Investor Engineering Discussions

[MC3OZ]

Seems that fast charging of sodium-ion batteries needs more R&D.

Just to clarify;-

The Na-based layered oxides (NaxMO2, 0 ≤ x ≤ 1, M = Mn, Ni, Co, Cu, etc.) are composed of transition metal oxide layers between which the Na+ cations are intercalated. This class of materials receives much interest due to its resemblance to the Li-based layered oxides, e.g., LiNi1-x-yMnxCoyO2 (NMC), which are widely used in lithium-ion batteries.

In 2018, Faradion reported a specific energy density of 150 Wh kg−1 for 32 Ah Na-ion pouch cell prototypes with a layered oxide positive electrode. 6,7 For comparison, LiFePO4/graphite cells typically achieve 160 Wh kg−1 and state-of-the-art NMC/graphite cells can offer ∼260 Wh kg−1 at the cell level. 8

Hard carbon (HC) remains by far the most adopted negative electrode in Na-ion batteries. The HC structure features graphitic domains and nanopores. The suitability of HC as negative electrode in Na-ion batteries is rooted in its cheap sources (organic precursors) and its high insertion capacity for Na+-ions exceeding 300 mAh g

I think the CATL Sodium battery is a bit different, what Jeff is researching seems more similar to typical lithium batteries.

 

[MC3OZ]

Lithium Metal Anode Licerion Batteries for Electric Vehicles: NiMH, Li-Ion, and What Comes Next​

Info on what some of the battery competition is doing.

 

[mongo]

Not much discussion about new 48 volt architecture for the Cybertruck/Nextgen car and the impact on margin. Besides reduced cost for less copper used and reduced weight are there other cost advantages? Any estimates on savings? Is POE going to be used?

Patents called out preformed flat harness assemblies which cut part cost and weight further along with easier installation.
Possibly dual high current 48V busses for redundancy and separate data lines. Another patent showed a ring topology with fault tolerance by rerouting traffic around a single break.
I'm interested to see where they placed the line between multifunction control nodes and self-sufficient end nodes.

 

[mongo]

Far fewer ECUs is one of main advantages. This is a very substantial cost savings. If they utilize a ring topology as mentioned by some here, additional redundancy allows for other ECUs to take over functions in the event of a failure of one.

The ring topology allows continued functionality if the normally monodirectional data lines are damaged, or a node fails. If a node fails though, all its functions are lost also.
As I recall, only safety crucial items (steering, brakes) have redundant controllers.

 

[Oil4AsphaultOnly]

The ring topology allows continued functionality if the normally monodirectional data lines are damaged, or a node fails. If a node fails though, all its functions are lost also.
As I recall, only safety crucial items (steering, brakes) have redundant controllers.

Man, this totally brings me back to the days of thin-nets! As long as Tesla is in full control of the ring, then this should be fine. But the reason 10base-T won out, was because of ease of troubleshooting packet collisions. Since CANBUS does NOT support ring topology, is Tesla planning to introduce their own network protocol?

 

[mongo]

Man, this totally brings me back to the days of thin-nets! As long as Tesla is in full control of the ring, then this should be fine. But the reason 10base-T won out, was because of ease of troubleshooting packet collisions. Since CANBUS does NOT support ring topology, is Tesla planning to introduce their own network protocol?

Thicknet vampire taps...
My hazy memory says it's custom with two pairs, one in each direction. Packet flow is clockwise on one, counter clockwise on the other. If a link breaks, the nodes on either side retransmit the packets down the other pair.

Unless it's one pair mono directional and switches to bidirectional on failure...

Gah, now I've gotta look it up...

Yeah, dual pairs
WIPO - Search International and National Patent Collections

 

[bkp_duke]

The ring topology allows continued functionality if the normally monodirectional data lines are damaged, or a node fails. If a node fails though, all its functions are lost also.
As I recall, only safety crucial items (steering, brakes) have redundant controllers.

