TESLA and new 22700 cylindrical battery coming?

[eas]

Motivations we don't have the insight to see from our limited perspectives would include ability to directly retro-fit into existing battery pack volumes.

Nail on the head. A year from now Tesla will have to upgrade the Roadster batteries for the first customers who opted for a future pack cell replacement. Six years from now the first of the Model S's will undergo the same treatment.

Unless in future they make two different battery systems one for the new cars and another separate battery system for the retrofits they are locked into using battery packs that always will be compatible with their older cars.

Meh. The sales volumes on the early roadsters was low enough that they probably do better by just using the exact same pack design and sourcing more 18650 batteries. I doubt the situation with the current Model S is dramatically different. If you are a relatively young and ambitious company investing in a big new battery plant, your primary interest is unlikely to be optimizing the economics of maintaining your old designs, you are going to be focused on future opportunities.

The reasons they didn't make a huge jump seem pretty obvious to me. Cell cooling is clearly really important. Significantly larger cell diameters are going to have lower surface/volume ratios and longer distances for heat transfer. Significantly longer cells are either going to result in a higher riding car or need to be oriented horizontally, rather than vertically. I'm not sure if there is a problem with horizontal orientation, but I'm pretty sure that a good structural engineer can make good use of a bunch vertical metal cylinders to enhance the vertical rigidity of a box like, say, a battery pack case (doesn't appear that its part of the existing design though).

As for big pouch cells, well, yes, they could simplify some things, like cooling paths, but they have their own tradeoffs. They might be slightly lighter but less than you might thing, since much of the weight of an 18650 (or similar) is in the electrochemical components, not the relatively thin metal can. Whatever weight savings you gain though comes with a reduction of structural strength, which could be an issue for both pack design (see above), but also quality control during manufacturing. The metal coat of 18650 cells have a coat that protects them during pack manufacturing, while pouch cells are much more vulnerable. Accidental damage in during pack assembly was apparently an issue when pouch cells started getting more use in (some) laptops.

 

[eas]

Oh I agree thermal management has been a critical advantage of Tesla's design, but as with their apparent intention to change the cell format, I whorls assume they would also have progressed cell chemistry and construction to improve electro/chemical efficiency to reduce heat generation at source, and thus reduce the level of waste heat that has to be dealt with. After all , heat is just wasted battery power !.
In short, I am expecting Tesla/Panasonic to have progressed cell Developemnt such that they can scale up the cell physically whilst improving performance and efficiency at the same time.

Your expectations seem quite unrealistic to me. Engineering and design of any non-trivial product or system require a huge number of variables. Compromises and tradeoffs are essential to getting anything done. At a high level, the classic dimensions are time to market, quality, and expense, but you can simplify things further to risk vs cost, which themselves can be traded off against eachother..

A great way to manage costs and risks is to focus on incremental improvements and novel combinations. I'm sure that doesn't sound sexy, but get the mix right, and it is. Apple does it, Pixar did it, and Tesla was clearly founded on these principals. Think about it. They started with an existing chassis design from another company; an existing battery chemistry, and form factor. A key insight was that the big challenge in electric vehicles was battery energy density, and that consumer electronics had already pushed the energy density of lithium ion batteries to the point where all-electric cars with decent performance and rage.

The business didn't depend on radical improvements in battery charging energy efficiency, and it still doesn't. It depends on a range of things, and I'm pretty sure that now, as then, energy/kg and energy/$ are more important. kWh from the power grid are relatively cheap. kWh/kg and kWh/$ out of a vehicle battery pack are another matter.

