4680 charging issues???

[HaulingAss]

In essence, the larger cell volume is offset by the reduced overall heat generated and better dissipation from the tabless design.

This is what I argued repeatedly in 2020 and 2021. Jordan Giesige's video (above) didn't come out until 10 months ago.

Sponsored

 

[HaulingAss]

As others have posted, the poor charging rate on the 4680 MY's is just the consequence of a conservative charging algorithm so they don't produce premature failures on 4680 test MY's.

This could either be intentional as a hard rate limit, or because the cells are hitting the thermal constraints which in turn derates it. It would be interesting to see a plotted charge graph, preferably with a battery temp value.

The videos of the standard range Model Y 4680 charge speed testing presented earlier in this thread probably have graphs available if you go to his YouTube channel, but I watched the video (which included speeded up data feeds of battery temps during charging) so I don't need to see the graphs in order to say this:

The first gen 4680 cells in the Model Y appear to be charging slowly even when the BMS does not feel the need to actively cool the battery. To me, this implies that there may be localized heat buildup that is not carried away from the inside of the cell quickly enough and would have negative consequences for battery longevity if it were ignored. I suspect the next generation of battery pack, and/or the cells within it, will address this problem in some manner. It could be a change in chemistry or a change in construction of the separator.

Remember, a cell does not have one temperature during charging, it has temperature gradients within each cell.

I would be very surprised if the charging speeds of newer generations of 4680 cells do not steadily increase with each new generation.

 

[Dusty]

The chart was illustrating a relative relationship between charging time and diameter (with and without the tabless design feature). It's not an absolute graph with numbers, even the charge time of 2170 cells is not fixed, it has steadily improved as the format has been updated with chemistry and construction tweaks. The 4680 cell will improve too.

The chart is useful without numbers to illustrate Tesla's expectation for the tabless design in a 4680 format when the format is mature...

.... The chart makes it clear the tabless 4680 was expected to charge slightly slower than a single tab 2170... I was shouted down by people who had not used this very chart to inform themselves better.

I see that chart as more "dubious" than "useful". My gripe is about the chart being meaningful. RVAC and JBee are dead on about the chart. It doesn't say anything because no one can say if the Y-axis peak value is 10 minutes or 3 hours. That could be in 'cricket chirps' for all we know. There's no way to tell the actual degree of any of the infections, because you can't interpolate a single piece of data from it. Like I said, at best it says "Kinda Sorta Like This".

But that aside, Tesla is scraping every bit of real world charging data returned from the 4680 Y's, and they need to get the right chemistry for the 4680 format. Until then, Tesla's going to be super conservative with their existing 4680 charge rates. On top of that, I wouldn't expect the current batch of 4680 Ys to get much better because they may be hard-capped by the battery's construction and chemistry. The last thing they want is a "battery issue" in the news.

I totally agree with your sentiment regarding gen2 and gen3 4680s reaping the benefits of the format. I think the 4680 is a solid 5 years from maturity. Then we'll start to see the true potential unleashed. I said a while back I only plan on keeping my CT for 5 years max because it's 1st generation. The CT they build 4-5 years from now will be the ONE. The competition still working on their formats are way behind.

 

[GhostAndSkater]

I'm the author of some of the graphs John used in his video

I recommend reading these two papers, the amount of ohmic losses heat generated inside the cells due to the tabless electrode is greatly reduced to almost nothing

Also as shown there, the most likely explanation with all the data we have is that Tesla is being extremely conservative with the 4680 charging curve and I tried to do some speculation/calculations oh how much it could improve

https://iopscience.iop.org/article/10.1149/1945-7111/abd44f
https://iopscience.iop.org/article/10.1149/2754-2734/ac97e0

 

[Crissa]

Geez guys, I take a day to sleep and play D&D and you're arguing about an almost three year old chart. I think my DM might let me do a Fey Blood Charisma build Druid Multiclass.

• The "tab-less" design enables the 4680 to generate almost as little heat as the single tab 2170.
• The chart is labeled, but unhelpfully.
• The 4680 is worse than the 2170 - but only slightly.
• The chart does show that the 4680 is a tiny amount worse but only slightly. The number of pixels is easily visible.
• The 4680 Model Y also has about 10-20% fewer cells than its theoretical maximum, which is also not helping its charge curve.
• Yes, Tesla is being conservative with their charge curves of the 4680 Model Y.
• We have no confirmation of the charge curve or the layout of the "Cyber-cell" 4680 vehicles.