Sorry, was a given assumption we were talking an Ethernet or Ethernet-like ring topology setup. Ethernet by it's very design has built-in redundancy with multiples wires within the same sheath that gives some protection against what you describe.

So, I was assuming with the 48V transition there would be a move to Ethernet communication more fully as well. 48V already works very well over ethernet (/me looks at all the PoE and 10G gear in his house).


EDIT - if I'm not being clear - I'm thinking of a loop of devices connected by Ethernet. We employ this in datacenter critical switch design and deployment, so that if some cage monkey pulls the wrong wire, a critical piece of switching gear doesn't go offline, and we get an alert that one of the 2 links to that device is down.

 

[mongo]

Sorry, was a given assumption we were talking an Ethernet or Ethernet-like ring topology setup. Ethernet by it's very design has built-in redundancy with multiples wires within the same sheath that gives some protection against what you describe.

So, I was assuming with the 48V transition there would be a move to Ethernet communication more fully as well. 48V already works very well over ethernet (/me looks at all the PoE and 10G gear in his house).


EDIT - if I'm not being clear - I'm thinking of a loop of devices connected by Ethernet. We employ this in datacenter critical switch design and deployment, so that if some cage monkey pulls the wrong wire, a critical piece of switching gear doesn't go offline, and we get an alert that one of the 2 links to that device is down.

Sure, and Automotive Ethernet is a thing, Tesla may be leveraging that.
I expect the power requirements exceed what makes sense for controlled impedance twisted pair.
Automotive Ethernet: The Future of In-Vehicle Networking

 

[mongo]

IMO the unboxed process process described on Investor Day is the best unboxed process.

I admit my speculation is partially informed and unlikely to be correct, I am only speculating because I don't understand why Tesla would deviate from a pure unboxed approach.

If it was possible, dropping a prebuilt roof onto separate front and rear castings effectively achieves a 1-step vehicle body construction, it is certainly what I would do, even if reinforcing the roof casting with high strength steel was difficult. Rather than diluting the the unboxed process, this is the perfect complement to the unboxed process.

Some hybrid approach with some elements of both approaches is also possible, but IMO the ability to replace the battery pack is essential, and a structural battery pack is ideal. That can be accommodated in your solution.

The unboxed has disconnected sides also. The rocker would need to be part of those unless either the pack is permanently attached.

The presentation may have had the highest access to all parts, but the trade offs might not be worth it.

 

[MC3OZ]

The unboxed has disconnected sides also. The rocker would need to be part of those unless either the pack is permanently attached.

The presentation may have had the highest access to all parts, but the trade offs might not be worth it.

Good idea to move the discussion to this thread.

I got a bit sloppy because it was the weekend... and because the original topic started in that thread.

Trade-off is the right term to use...

I think of it this way, anyone doing serious work on an ICE engine would remove the engine from the car as the first step, because after that, access is much easier from all sides.

The additional point is that a large casting and a very large casting machine will not be cheap to do. I'm not sure merely joining the front an rear castings has sufficient payback, especially when it results in some restricted access.

The construction method I suggested may add a full length rocker after the roof is joined to the front an rear castings, The roof and perhaps some temporary bracing or the battery pack holds the castings in place until this can be done.

Ultimately Tesla are smart enough to make any approach work, they have the full picture on the trade-offs.

I went down this rabbit hole in part because those trolleys at Austin looked like a good fit with the unboxed process described on investor day.

And because a full length floor casting didn't seem like a good fit with the investor day process and perhaps wasn't a good fit with the trolleys.

I'm still only guessing what the trolleys might have been used for.

 

[MC3OZ]

Tesla granted the patent for the structural battery pack. Seems like an easy one - it’s so novel and useful.

https://image-ppubs.uspto.gov/dirsearch-public/print/downloadPdf/11764432

This seems to be a small battery brick/module design, not necessarily a structural pack (though it calls out potting the cells in one embodiment). Main new thing seems to be calling out a PCB for one of the current collectors which could integrate other functions.
It has the same themes of welded tube can or one piece stamped sheet/cans we've seen before. Same with option for immersed cans for cooling.

