Tesla Model 3 Battery Current Collector: Radically Different Design


Compared to the Tesla Model S and X, the Model 3 current collector design is quite different.


We have a reason that we will explain later in the article, but, for now, let’s just review what Tesla did to the current collector design in the Model 3. Then we can speculate on why they were made.

What are current collectors? They do exactly what they are called. They collect current off the individual cells. Many amps worth, and they are not big wires. They are aluminum plates in the Model S and aluminum fingers in the Model 3. Tiny micro fuses attach each cell to the current collectors.

In the Model S there are current collectors on the top AND bottom of each module. Also in the Model S, each parallel cell group flips polarity up and down. Some groups are up and some are down.

Here’s a screen shot from a video by Net Positive LLC on Tesla electric boat battery teardown that shows a module with cell groups switching direction (polarity) up and down. I’ve referred to this video many times as it gives you a great understanding of how the Model S and X battery modules are put together.

In Model S modules the cell group polarity flips up and down. Credit: Net Positive LLC Tesla electric boat battery teardown.

It’s a bit difficult to see in the figure, so we made a simplified schematic that illustrates the current collectors on both sides and the polarity switching up and down in Model S and X battery modules.


In the Model S and X the current collectors are on both the top and the bottom of the module and cell group polarity switches up and down.

What are the big changes in Model 3 modules?

In the Model 3, the cells all point in the same direction. They don’t flip up and down as in the Model S and X. Also, ALL the current collectors are now on the same side of the cells … and get this: the cells all face positive-end down, with current collectors on the BOTTOM of the module.

Inquiring minds might want to know, since both negative AND positive current collectors are on the same (positive) end of the cells, how does the negative end of the cell get connected to the current collector on the other side? Does Tesla run a wire from the negative end of the battery to the collector? No, it’s a totally simple, totally out-of-the-box solution. Remember that the negative side of the cell includes the whole cell case, not just a button on the end. Tesla attaches the micro fuse to the EDGE of the cell case.

You can see that in the following screen shot from Autoline After Hours with Sandy Munro.

Tesla attaches the negative micro fuse to the EDGE of the battery case. Credit Autoline After Hours, Munro and Associates

Here’s another marked up screen shot that shows how the collector gathers all the current from the positive side of the cell group. Then overlapping the next parallel cell group, another set of micro wires feed into the negative ends of the batteries in the next cell group.

Screen shot showing how the current collection gets done all on one side of the module in the Model 3. (Original photo credit Munro and Associates: Markups by Bower and Ritter)

Great example of thinking outside the box.

The screen shot is a bit difficult to see, so we made another simplified sketch showing all the current collectors on one side, with all polarity going the same direction and the current collectors on the bottom.

Schematic showing Model 3 module with all current collectors on the bottom and with all cell’s polarity in the same direction.

Below is a screen shot that verifies that the current collectors are on the BOTTOM of the module in the Model 3. You can see that the little attachment feet on the module, which are on the bottom, are on the same side as the current collectors (the down side).

Screen shot verifying that current collectors are on the BOTTOM of the module in the Model 3. Credit Munro and Associates.

Why did Tesla do this? We can only guess. Our guess is that the new Bandolier design of the Model 3 modules and Grohmann’s automated assembly machine worked better with this set up. Another reason: Lighter current collectors, since they are now fingers instead of big plates, and thinner resulting in a shorter pack since there’s only one plate thickness and not two like in the Model S.

Now here’s another totally off-the-wall reason why Tesla made the changes:

It leaves the design open to be easily adaptable to a flat plate cooling scheme.  The smooth negative side of the cells are available to mate up to a flat cooling plate. You ask: Why would Tesla want to switch to a flat plate design that we know from a detailed thermal analysis is inferior to Tesla’s cooling snake?

Ease of assembly and cost.  Also, the flat plate design could double as a structural member between the multiple layers of battery in the semi-truck design and would allow tighter spacing between the cells, exactly like the Rivian pack.