-Crissa

 

[SSonnentag]

I expect the 4680 cells will allow faster charging due to less heat being generated during the charging process. I know my S ramps down really quickly once the cooling fans start to scream. The lower internal resistance of the 4680 cells could help.

 

[RVAC]

The graph is not really representative of much without charge time values, and shouldn't be considered evidence of the 4680 being slower at charging.

There are two factors here, one is that the max charge rate is limited by cell temperature, because making the cells hot causes irreparable degradation, and two; cell temperature whilst charging is created through a combination of conductor and chemical heating that is commonly expressed as battery resistance.

The graph is representative, just not quantifiable. Whether the difference shown is meaningful is a different matter. I only brought this up because of the often repeated claim that the tabless 4680 charges faster, relative to a 2170.

Thank you for proving my point! Charge time for a tabless cell increases negligibly as diameter increases from 21mm to 46mm.

You're missing the point. You can't say it's negligible without any scale on the y-axis. What we can say for sure is that a tabless 4680 does not charge faster than a 2170, which was your claim and what prompted my reply in the first place.

Note the dashed red line, representing a tabbed cell, increases significantly over that same span.

Don't see the point of noting that when I've stated as much, more than once.

In essence, the larger cell volume is offset by the reduced overall heat generated and better dissipation from the tabless design.

See my first reply to you:

It's a bit of a misconception, the tabless design does reduce internal resistance but it is negated by the increase in cell diameter. This graph from battery day actually shows a slight supercharge time increase, although it's hard to quantify given the y axis is not labeled.

 

[JBee]

I'm the author of some of the graphs John used in his video

I recommend reading these two papers, the amount of ohmic losses heat generated inside the cells due to the tabless electrode is greatly reduced to almost nothing

Also as shown there, the most likely explanation with all the data we have is that Tesla is being extremely conservative with the 4680 charging curve and I tried to do some speculation/calculations oh how much it could improve

https://iopscience.iop.org/article/10.1149/1945-7111/abd44f
https://iopscience.iop.org/article/10.1149/2754-2734/ac97e0

Brilliant information. Thank you very much for the work and sharing this.

I wish this forum had more such great posts to digest.

To many armchair experts here, which might be fine for consumer level banter, but more often than not has nothing to do with actual engineering.

PS any idea why the 4680 to date are still using side of cell cooling instead of top across the tabs, which would seem better given the shorter thermal path? Is it only because of the surface area to extract heat, and/or because it's easier to manufacture? Or is it because it just doesn't need any more cooling capacity than that anyway because there is less ohmic heating?

 

[GhostAndSkater]

Brilliant information. Thank you very much for the work and sharing this.

I wish this forum had more such great posts to digest.

To many armchair experts here, which might be fine for consumer level banter, but more often than not has nothing to do with actual engineering.

PS any idea why the 4680 to date are still using side of cell cooling instead of top across the tabs, which would seem better given the shorter thermal path? Is it only because of the surface area to extract heat, and/or because it's easier to manufacture? Or is it because it just doesn't need any more cooling capacity than that anyway because there is less ohmic heating?

Thanks man

Yeah, because it's more effective, the second paper goes into detail on the why, and also recommend this video that is a summary of the paper

But the TLDR is, while cooling the bottom seems more effective because the thermal path is shorter, the performance is worse due to the smaller surface area

You have to think about not only about thermal transfer, but also electrical insulation. I have no data to back this up, but I imagine if you could have a finned stack attached to the bottom of the cell, or being the bottom of the cell with coolant flowing through it, bottom cooling would be the most effective one

But when you consider that whatever you are using to cool the cell has to be electrically insulate, and that usually goes against thermal conductivity, it's better to go to the longer thermal path but allowing more surface area to make up for the worse conductivity between the cell and the cooling ribbons

From worse to best according to the paper it goes:

Bottom cooling (B) < side cooling (M) < top and bottom cooling (T)

Top and bottom is way too complex for a high volume manufacturing, take a look at the image bellow, I highlighted the most interesting part, these are shortest charge times for the 4680 with the tabless electrode on the last row, from page 8 of the second paper, keep in mind as the author says, these are theoretical times that likely won't be achieve on real life

Cooling from the side 18 minutes, from the top and bottom 17 minutes and from the bottom 20 minutes, it's clear that there is a diminishing return point

In the end, all we can do now is wait for Cybertruck to release and have someone test it. I seriously doubt the charge times will be anywhere close to the 4680 Model Y, all the data we have so far points it should charge way faster, so I foresee it being at least on par with current best charging Tesla vehicles at 10-80% under half an hour, maybe even less on whatever V4 Supercharger ends up being