Looks like something that could be useful for Optimus or Powerwall (if not prismatic LFP).

There are some changes to the ways cell are electrically connected which simplifies cell design and lowers costs.

I agree that the liquid coolant probably rules out this being a structural battery pack.

If Tesla are going to build Megapacks with 4680 cells then this architecture seems like a good fit, they may be able to have one cooling loop for the entire Megapack.

 

[mongo]

There are some changes to the ways cell are electrically connected which simplifies cell design and lowers costs.

I agree that the liquid coolant probably rules out this being a structural battery pack.

If Tesla are going to build Megapacks with 4680 cells then this architecture seems like a good fit, they may be able to have one cooling loop for the entire Megapack.

There are so many claims with different configurations, including a potted (which could be phase change) and liquid cooled.
One cooling loop would result in a large temperature delta across all the modules. Megapack and LFP prismatic with plate cooling might be the best option.

PCB as BMS and interconnect would be neat.

 

[MC3OZ]

There are so many claims with different configurations, including a potted (which could be phase change) and liquid cooled.
One cooling loop would result in a large temperature delta across all the modules. Megapack and LFP prismatic with plate cooling might be the best option.

PCB as BMS and interconnect would be neat.

Yes, it will ne interesting to see if they use this patent and what they use it for,

When a similar patent is filed multiple times, I typically assume some intent to use it eventually.

Tesla Semi might also use this, it will never have a structural pack..

 

[MC3OZ]

There are so many claims with different configurations, including a potted (which could be phase change) and liquid cooled.

Do a bit more reading the phase change is from liquid to gas as a way of preventing a local build up of heat during thermal runaway, as gas will move the heat away from the local area faster.

I am now wondering if a conventional water/anti-freeze mix could be one of the cooling options.

The patent also talk about interfacing to HVAC a heat pump and octovalve could be part of the cooling loop.

A higher flow rate could compensate for a single loop. Or perhaps there could be a branch / re-merge topology.

Might end up similar to a ICE engine cooling system?

 

[Buckminster]

Bionic silkworms with spider genes spin fibers 6x tougher than Kevlar
6x Kevlar strength

 

[BioSehnsucht]

https://twitter.com/x/status/1706768319825506584

https://twitter.com/x/status/1706774184708473084

I wonder if V3 to V4 upgrades will be easy, or if the placement of conduit within the footprint and/or length of wiring will cause a problem.
"5. Identify the charge post's anchor locations (which are identical to the V3 charge post locations):" (page 11)

I imagine that the large opening space for conduit at the base means the conduit would likely be fine, and at worst they'd need to replace the site wiring within the conduit, to do a V3 to V4 upgrade (at least as far as pedestals are concerned, not sure what would need upgrading at the supply end).

Interestingly, from reading the install guide, apparently what we typically think of as high voltage DC between pedestal and supply (and the car, and it's battery) is just "MVDC" (Medium Voltage DC, I'm guessing - page 18).

Tesla appears to be using some possibly bespoke, possibly off the shelf, wiring (V4 Comm Cable, page 20). I've never seen any with such a combination, but it's possible it was something they were able to source that already exists. Sure makes pulling it easier to pull one cable with all the various wires than pulling several cables in the same conduit, though.

It combines 24V power for the pedestal electronics, two-wire CAN bus for communication between pedestal and supply (and car, presumably), a separate "enable" wire (Guessing this is used to close a relay to power up the rest of things, rather than simply powering on with 24V immediately), a bare copper drain (ground, basically, for the comm cable itself, to improve shielding) wire, and what appears to be 2-pair ethernet wiring (should be good for 100Mbps, even with only 2 pair vs normal 4 pair).