Unfortunately, we see the flat plate design only being suitable for the semi-truck battery pack because cooling should not be as big an issue on the semi-truck since it only discharges at a C rate of around one which should minimize heat rejection. Sedans like the Model 3 would still need the cooling snake because of the higher discharge rates and track mode.

Do we really think the semi will use a flat plate design? Probably unlikely since it would destroy commonality between semi and Model 3 modules, but it’s fun speculation nonetheless.

Let us know in the comment section why you think Tesla made the big changes to the Model 3 pack we just described.

This article was a co-production with the author and Keith Ritter

*Additional Tesla Model 3 discussions at our InsideEVs Model 3 sub-forum here

Categories: Battery Tech, Tesla

Tags: , ,

Leave a Reply

80 Comments on "Tesla Model 3 Battery Current Collector: Radically Different Design"

newest oldest most voted

If batteries impressive they won’t need as much cooling.

Isn’t cooling a big issue during charging, too? Semi will charge just as fast.

As this is widely recognized as the best automotive technology in the world, what would you find impressive?

Was supposed to be “if batteries improve”. Autocorrect strikes again. They haven’t improved energy density in years, presumably choosing to focus on cost, cycle life, etc. If they can also improve internal resistance and/or temperature sensitivity, cooling would become less important and flat plate or other simpler method might suffice.

What George and I still haven’t quite figured out is what Porsche is doing with the Taycan to achieve 300+ kW (almost 4C) charge rate without toasting the cells using just a glycol-cooled bottom plate to cool the modules. I suspect it is a combination of using the pack’s thermal mass via pre-and-post-charge cooling, advanced cell chemistry, and perhaps cell tab-cooling via either natural convection with engineered fluid within voids at the ends of the modules or conduction via filling the module with a super-high-thermally-conductive silicon gel.

I keep wondering if they are simply going to allow a super fast charge every so often, and let a bit of damage through.
It really does not make sense how they are going to accomplish it, unless it is simply F’ing magic.

It may be able to charge fast but that is not that much of a big deal. The big deal is why can’t Porche get the discharge rates out of their cells that Tesla can? The acceleration on the Taycan is nowhere near as high as P100D ludicrous. And Porche is coming to market many years after Tesla could do this!

You cool the battery tabs and pull the heat directly from the metal foils in the cell.

Porsche is using 800 Volts: 350 kW / 800 V = 437 A
instead of: 500 Volts: 350 kW / 500 V = 700 A
Remember: C-Rate uses Amp-hours not kWh.

why the down votes? It s math…

It’s bad math. Higher pack voltage just means more cells in series and less in parallel. The current and thus C-rate per cell is *exactly the same*.

Uh, Tesla’s chemistry is supposed to have improved ED over Panasonic’s, as well as lower costs, etc.

Tesla has a sophisticated battery cell analysis lab, but it’s Panasonic — not Tesla — which keeps improving the chemistry year after year. And it’s Panasonic — not Tesla — who makes the battery cells which go into Tesla cars.

Tesla does have patents on the internal arrangement of cells, and it’s almost certain that Tesla’s battery analysis team works in close cooperation with Panasonic’s battery designers. But to call it “Tesla’s chemistry” seems to me to be quite an overstatement.

Musk was asked on the call how they could use a Chinese manufacturers cell. He said yes it doesn’t matter that all the cells are the same.

While great cells now, hardly the best in the world.
LG cells are far superior, 2x higher output, lower resistance thus make less heat and likely last longer thought both over 20 yr life trend.
And far easier to make a pack from, cool.
Volt module cooling is the world’s best cooling. Thus why most racing EVs use Volt packs combined with LG cells.
On cooling the major thing is keeping all the cells at the same temp so they act the same and not get out of balance.

Sorry but no car with LG cells has anything like the performance of a Tesla.

