Investor Engineering Discussions

[Buckminster]

New Goodwood hill climb record from an EV. Cool downforce fan tech:

https://twitter.com/x/status/1541090417319575561

 

[nwdiver]

I don't think this is Tesla but they probably have similar equipment. Crazy how quickly it escalated.

 

[Oil4AsphaultOnly]

I don't think this is Tesla but they probably have similar equipment. Crazy how quickly it escalated.

Ummm ... "that escalated quickly"!

Dude barely had enough piece of mind to snatch whatever was on that computer console and RUN before the whole ceiling came crashing down!

 

[Doggydogworld]

Ummm ... "that escalated quickly"!

Dude barely had enough piece of mind to snatch whatever was on that computer console and RUN before the whole ceiling came crashing down!

It was his phone, I think. It was a bad enough fire when the oil on the floor and equipment was burning, but when the flames reached the leak itself it turned into a huge flamethrower. Imagine an IDRA press leak, aka Giga-flamethrower.

 

[mongo]

It was his phone, I think. It was a bad enough fire when the oil on the floor and equipment was burning, but when the flames reached the leak itself it turned into a huge flamethrower. Imagine an IDRA press leak, aka Giga-flamethrower.

Aerosolized flammables are never a good time. Unless there are shutoff valves upstream or dump valves, any accumulator volume can escape uncontrolled even if pumps are shutdown.
Giga has the advantage of being in a concrete or firebreak wall room with high ceilings and minimal other combustibles.
Fremont did have one leak/ fire in March 2021 and needed some repair.

 

[mongo]

Not only did I work on labeling at Tesla (I set up the first team), but also during my time at Google Maps and yes, they need tight coupling. Quick pivots are key in labeling as is mass updating when code changes are made to look at new features in new ways.

Quick implementation pivots sure, but does that require colocation of all 1,500 labelers?

If I remember correctly Andrej has mentioned that it’s critical to have the labeling team colocated with the AI engineers in the SF Bay Area so that they can all work together directly.

 

[mongo]

So this is odd. The SR+ uses LFP cells while the LR and Performance use 2170s. They would have to be stockpiling LFP cells or perhaps they had a shortage.

At some point this is going to have to swing the other way. I wonder if Tesla is favoring 2170 base cars for a bit to boost margins and will have a surge of LFP shipments in early Q3 when they are less worried about hitting numbers.

Wasn't the plan to move stationary storage to LFP? And Isn't Tesla building a megapack factory in CA right now? Not saying those things are connected, but just off top of my head would be a good use for "extra" LFPs they're not putting in SRs right now in favor of building LRs.

I thought SR+ packs used fully up modules from CATL. I don't think Megapack has the same form factor.
Maybe CATL can produce Megapack version on same equipment ?

 

[Knightshade]

Honestly no idea- that was just the immediate thought I had about "what could Tesla do with a bunch of extra LFP piling up.

I suppose the alternative is if they've decided to start shifting a bunch of 2170 supply to Austin as they appear in the process of doing to ramp LR/2170 there with 4680 going slow it'd open more SR build slots at Freemont and were stockpiling for it.... but not until entire LR backlog was caught up on....

 

[Ogre]

I thought SR+ packs used fully up modules from CATL. I don't think Megapack has the same form factor.
Maybe CATL can produce Megapack version on same equipment ?

Honestly no idea- that was just the immediate thought I had about "what could Tesla do with a bunch of extra LFP piling up.

I suppose the alternative is if they've decided to start shifting a bunch of 2170 supply to Austin as they appear in the process of doing to ramp LR/2170 there with 4680 going slow it'd open more SR build slots at Freemont and were stockpiling for it.... but not until entire LR backlog was caught up on....

I don't know, but if they had a surplus of 2170s and an alternative use for those LFP cells it would connect a lot dots here. Increasing revenues on both sides of the table during a down quarter would be a huge win.

Not sure if this is just hypothetical or not, but the idea they are able to shift around their capacity like this is very bullish. Even if it's simply the fact that their LFP shipment was late and they shifted production to LR/ Performance, that's a big win compared to Ford/ GM who would have to grind their assembly lines to a halt if they missed a shipment.

 

[JRP3]

Tesla never showed bottom cooling.
The cell bottom is not uniform and is also the safety vent.