Strangely, while I see everything else being called out to be hooked up, I don't see mention of the 2-pair wire being terminated anywhere, and the previous page shows regular Cat5 ethernet being pulled to the pedestal for the payment display (which is located further up and to the "left"). In theory they could have used the apparently unused 2-pair in the V4 Comm Cable to run the payment display, but perhaps that is so off-the-shelf that it isn't actually tied into the rest of the pedestal anywhere and thus is being powered from PoE and needs a full Cat5 hookup. The instructions also don't mention actually hooking up the Cat5 cable to the payment display, even though it calls out the length the cat5 needs to extend up into the pedestal area, and that it is for the payment display. Perhaps it's assumed the installer knows how to terminate a normal Cat5 cable properly and they can figure out how to plug it in without further instruction? Seems an odd oversight. Could be this is a US version of the manual and was partially edited down if they're not bothering with payment terminals here vs in EU?

The whole extra cat5 vs using existing 2-pair situation seems like a missed opportunity for install simplicity - pre-wire the display over to where the rest of the pedestal electronics are, with a built-in PoE injector powered from the pedestal electronics, and skip pulling a dedicated Cat5 wire by using the V4 Comm Cable's 2-wire section. If a future upgrade wants to use Ethernet in the pedestal electronics itself, just give the new electronics a cheap switch chip. Not sure what the added cost per pedestal would be to internalize PoE but you would probably save money overall versus the extra Cat5 pull and the "wasted" 2-wire portion of the V4 Comm Cable.

Even an 8-pedestal location is probably going to need around $35-70 of Cat5e cable at minimum (depending on whether they go cheap CCA or good solid copper), and the component costs for PoE injection from the pedestal's 24V supply (DC to DC buck converter to get from 24V to 44-57V, controller if it needs to support the fancier flavors of PoE+/++, etc) should only be a few bucks - I'd say $5 per pedestal would be reasonable, even if supporting 802.11bt (PoE++ 60W), though even 802.11at (PoE+ 30W) should be more than plenty, and likely even ye olde 802.11af (15W) would be good enough. At 8-pedestal sites it might be roughly break even (pedestal costs being slightly higher, but labor reduced), and at larger sites the labor and material savings (bigger sites quickly need much more Category 5 wiring) should be in favor of an integrated design.

Perhaps though the math worked in favor of keeping the payment display separate, simply because they expect to deploy so many V4 pedestals in the US where they might not bother with the payment display at all, versus in the EU? For all those US installs, if they really lack the payment display module, they won't have the Cat5 cable to pull or terminate anyways, and adding onboard PoE injection for something that isn't there would be a further waste of money.

 

[Knightshade]

I wonder if V3 to V4 upgrades will be easy, or if the placement of conduit within the footprint and/or length of wiring will cause a problem.

Do we expect they plan to do that? Didn't they mostly just keep installing new V3s when those came out rather than upgrade V2s? Of the 11363 locations listed at supercharge.info only about 2/3rd of em are V3s despite launching back in 2019 so seems like they mostly left the older ones as is to increase total # of locations.... with lots more brands gaining NACS access in the future I expect that'd remain their focus rather than in-place upgrades?

 

[mongo]

Interestingly, from reading the install guide, apparently what we typically think of as high voltage DC between pedestal and supply (and the car, and it's battery) is just "MVDC" (Medium Voltage DC, I'm guessing - page 18).

Yeah, we tend to talk about the vehicle which has a low voltage and high(er) voltage system. Generac calls <600V medium, whereas transmission lines of 1kV to 35kV are medium (sometimes).

Tesla appears to be using some possibly bespoke, possibly off the shelf, wiring (V4 Comm Cable, page 20). I've never seen any with such a combination, but it's possible it was something they were able to source that already exists. Sure makes pulling it easier to pull one cable with all the various wires than pulling several cables in the same conduit, though.

There are places that will make custom cables from standard off the shelf ones. The drawing seems to indicate that sort of setup.