Specifically what “performance” are you alluding to? Racing speed? I believe we are discussing battery technology. Do you have any facts to assert in your conclusion that only Tesla (panasonic) has the best batteries?

there are 4(3*) makers of cells that Make quality High end Cells
the last 2 really are kinda one. but for WH/kg Panasonic has the highest numbers without cheating like Samsung did years ago with the 4.35 cells which were really 4.2 rated cells Overcharged a little. im still not sure why they have not Thinned out the case and gone with something Like the way an Alum can is made in that small form factor. you could decrease the weight which increases the wh/Kg ratio and Increase the heat Dissipation… its not like the cells need to be as strong as they are not a Structural component… there has to be a reason for this… i’m at a loss as to why its not done.

Sanyo is now part of Panasonic many years ago.

You know that is an interesting question. I must take apart a Panasonic 18650 battery and weight the casing. In a Tesla power pack it does not have to be thicker than a foil.

Based on internet search it looks like the casing is around 8 grams in weight. Or about 15%. So there are possible savings there too. But maybe this will make handling the cells too delicate.

Using a pouch-type foil casing, instead of the hard metal “can” around the cylindrical cells which Tesla uses, may make the cells themselves lighter, but that’s decidedly not an advantage when the pouch cells have to be glued into a rigid framework to prevent them from shifing around. Surely that framework weighs something, not to mention it’s an unnecessary expense for any EV maker using cells which are encased in a rigid metal can or casing.

“LG cells are far superior, 2x higher output, lower resistance…”

Wherever you are getting your info from… you should start ignoring it. No part of that is even remotely true.

It’s hard to be certain because both battery cell makers and EV makers hold details about batteries as trade secrets, but there is a pretty strong consensus among EV industry watchers that Tesla is using the most advanced, and lowest cost per kWh, cells in the industry.

NMC cells (not just LG ones) generally have lower resistance than NCA ones. That doesn’t make them “superior”, though — it’s just a different trade-off…

Thank you, Keith Ritter and George Bower for your continued research on EV batteries, particularly as it pertains to manufacturing. This is where at least half of the cost improvements will come over the next decade. While chemistry improvements make up the other half, it is exciting to this old manufacturing engineer to watch the changes through your articles.
With at least 150,000 Model 3s being produced per year, I am also intrigued by the performance data that will follow. How have these manufacturing changes improved safety? It appears in the early data that they have achieved a new level of reducing thermal events. How will the BMS manage vampire drain? I feel like just the last OTA update has reduced this once again. Will the battery life exceed that of the Model S&X which were already quite impressive. It’s a new industry and rapidly replacing the old guard. Tech vs commodity, it wins every time!

I believe they are over 300k already.

Like you, I believe that Tesla’s REAL IP is their manufacturing. In fact, I’m sure that Elon has disregarded the board and is now building an entirely new design on MY. I also suspect that the board will back him since it will mean more automation, faster ramp up and lower costs.

Elon said on the last earnings call that Model Y shares 76% of Model 3 parts.

They might have a new pack architecture — there was some mention of that on an earlier earnings call — but these would undoubtedly be applied to both models. In fact it sounded like they might debut with the standard range Model 3…

Thinking about this for the future of EVs, as the cells and manufacturing get better, the ability to make a pack with larger cells, but more degradation due to less efficient cooling of entire cell will help enable cheaper EVs. They won’t last as long as a Tesla or have the same performance, but the packs should be cheaper eventually. Tesla has set the benchmark for battery packs and I applaud them for it, but the vehicle itself will not last as long as the pack, so inevitably the packs will be designed to last about 10 years before x% of degradation.