While usually Tesla releases new packs with a nerfed curve and with time improves it, on the Model S/X refresh² it launched with it's awesome charge curve from the start, so it's likely Cybertruck goes this way, either will be mine and everyone else disappointment

The next thread I will do is Cybertruck charge times using all the data from the previous posts for a various charger power, from Supercharger V2, to V3, to 350 kW V4 and 500 kW V4

 

[Ak85]

I came across this thread and would like to revive it as there is some great insight and information here. Now that we have the confirmed poor Cybertruck charge curves, does it provide any more insight into what's holding back the charge rate? Theoretically, the 4680 tabless should be able to have similar charge rates to the 18650 and 2170, but it doesn't as of now (not even close). Tesla and panasonic would not pursue the 4680 tabless if it was not an inherently good design. It must be some practical implementation limiting the cells.

From this thread and other sources these are the hypotheses that are floating around in my head:

1. the dry application is proving difficult (perhaps inconsistent/not-precise application) which is resulting in non uniform cell film whereby fast charging creates hot spots or differentiating SoC within the cell which is forcing Tesla to reduce the performance to avoid premature degradation. This would mean in order to get better charging performance, the manufacturing needs to be dialed in (if possible) or the technology needs to be improved (if possible) and/or Tesla would need to return to the more proven wet film application.

2. Maybe Tesla is just not as good at manufacturing batteries yet as LG or Panasonic. So there is no inherent innovation that needs to be made just that the manufacturing processes need to be dialed in which will take time.

3. Tesla is using the 4680 in vehicles as a test fleet and is learning the limits. I find this highly unlikely. I'm sure they have tested 4680s immensely. Sure a marginal increase in performance over time OTA as data comes in to get closer to the limit is imaginable. But a stepwise increase because they do not know yet where that limit is seems unlikely. What were the improvements in charge rates in percentage terms of major OTA upgrades in the past?

4. The 4680s could perform much better if cooled at the continual tabs side. (See the papers) However, Tesla is adapting its previous cooling methods to cool the cells along the side of the cells (like in its other models), perhaps for cost savings or ease of manufacturing, and they are OK with this tradeoff.

Does anyone have any more information to try to figure out why the 4680s are so bad right now?

To be clear the 4680s were never meant to outperform the 2170s, the primary goal was to match the performance but at a much lower production cost.

 

[Crissa]

2. Maybe Tesla is just not as good at manufacturing batteries yet as LG or Panasonic. So there is no inherent innovation that needs to be made just that the manufacturing processes need to be dialed in which will take time.

Well, we can't say that, since neither of them have a 4680 in production.

Does anyone have any more information to try to figure out why the 4680s are so bad right now?

To be clear the 4680s were never meant to outperform the 2170s, the primary goal was to match the performance but at a much lower production cost.

Yes, they're meant to be cheaper per kWh to produce, not otherwise better than the 2170s.

But that means it's early times, and we should be comparing the 4680s to first gen 2170s, not current gen. It takes time to learn the charge curves and not blow them up.

The latent thermal mass of a bigger cell just sheds heat more poorly, so it's going to be a tough learning curve.

Add to that, we're not testing it at its native voltage which means the charge circuitry will be thermally limited.

-Crissa

 

[Eddyrelik]

Honestly, I would think Tesla is just being conservative with their current CT charge rates since its a whole new design/battery pack/model. They must be setting the software to be conservative initially for safety and battery health....and as they get more real world use data, they can "unleash" the speed with updates as they confirm real-world cycling, thermals, etc.

 

[Ak85]

The latent thermal mass of a bigger cell just sheds heat more poorly, so it's going to be a tough learning curve.

I originally thought this as well. However as you can see with the ribbon cooling, only a small portion of the can is in contact with the cell. If latent heat dissipation was the limiting factor, I would imagine tesla would cool more than what looks like 1/6th the outside of the can. (perhaps implementing double strip cooling so that both sides were cooled)

The tabless design also generates a lot less heat as there is less ohmic losses. The research papers both go through this.

Through reading the papers heat is not really difficult to deal with (4680 actually produce a lower heat load than equivalent size pack of the other cells). Hot spots, or heat differentials, within the cell are what cause more problems (which the tabless design is supposed to address).

In the past, how much were charged rates bumbed up in percentage terms by Tesla through OTA updates?

Sponsored