 

[Hillhater]

Tesla's busines from now on depends on sales success, and that to a large measure means being cost competitive and hence cost reduction in their manufacturing and assembly functions.
Tesla knows that the 18650 format is a barrier for cost efficient mass production of large capacity packs.
Increasing cell capacity (2-3 fold) gives a huge reduction in component count , complexity , and assembly time.(=costs)
Incremental improvement in cell capacity and performance has been continuous at Panasonic since the first Tesla packs , even before the establishment of the cell development team at Tesla who will have accelerated those improvements in the past few month in preparation for the new format.
I don't think any of us are surprised to know that there are rumours of a 4-+ Ahr 18650 cell being tested in labs, and I for one fully expect that thermal performance will also have been part of any development program.
Progress those "incremental improvements". Into the larger 22700 cell format and you have a serious gain in production and assembly costs
Refining the thermal management system of the pack would be another target to dramatically simplify pack construction and reduce costs.
Remember , there are 2-3 years before these changes will be introduced in quantity.

 

[justin_le]

Volumetric efficiency is only one piece of the puzzle. By increasing capacity per cell you can drastically reduce the number of interconnections, reduce monitoring complexity, move the split of the BOM towards a higher percentage of 'active' ingredients, decrease time required to assemble packs and probably 2,000 other good reasons I can't think of immediately. Makes sense to me..

You miss one of the huge reasons why in practice high capacity cells are a bad idea, which pretty much trumps all the thousands of things that would otherwise be in their favor. And that is that is with a number of small cells in parallel you have a much finer grid or comb for doing QC during production and detecting even the slightest defects and imperfections that can spell disaster down the road.

Think of it this way, a 50Ah cell has the same active electrode area as twenty 2.5Ah cells. And so you have the same probability that there will be some small piece of contamination, some piece of electrode broken off, some section with more or less plating thickness etc. Say this manifest itself as a reduction in capacity of 0.2 Ah. On the single 50Ah cell, +- 0.2Ah would be indistinguishable from the expected spread in capacity and you'd never clue in that there was a material defect in the cell assembly. While with the twenty 2.5Ah cells, you would have one cell that tested to 2.3Ah instead of 2.5Ah, and that would go straight to the reject bin and never make it to the production pack.

The first 4-5 years of lithium batteries in the ebike industry used almost entirely large format ~10Ah cells, be they polymer cells or prismatic cans or large cylindricals like headway/ PSI. We dealt with over a dozen different sources in our shop and had nothing but massive headaches after headaches. Meanwhile BionX built their batteries from a bunch of small 18650 cells in parallel, and when I saw this at the time I thought it was just ridiculous and made no engineering sense. But the BionX batteries lasted for years without any cell failures, while we kept pulling our hair out.

Nowadays, every single credible ebike on the market has a pack made from 18650's. Even a number of cheap import ebikes are boasting brand name 18650 cells in their packs. I think we may see some creep up to somewhat larger cylindrical cells where the tradeoff individual cell reliability isn't too bad, but large format lithium is a road that was already taken and abandoned by the EV industry, and I don't think many players who experienced it want to go back there again.

 

[liveforphysics]

One can go no bigger than one can ensure can be made right.

I know first hand it's possible with amazing automation equipment that uses continuous QC processes at every step to make pouches up to at least the size of a magazine and get them right. Doesn't mean its commonly achieved or easy, but it is possible.

 

[Hillhater]

.....but large format lithium is a road that was already taken and abandoned by the EV industry, and I don't think many players who experienced it want to go back there again.

Very good points, but I suspect there has been much learning and development of cell manufacture, process control, and quality management ,..over the past few years of EV programs.
When you have to produce, inventory, handle, and assemble many millions of cells into large capacity packs, assembly time and associated costs, become factors that are hard to ignor.!
I don't see Nissan, BMW, GM, Ford, etc etc moving to smaller cell formats !
Tesla is pretty much out on a limb with small format cells , and it seems they are at least moving toward something larger .

 

[arkmundi]

You miss one of the huge reasons why in practice high capacity cells are a bad idea, which pretty much trumps all the thousands of things that would otherwise be in their favor.... more discussion .... Nowadays, every single credible ebike on the market has a pack made from 18650's. Even a number of cheap import ebikes are boasting brand name 18650 cells in their packs. I think we may see some creep up to somewhat larger cylindrical cells where the tradeoff individual cell reliability isn't too bad, but large format lithium is a road that was already taken and abandoned by the EV industry, and I don't think many players who experienced it want to go back there again.