Efficiency in recycling the packs is the next major step.

you should really look into cycle count of the Cells Tesla uses. dont expect more then 10 years out of them… Panasonic doesn’t. and they make the damn things charge them up to 4.2 discharge them down to 2.8 and you are lucky to get 500 cycles. whats the range on 500 cycles? oddly enough… its 100k miles. Right around the packs Warranty period. Convenient right? we cant even see alot of data on the degradation of the cells in a true manner since no 2 people charge alike. and noone is logging the data but of the 100k MI cars in So-cal ive found. not a single one of them gets even 90% of their reported range… sure its good to see that. but at the 7 or 10 year mark that car is going to give 70% or less… eventually it will be like guessing how many miles you are going to get since the car cant even really be sure. its only monitoring voltages.. this does not even take into effect the HUGE damage Supercharging does to the 3250 and 3750 cells in the S/X. you are charging 1.6-1.2C rate. which is worse then a fast discharge.

The problem with what you said is we already know of ten year old packs that have done a lot more mileage and still keep up their SOC. It looks like the batteries last a lot long than what they are warrantied for,.




The actual data we have on the aging of Tesla packs, pretty strongly refutes every part of your Tesla bashing claim, which appears to be intentional FÜD.

For example, the Tesloop cars probably have the highest mileage of any Tesla cars owned by the public, and their use is hard on the battery pack; they are charged daily at Superchargers rather than slow chargers.

“Back in June 2018 one of Tesloop’s Model X 90Ds, dubbed Rex, achieved 300,000 miles on its original battery and drive units in 1.75 years. Battery degradation over the 300,000 miles was ~10%…

“Tesloop also has had 5 other Model Xs that have each traveled over 200,000 miles on their original batteries and drive units. The vehicles are under the standard 8 year warranty. Tesloop does not have and never has had any extra service deals, warranties or agreements with Tesla.

“The company believes that the Model S can drive another 600,000 miles over the next 5 years, while under warranty.”


Aside from the other nonsense already pointed out by others, “only monitoring voltages” is plain BS, as anyone working with Li-Ion batteries knows.

What if its to future proof the battery modules for additional cooling, for double staked (Back to Back) battery pack modules. Such as may be used in the future Roadster

This is a good though as well, it may not be a wholesale switch to a cooling plate, but the plate may offer the ability to stack the packs without causing too much heat buildup.

I don’t see how it would make any difference… The point of a plate is to move heat from cells to the cooling system. If it’s not attached to the cooling system, it does nothing.

Fascinating stuff. Thanks for the insights…

@ George Bower —

“Why would Tesla want to switch to a flat plate design that we know from a detailed thermal analysis is inferior to Tesla’s cooling snake?”

? Ok. I’m confused. I thought the primary cooling design in the model 3 is the curvy “radiator bands” (about the same height as the cells) that the cells are glued to, .. the radiator bands are hooked up to a manifold, moving the cooling fluid through from one end to the other of the module (i.e. the cooling fluid no longer snakes through).

/ you can see the “radiator bands” cut through in the last Jack Rickard “Tesla model 3 battery secrets revealed” post.
//the model 3 design looks like a better cooling system to me. More uniform temperature control throughout the pack (manifold vs snake) and more contact area with the cells (radiator ribbons touching the cylinder part of the cells on both sides + the blue goo which is likely a thermal conductive material).

Edit: Jack shows 7 cooling channels (radiator bands/ribbons) connected to 7 manifold outlets starting 9:15/1:59:22 in the video,
/each cell has contact with just one channel (not both sides)

Yeah, I was pretty sure the cooling ribbon (“snake”) was in contact with just one side of the cell, altho in the Model 3 it runs from top to bottom on that side. At least, that’s what photographs show.

