla, SpaceX et al, I'd have to think Drew is already leading much of the time in practice. Drew is one of the earliest Tesla employees, having joined in 2006 and worked alongside Elon and JB the whole time.

Those at the top of the food chain can't stop this trend, which has been going on longer that Elon has been alive, and if he dies I guaranteed that the martyrdom effect will kick in and galvanize the renewable energy community for years.

View attachment 810584

 

[MC3OZ]

The whole million mile battery concept is overblown. It's simply the convergence of energy density, cycle life, and cost. LTO has been capable of a million mile battery for years but the energy density and cost make it a poor choice. The current Model 3 with LFP is likely capable of it now as mentioned, (250+ mile pack), as is a 400 mile NMC pack.

From memory the term was coined by the media, most of the hype started outside Tesla.

Jeff Dahn is researching how long batteries can last for in lab conditions. That has odvious advantages, but doesn't necessarily mean a new product is coming.

 

[mongo]

The other benefit is that you could build the power electronics ( inverter & motor) in a smaller package for the same output. In case of the inverter I heard you could even save some circuits, which in the current shortage could allow for a higher production number.

That's debatable. High voltage lower current gives the same switching losses and higher voltage switching devices will tend to have higher resistances (in the FET space), so efficiency gains are minimal to negative. Unless the motor gets rewound, ripple will be worse or switching frequency needs to increase which increases switching losses. So cooling doesn't get better which may constrain package shrink (not that it's large now).

They had 600 or 650V rated switching parts, like you stated, higher then requires everything to bump up which means bigger caps (physically) and more space between HV traces.

Connectors and wire places that would most benefit from twice the voltage, half the amperage. Which is what makes a 48V LV system appealing (along with easy interface to loads like LEDs).

 

[Doggydogworld]

Tesla has said outright the 4680 currently shipping is not using any of the other advancements they talked about on battery day. They are focusing on scaling first then will introduce the rest of the changes over the next year.

Who said this, and when? Links? And which changes will they introduce within a year? The SilLion anode? Cobalt-free NM cathodes? 54% range increase? 100 GWh production this year? $25k car next year? Lithium springing from the Nevada soil after a few shakes of the salt shaker?

If charging speed is limited at the cell level by C rate what significant advantage do you imagine is provided by running at 800V?

How did you twist my statement of "minor 800V benefits" into a claim of significant advantage?

 

[Ogre]

Who said this, and when? Links? And which changes will they introduce within a year? The SilLion anode? Cobalt-free NM cathodes? 54% range increase? 100 GWh production this year? $25k car next year? Lithium springing from the Nevada soil after a few shakes of the salt shaker?


How did you twist my statement of "minor 800V benefits" into a claim of significant advantage?

Traveling so no links. Listen to the latest earnings call or search for a transcript. They are pretty clear on the state of the 4680.

 

[Doggydogworld]

Traveling so no links. Listen to the latest earnings call or search for a transcript. They are pretty clear on the state of the 4680.

Yeah, the earnings call was the same kind of vague stuff we've been hearing for two years. I thought maybe you'd heard something more specific since then. On a different note, looking back through the transcript I saw where Drew said Tesla is not cell-constrained this year. I missed that the first time. That argues against the idea that the 279 mile pack is only partially populated to stretch their cell supply. More and more I'm thinking it's 828 cells, each roughly 83 Wh.

 

[JRP3]

How did you twist my statement of "minor 800V benefits" into a claim of significant advantage?

You:

Of course if Tesla were 800V and the others 400V it'd prove they were "5 years ahead", lol.

 

[mongo]

You:

My third party viewpoint was that the statement was a joke about the pro-anything-Tesla-does spin machine trope. If Tesla did it, then it would be they were 5 years ahead, if Tesla didn't do it, it's dumb to do it. (Thus the double quotes and lol)
Not that 800V is actually being 5 years ahead.

 

[Ogre]

Yeah, the earnings call was the same kind of vague stuff we've been hearing for two years. I thought maybe you'd heard something more specific since then. On a different note, looking back through the transcript I saw where Drew said Tesla is not cell-constrained this year. I missed that the first time. That argues against the idea that the 279 mile pack is only partially populated to stretch their cell supply. More and more I'm thinking it's 828 cells, each roughly 83 Wh.

This quote from Drew is what I was referring to:

Our priority was on simplicity and scale during our initial 4680 and structural battery ramps and as we attain our manufacturing goals, we will layer in new material technologies we are developing and higher range structural pack revisions.

This is likely about as detailed as you will get from public statements. But they were clear the current chemistry is not incorporating any of their improvements aside from the DBE and other changes specific to migrating to their new process.

Musk followed and made it more clear to expect improvements to chemistry in the next 12 months:

I think maybe in a nutshell, I think it probably is fair to say that 4680 and structural pack will be competitive with the best alternatives later this year and we think will exceed the best alternatives next year.

Again, they aren't likely to drop specifics in public statements.