Geeze, the fabled justin_le appears. Haahaaa, just wasn't sure whether you're a myth or real. Haaahaa, Seriously, no offence.

When I went from using the A123 AMP20 cells to the 26650, it was this kind of consideration. I asked myself why Tesla/Panasonic was using the 18650 rather than larger format cells like the AMP20, made by A123 for the electric car industry. I still ponder format questions. Thanks for the perspective.

But, every single credible ebike on the market has a pack made from 18650's. Is this really true? What about Ping, cell_man and other makers? What about the Copenhagen Wheel?

 

[riba2233]

Geeze, the fabled justin_le appears. Haahaaa, just wasn't sure whether you're a myth or real. Haaahaa, Seriously, no offence.

When I went from using the A123 AMP20 cells to the 26650, it was this kind of consideration. I asked myself why Tesla/Panasonic was using the 18650 rather than larger format cells like the AMP20, made by A123 for the electric car industry. I still ponder format questions. Thanks for the perspective.

But, every single credible ebike on the market has a pack made from 18650's. Is this really true? What about Ping, cell_man and other makers? What about the Copenhagen Wheel?

Copenhagen wheel uses small cylindrical cells, cellman uses samsung 18650, and ping is not that credible, and they use lifepo4.

 

[speedmd]

One can go no bigger than one can ensure can be made right.

We see this in a bunch of products. Most industry has standardized to smaller hardware sizes that heat treatment can not mess up easily or at all. Many large industrial machines now use multiple smaller screws and nuts as opposed to large custom hardware, for both assembly and hardware consistency considerations. Even in construction they are standardizing on smaller thicknesses in beam walls and re-bar that quality can be assured much easier and reliably.

Would think a 22x70mm size cell not significantly more difficult or complex to assure 100% quality on than a 18x65mm cell unless it is very different construction.

 

[Hillhater]

One can go no bigger than one can ensure can be made right.

Would think a 22x70mm size cell not significantly more difficult or complex to assure 100% quality on than a 18x65mm cell unless it is very different construction.

Certainly the battery industry has plenty of experience with producing larger cell sizes...26650, "C", "D" "F" etc etc, in various chemistries .
The challenge will most likely be to incorporate any new chemistry, and manufacture at high speed whilst retaining quality levels.

 

[justin_le]

But, every single credible ebike on the market has a pack made from 18650's. Is this really true? What about Ping, cell_man and other makers? What about the Copenhagen Wheel?

Copenhagen wheel uses small cylindrical cells, cellman uses samsung 18650, and ping is not that credible, and they use lifepo4.

Thanks Riba, that's pretty much spot on. It's not often that I'd use absolutes like this, but seriously if you look at all the ebikes on exhibit at eurobike or interbike this year, and that's an awful lot of makes and models, they are across the board using 18650 packs. And the ebike OEMs I know who used to deal in larger format cells (from cell manufacturers like Phylion, AEenrgy etc.) have switched to reputable 18650's as well. You'll see some small cottage industry players that will have ebikes running PING batteries, but nobody in their right mind would try to scale that into a serious production. This isn't because people are just jumping on the bandwagon and copying Tesla or whatever, it's because with currently available cell technology this is the only way to make a large lithium battery that has the reliability needed for a consumer product.

 

[justin_le]

Certainly the battery industry has plenty of experience with producing larger cell sizes...26650, "C", "D" "F" etc etc, in various chemistries .
The challenge will most likely be to incorporate any new chemistry, and manufacture at high speed whilst retaining quality levels.