You are correct that the current Model 3 module uses aluminum micro-channel “snakes” glued to the cells to cool them. And it is an amazingly effective cell-cooling system. It is a vast improvement over the Model S modules. The problem with this design is that it has a lot of parts and pieces with a lot of potential failure points, both during assembly and in the field. Tesla’s first attempt at automated assembly of this module design was a total failure, which was why production was so slow in 2017/spring 2018 until Grohmann developed a hybrid automated/manual production line. I suspect it still is a challenge, with a lot of scrap or hand re-work and likely a lot more expensive per pack than Tesla expected. And why we haven’t seen the standard range M3 yet, as Elon has essentially stated. Most of the other manufacturers are going to bottom-plate cooling because of it greatly simplifies module and pack assembly. Porsche is achieving almost 350 kW charge rates with it for the Taycan. So there are ways to use bottom plates to effectively cool the cells at high current charging rates. Tesla may be figuring out those ways, too, which would… Read more »

Ok. thanks

If only some generous soul would donate a Taycan to Weber State, would love to see John Kelly do the tear down on that drivetrain and battery pack.

Achieving a high charge rate is impressive but doing it while maintaining battery life is even more impressive. I’m interested in finding how well Porsche system will perform over an extended time period. Is there a trade-off? Is it worth it?


I think pretty much all of us EV enthusiasts would like the answer to that question! I think the jury is very much still out on whether or not Porsche can deliver on its promise of being able to charge that fast on a daily basis without prematurely aging the battery pack.

The trade-off is that the higher current cells have somewhat lower density, i.e. the pack is somewhat heavier and more expensive for the same capacity.

For one thing, they are all using 800V architectures, which means that they don’t need as much current as Tesla would to charge at the advertised rates. Why their acceleration can’t beat Tesla’s, I don’t know. I suppose that C-rate is a limiting factor, but not the only factor in driving performance (acceleration and regen).

Pack-level current vs. voltage doesn’t affect charging speed in any way. It just allows for thinner wires. Charging speed is all about the cells. (Cooling playing some role in that — but it’s more about the internal chemistry and architecture of the cells.)


“Most of the other manufacturers are going to bottom-plate cooling because of it greatly simplifies module and pack assembly…. Tesla may be figuring out those ways, too, which would make their module and pack assembly a lot easier.”

Well, Elon did talk a few months back about a very different new design for the Model 3 battery pack, one they intend to use in the Standard Range version.

However, I can’t imagine how Tesla could switch to a simple bottom-plate design (or actually top-plate, since all the connectors are on the bottom), which would result in significantly less surface area on each cell; far less area in which to transfer heat out of the cell. But it wouldn’t at all surprise me if Tesla can figure out a simpler and cheaper way to do it than what they are doing now with the Model 3, which is to glue each individual cell precisely in the correct position on a metal cooling ribbon which has to be curved to fit the side of each individual cell.

The major limiting factor is heat transfer *within* the cell, not from cell to cooling ribbons/plates. And bottom cooling cylindrical cells is actually more efficient in terms of internal heat transfer, as the heat doesn’t need to move from the bottom of the can (where the hot electrode is attached) to the side walls.

Porsche’s stated charging time figures, as well as prototypes spotted in the wild, all suggest that the Taycan will actually have a ~250 kW peak charge rate. The 350 kW charger speed figure is likely intentional marketing nonsense, just like 80 kW for the Bolt, 100 kW for Hyundai/KIA, etc.

As for Tesla switching to a bottom cooling plate, there are indeed various hints at such a possibility. For one, it’s mentioned as one possible configuration in Tesla’s patents. Also, there are rumours that it’s actually the design used in the Powerwall 2 / Powerpack 2 — though I still haven’t found any real confirmation of that… And of course Elon’s claim of a much simplified pack architecture would fit with that as well. (Though it would pose the question why they didn’t go that route in the first place?…)

However, I don’t agree with your suggestion that the current collector design is for future proofing. With bottom plate cooling, the entire manufacturing process would completely change anyway. The actual reasons might include space savings and material savings (which you already pointed out); simpler manufacturing due to all bonding on one side; as well as better fire safety design, with all cells venting in the same direction.

I think the new design can be as simple as ease o manufacture, with all of the current collectors on one side it means only needing to weld the connections from one side. i would expect that would be a huge benefit to ease and speed of manufacture.