It combines 24V power for the pedestal electronics, two-wire CAN bus for communication between pedestal and supply (and car, presumably), a separate "enable" wire (Guessing this is used to close a relay to power up the rest of things, rather than simply powering on with 24V immediately), a bare copper drain (ground, basically, for the comm cable itself, to improve shielding) wire, and what appears to be 2-pair ethernet wiring (should be good for 100Mbps, even with only 2 pair vs normal 4 pair).

Enable may be a signal to the cabinet for rapid fail safe. Previous pedestals don't have any switching (v4 has cooling, but that seems like it would be locally controlled).

Strangely, while I see everything else being called out to be hooked up, I don't see mention of the 2-pair wire being terminated anywhere, and the previous page shows regular Cat5 ethernet being pulled to the pedestal for the payment display (which is located further up and to the "left"). In theory they could have used the apparently unused 2-pair in the V4 Comm Cable to run the payment display, but perhaps that is so off-the-shelf that it isn't actually tied into the rest of the pedestal anywhere and thus is being powered from PoE and needs a full Cat5 hookup. The instructions also don't mention actually hooking up the Cat5 cable to the payment display, even though it calls out the length the cat5 needs to extend up into the pedestal area, and that it is for the payment display. Perhaps it's assumed the installer knows how to terminate a normal Cat5 cable properly and they can figure out how to plug it in without further instruction? Seems an odd oversight. Could be this is a US version of the manual and was partially edited down if they're not bothering with payment terminals here vs in EU?

This installation manual is only the first step of commissioning. Note that the front logo light doesn't get plugged back in since the panel is coming off again.
Payment system connection is a follow up step.

The whole extra cat5 vs using existing 2-pair situation seems like a missed opportunity for install simplicity - pre-wire the display over to where the rest of the pedestal electronics are, with a built-in PoE injector powered from the pedestal electronics, and skip pulling a dedicated Cat5 wire by using the V4 Comm Cable's 2-wire section. If a future upgrade wants to use Ethernet in the pedestal electronics itself, just give the new electronics a cheap switch chip. Not sure what the added cost per pedestal would be to internalize PoE but you would probably save money overall versus the extra Cat5 pull and the "wasted" 2-wire portion of the V4 Comm Cable.

Best part is no part?
Extra cable pairs vs PoE injector, CAN bridge.
Then you have the issue that the payment unit is able to bring down the charging functions.

Like you call out, wire isn't free.
The prefab 4 post + cabinet units are < 40 ft wide, so total cable is ~100 feet/4= 25 foot per pedestal. They call out 18 AWG which is a lot (4x) more copper than normal 24AWG.

 

[willow_hiller]

I think this is about as close to closure on LK99 that we'll get for the moment.

Just in case anyone wanted a more definitive follow up on this story, the co-authors are seeking to retract their original paper: WSJ News Exclusive | Co-Authors Seek to Retract Paper Claiming Superconductor Breakthrough

 

[mongo]

Yeah, we tend to talk about the vehicle which has a low voltage and high(er) voltage system. Generac calls <600V medium, whereas transmission lines of 1kV to 35kV are medium (sometimes).



There are places that will make custom cables from standard off the shelf ones. The drawing seems to indicate that sort of setup.


Enable may be a signal to the cabinet for rapid fail safe. Previous pedestals don't have any switching (v4 has cooling, but that seems like it would be locally controlled).
View attachment 977544



This installation manual is only the first step of commissioning. Note that the front logo light doesn't get plugged back in since the panel is coming off again.
Payment system connection is a follow up step.



Best part is no part?
Extra cable pairs vs PoE injector, CAN bridge.
Then you have the issue that the payment unit is able to bring down the charging functions.

Like you call out, wire isn't free.
The prefab 4 post + cabinet units are < 40 ft wide, so total cable is ~100 feet/4= 25 foot per pedestal. They call out 18 AWG which is a lot (4x) more copper than normal 24AWG.

Just reread. The dual pair Ethernet looks like optional population option in the multicable.
Payment system runs off a separate cable.