Here I'm talking way outside of my firsthand expertise and knowledge, but I get the impression that lithium cell manufacturing is an entirely different beast than all the rechargeable chemistries that preceded it. There are massive format Lead Acid batteries, NiCad packs etc. which seem to scale up just fine in size. (Mind you, we did deal with more than our share of 'F' sized NiCad and NiMH cells that burst apart violently in use, while I've never seen this from the consumer sized AA NiMH/NiCad). But with lithium there seems to be a lot more at stake, and with way less deviation from perfect that can tolerated. And that gets much harder to manage the larger the final cell size.

The other thing to mention is that while it seems like having a few large cells to connect is easier than tons of small cells, we also experienced WAY more interconnect issues with the large cell packs. Having a single thick tab that is spot welded to a single cell and needs to deal with the full ~40A ebike current is a big challenge, thick tab material puts much more strain on the spot weld point and is more likely to pop off. Cells that have terminal bolts are prone to all kinds of loosening and corrosion/resistance problems at the bolt interface. Meanwhile, the 18650 strategy distributes the full battery amperage over a large number of parallel spot weld joints, joints done in a relatively thin tab material that fuses almost perfectly to the cell can with no tendency for the spot weld to break off with stress and fatigue.

So for a home builder it seems inconvenient to deal with all the hundreds tab weld connections from an 18650 pack, and you would intuitively think this is then many more potential points of failure. In practice, the reliability is orders of magnitude better with 100's of 18650 spot welds in thin nickel sheet than with 10's of joints that each need to handle the full battery current. To give you numbers, I recall maybe 2 or 3 battery RMA's that we worked on where the fault or low capacity was traced to a delaminated 18650 tab weld. But we had at least a hundred large cell NiCad/NiMH packs with failed tab welds and, and similar problem percentages with threaded connections on the bolted cells. This isn't to say you can't make a reliable high current connection to a large format cell with proper design and engineering, just that it's not nearly as straightforward as you'd think. And 100's of 18650 spot welds is a heck of a lot more reliable than one would think.

One can go no bigger than one can ensure can be made right.

We see this in a bunch of products. Most industry has standardized to smaller hardware sizes that heat treatment can not mess up easily or at all. Many large industrial machines now use multiple smaller screws and nuts as opposed to large custom hardware, for both assembly and hardware consistency considerations. Even in construction they are standardizing on smaller thicknesses in beam walls and re-bar that quality can be assured much easier and reliably.

That's an interesting point. You'd cynically think it's just to cut cost but the same principles could be involved in some of these other areas too.

Would think a 22x70mm size cell not significantly more difficult or complex to assure 100% quality on than a 18x65mm cell unless it is very different construction.

I'd agree here, and at the level of perfection the industry seems to have dialed in the 18650 you'd expect them to scale to a somewhat larger size like the 2270 with confidence. And then yes this could mean half the number of welds and joints, but still having the reliability and QC benefit of the small cylindrical cell. But to jump from 18650 standard to something like a large 20Ah cell, that would seem pretty ill-advised from everything we've seen.

 

[arkmundi]

For large automated production runs I can see the point, and how in practice, the manufacturers of eBikes and kits have gravitated over time to the 18650 cell format. There's also likely a cost savings when using the cell of choice for consumer products of all sorts, laptops and whatnot. An OEM is going to go to a cell manufacturer, like Panasonic or Samsung, and in negotiating quantities and pricing, will be led to the cells of highest manufacturing volume, the big factories in China, running 24/7.

But.... I and many of the members of the ES forum are making our own batteries, for various reasons. We're not doing production manufacturing. And so we can pick & choose the cell and its sourcing on the basis of a number of criteria. Like, do you have or access to a spot welder or not? Having made my choice in favor of the A123 nanophosphate LiFePO4 cells, there are three availability choices: the 18650 or 26650 cylidrical or the AMP20 pouch. My two pack builds were the AMP20 and 26650 respectively. The AMP20 pack used Agniusm's kit, utilizing a screw down plate mechanical bond of the tabs. The 26650 build was soldered, because of the lack of a spot welder. Both packs are whole and in use.