Also in the new 4680 teardown video from The Limiting Factor the researcher says the bottom is electrically isolated from the electrode and the cell can which means it's also thermally isolated, i.e. bottom cooling isn't even possible.

 

[mongo]

I don't know, but if they had a surplus of 2170s and an alternative use for those LFP cells it would connect a lot dots here. Increasing revenues on both sides of the table during a down quarter would be a huge win.

Not sure if this is just hypothetical or not, but the idea they are able to shift around their capacity like this is very bullish. Even if it's simply the fact that their LFP shipment was late and they shifted production to LR/ Performance, that's a big win compared to Ford/ GM who would have to grind their assembly lines to a halt if they missed a shipment.

Yah, it may be the cells in the modules can be used for Megapack if taken upstream from module assembly.

 

[mongo]

I'd still expect to see some in the large central void.

They only need to wet the separator and electrodes, similar to an AGM lead acid battery

Also in the new 4680 teardown video from The Limiting Factor the researcher says the bottom is electrically isolated from the electrode and the cell can which means it's also thermally isolated, i.e. bottom cooling isn't even possible.

Indeed. I was going to post about that too
So there is no deliberate connection from the copper current collector to the can bottom, and heat travels from that piece to the side walls then to cooling ribbon.

Was it? I thought DBE was unrelated to electrolyte. DBE is the process step that takes the cathode material (nickel, manganese, cobalt, lithium) and squishes/mixes/chemically bonds them together to make something that can be laid on the aluminum collector film. The normal process is to make a liquid slurry of these elements, smear them on the aluminum and then bake it in ovens to drive off the solvent. You then construct the jelly roll and THEN you squeeze in the liquid electrolyte once everything is in the can. DBE doesn’t change that part of the process (injecting liquid electrolyte into the can). Now, I’m not a battery expert and I could be completely wrong here, so let me know.

Yeah, DBE is separate, but they didn't say it wasn't . DBE was an aside in their statement, not linked.

"I'm pretty sure that part of the process for the new cells, in addition to DBE, was to reduce total liquid electrolytes."

 

[Oil4AsphaultOnly]

Also in the new 4680 teardown video from The Limiting Factor the researcher says the bottom is electrically isolated from the electrode and the cell can which means it's also thermally isolated, i.e. bottom cooling isn't even possible.

There's thermally conductive epoxy "potting" compound that's used to protect electronic circuit boards without shorting anything together. I imagine that the existance of this compound pretty much eliminates that line of reasoning. So although there is indeed no bottom cooling, but it IS definitely possible.

 

[JRP3]

They only need to wet the separator and electrodes, similar to an AGM lead acid battery

Looks as if electrolyte gets injected into the central void left by the winding mandrel:

 

[mongo]

There's thermally conductive epoxy "potting" compound that's used to protect electronic circuit boards without shorting anything together. I imagine that the existance of this compound pretty much eliminates that line of reasoning. So although there is indeed no bottom cooling, but it IS definitely possible.

Yeah, it 'could' be thermally conductive and electrically insulative, but the material used doesn't look optimized for that.

 

[mongo]

Looks as if electrolyte gets injected into the central void left by the winding mandrel:

Yeah, and then on the 4680 the copper plug is inserted. That video looks like a lot of extra electrolyte was used.

 

[JRP3]

There's thermally conductive epoxy "potting" compound that's used to protect electronic circuit boards without shorting anything together. I imagine that the existance of this compound pretty much eliminates that line of reasoning. So although there is indeed no bottom cooling, but it IS definitely possible.

I believe those are meant for flat surfaces with minimal bondline thickness which the tabless structure would not provide. So I would say it's so sub optimal that it is not functionally possible.

 

[JRP3]

Yeah, and then on the 4680 the copper plug is inserted. That video looks like a lot of extra electrolyte was used.

I'm not understanding how you could use the central void to fill the cell but not have electrolyte in it.

 

[MP3Mike]

I'm not understanding how you could use the central void to fill the cell but not have electrolyte in it.

Because all of the liquid would be absorbed/wicked into the cathode/anode material. (They are injecting a precise amount, which is what one of the companies Tesla acquired was a specialist in.)

 

[mongo]

I'm not understanding how you could use the central void to fill the cell but not have electrolyte in it.

If the optimal amount of electrolyte is used, it will all get wicked into the layers leaying none free.
On the fancy end of things, they could pull a vacuum, inject electrolyte, wait, then pull out the excess.