But from an ease of assembly and ongoing maintenance perspective, I far prefer the AMP20 pack. I recently rebuilt that pack to deal with the compression issue raised by wb9k and in considering approaches, have one that makes the build straightforward for the guy at home with a small workshop and common tools.

The 18650 cell pack build would have been worse for me, because of the higher number of cells needed to make capacity, compounding the issues faced in the make. My next pack build will be back with the AMP20 cells. For Tesla, its a new 22700 cylindrical format. Two ends of the spectrum between lowest volume (three batteries), and a half million a year.

 

[speedmd]

I'd agree here, and at the level of perfection the industry seems to have dialed in the 18650 you'd expect them to scale to a somewhat larger size like the 2270 with confidence.

This most likely will still be a challenge on the larger format or any significant change short term. Lots of things can and most likely will go wrong on even simply changes. A few years back I was involved with a small production mold trial that was running 16 cavity seal caps for standard AA alkaline batteries for one of the majors. It was difficult getting the process and all the fits into very tight dimensional specs. and that was the easy part. After we aged parts with thermal cycling, it got real interesting. This was all before they were even considered to be assembled into finished test cells. Leaks were taken very seriously in that product as it was the most expensive losses most of the companies faced at that time. Would think the 22x70 will be much more critical in many more areas.

With cells going into large pack assemblies and used in very expensive products, it takes the quality demands to a whole new level. Costs of having a failed product in the field are HUGE in so many ways. Saving a few percent in cost of assembly or cell build are not even close to being viable trade-offs to adding any potential quality / failure issues. Just poor for long term business success to allow any added potential for failures. I would imagine they will test the hell out of them for some time before making the change over.

 

[arkmundi]

My next pack build will be back with the AMP20 cells. For Tesla, its a new 22700 cylindrical format. Two ends of the spectrum between lowest volume (three batteries), and a half million a year.

I tend to think things thru too much, I suppose, a liability of being old, having too much time to mull things. And become so spun around by options and facts that I loose my head. I considered a lot of options for battery for my current build on a Ross Mt Whitney frame, including 26650 and 18650 from A123, Panasonic and Samsung, as well as prismatic cells from Electric Car Parts. This latter has a shipment arriving soon from China with NEW ENERGY >3700 CHARGE CYCLES - 20 AH prismatic cells in Aluminum shells for compression. Figured it might be a good option to the A123 AMP20 prismatic cells.

Anyway, I keep gravitating back to A123, and then back to the AMP20's. So got my head unstuck and placed an order ... see PROMOTING OSN A123 20ah packs.

 

[Doctorbass]

One of the big player in this industry is the cordless powertool.

The mechanical engineering of these deserve a lot of attention I think.

I have big experience of disassembling these Makita pack and many other brand like Dewalt, Bosh etc.. and built a lot of ebike battery with these. Over the years I saw really big improvement of the technology they use in their pack.

Personally I think that due to the very hard environment these pack are used in and with all these severe impact, hard use and also the vibration these tools are taking this give us a great example of how their manufacturing process need to be.

Ex, for the Electric car battery, they use heavy pack and have their cells taking much less vibration than powertool. The frequency of vibration they get is pretty low compare to impact of smaller pack like powertools. The powertools pack need to be able to take impact of many G at high freq and their spotwelded cells need to sustain that while electric car battery require more attention on the thermal management I think.

One big exemple of the technology improvement is the way the powertools cells are spotwelded.

The Makita BL1830 and BL1815 pack in 2008 and older had thick nickel connections with only two spotweld to each cell tab. Than in 2009+ they choosed a thinner nickel connection with 4 spotweld. than on most of all the 2009+ defective pack I got the broken spotweld problem nearly dissapeard.

The defective pack I had in the 2008- had a lot of the cells with broken spotweld and it was often the cell in the end corners of the battery shell wich are more subject to take impact energy I the 3 axis.

This is only a small exemple of all the other improvement they did at Makita. The lastest improvement of Makita is their jump to the 2.5ah cells capable of high c rate!

These pack have a lot more design technology than we think!

Doc

 

[grindz145]

It will be interesting to see how it all shakes out. I can't believe that tesla has made 18650s work as well as they have. It's absolutely incredible. I have to believe that the larger form factor has to become dominant at some point. The overall system reliability (at least from a classical MTBF standpoint), will ultimately be greater with fewer interconnections. I can't believe all these years later 18650 is still the dominant form factor.

 

[Punx0r]

It will be interesting to see how it all shakes out. I can't believe that tesla has made 18650s work as well as they have. It's absolutely incredible.

Absolutely agree.

Justin's point regarding interconnect reliability on small Vs. large cells is a good point. However, I feel the multifarious reliability advantages we see for 18650 cells is due to the extensive development work they have received. I can't see any reason why, for example, an equally reliable spot-welded interconnect couldn't be made for any size cell you wish. It's just the time and resources to required to develop it.

 

[liveforphysics]

It will be interesting to see how it all shakes out. I can't believe that tesla has made 18650s work as well as they have. It's absolutely incredible.

Absolutely agree.

Justin's point regarding interconnect reliability on small Vs. large cells is a good point. However, I feel the multifarious reliability advantages we see for 18650 cells is due to the extensive development work they have received. I can't see any reason why, for example, an equally reliable spot-welded interconnect couldn't be made for any size cell you wish. It's just the time and resources to required to develop it.

The path forward uses no welds in pack construction. Welds are an unneeded hot spot on the cell and a relic of antiquated pack architectures IMHO. This is not to say they can't be used to make a successful pack, the Tesla pack is incredibly good in practice. It simply could be smaller, and even more robust if it didn't involve a bunch of welded tiny cells (though the individual cell-level fusing is awesome!).

 

[speedmd]

If I read Justin's post correctly, spot welds are not the major failure mode on the small cells and more a issue in the large cells. Also Doc notes issues with the thicker tab material in older makita packs. If tab material is much thicker than the battery cap material, it will be troublesome getting a good weld bite into the thicker material. A small stress relief (tiny U bend) in the tab between cells should take care of shock and vibration loads. Wondering if the end welds that came apart are torn or just unstuck. We saw this happening with power factor correction capacitor banks in the early designs, and it was greatly improved by matching material thicknesses closer of the end caps and tab. After extensive review it was determined that they were actually failing by migration of the weld metal over time when dumping high current. In the larger cell packs, this may be the case also on less than perfect welds.

 

[Punx0r]

Exactly, match the material thickness by increasing cell cap thickness, or decrease interconnect thickness and use greater width and more welds.

I imagine Luke is alluding to pressure-contacts? It will be interesting to see how that technique pans out. I imagine there are a few challenges in maintaining low resistance over many years of vibration and environmental attack.

By incremental improvements we progress

 

[grindz145]

It will be interesting to see how it all shakes out. I can't believe that tesla has made 18650s work as well as they have. It's absolutely incredible.

Absolutely agree.

Justin's point regarding interconnect reliability on small Vs. large cells is a good point. However, I feel the multifarious reliability advantages we see for 18650 cells is due to the extensive development work they have received. I can't see any reason why, for example, an equally reliable spot-welded interconnect couldn't be made for any size cell you wish. It's just the time and resources to required to develop it.

The path forward uses no welds in pack construction. Welds are an unneeded hot spot on the cell and a relic of antiquated pack architectures IMHO. This is not to say they can't be used to make a successful pack, the Tesla pack is incredibly good in practice. It simply could be smaller, and even more robust if it didn't involve a bunch of welded tiny cells (though the individual cell-level fusing is awesome!).

It seems appropriate for the level of technology as it is right now, for a 60kWh battery pack, but it certainly seems like it will be antiqued.

(side comment, the term "Antiquing" , as in grabbing a handful of baby-powder or flour, and throwing it at your unsuspecting friends is great fun, but not what I am referring to here)