^^THIS. No need for robot handling to invert half the cells prior to welding, that machine is gone. No need to invert the entire pack after one side has been welded to do the other side. That machine is also eliminated.

The first graphic says the Model S has 210 cells.

Typo, should say 2170 cells referring to size, not quantity.

The “21700” (correct name) means the cell diameter is 21 mm and the length is 70 mm. This is the Model 3 cell size. The Model S and X use the more common “18650” cell size or 18 mm diameter and 65 mm length.

Panasonic has dubbed them “2170” cells, dropping the final (and redundant) “0”.

Other battery makers are still using the “21700” designation, so it may be best to think of “2170” as a brand name specifically for this type of Panasonic cell, rather than the name of a form factor.

The final ‘0’ is not really redundant. It’s the nomenclature used for all round (cylindrical/button) cells across the entire industry. 2170 would mean 7 mm height. (Which is actually a plausible size of button cells.) Panasonic is just being silly.

They might mean modules. The S does not use the 2170 cells.

I truly believe that Tesla is experimenting with the New* John Goodenough Solid State Glass Battery on the Semi Truck & the Roadster. Tesla May Already have “Small Scale Production” Going . Goodenough said they did 15000 Charge/Discharge Cycles with the glass battery and that it gained capacity with each cycle. “It does work” thus far. However., Goodenough said that it was up to whomever took it on to mass produce the Cells , to put together a Production method to Mass Produce the Glass Battery to make it massively viable . Goodenough re iterated that the manufacturing Process was be very similar to the process used to make our present days lithium Battery Packs. I honestly believe/think this is where Tesla is headed battery Wise.. ie… There are Certain Areas in the gigafactory that are Guarded* (((3rd Flr))) Restricted to All * Except to a Chosen few of Tesla Employees…

Given recent changes at Tesla, I’d say it’s a cheaper way to produce more arrays per hour, or have a lower rejection rate at the testing phase.

Not to be too picky, but your diagram of the Model S battery pack says “2170” for the cell size. Perhaps Tesla has changed the design, but I was certain that only the Model 3 used that cell size (21700, Tesla — please stick to standard naming conventions! 🙂) and that the other models used the 18650.


Isn’t the reason for the design as simple as “just one side to do wire bonding on?”
Beats flipping it over and working on both sides.

Partly or mostly I think you’re right, but using current collector “fingers” instead of a solid plate certainly cuts down on weight, and possibly cuts a bit of cost too.

Well, the plates weren’t entirely solid either. (There are holes above each cell for attaching the wire.) I think material savings mostly come down to shorter distances and/or lower peak current per area.

What happens when fuses open up?

Ooops ! …… lol

Presumably that is a shorted cell resulting in lower capacity.

the fuse or current limiter opens stopping the flow of current. The monitoring computer should then enunciate the malfunction to alert the owner / operator and maintenance. Those batteries and have short detection and protection.

Nope.Tesla packs have always been designed to cut cells which have gone bad out of the circuit and keep on functioning. One advantage of using lots of small cells instead of a few large ones is that a Tesla pack can have one or two cells go bad yet have the degradation barely detectable, and certainly unnoticed by the driver.

That’s just one of many ways in which Tesla’s EV tech is superior to everyone else’s. (If you don’t believe me, just listen to what Sandy Munro has to say on the subject!)

I highly doubt that’s actually the case. Balancing needs would become horrible if one of the cell groups suddenly lacked an entire cell.

Also, even if a short-circuited cell is cut out of the main circuit, it’s still a fire hazard. Continuing to use the battery would be crazy.

That cell is disconnected. The battery pack may lose a tiny amount of capacity (watt-hours) per cell, but the battery will keep going.

If a fuse blows due to a shorted cell, you lose 1/4512th of your range and 1/47th of the peak power in an M3 LR. If a second one blows, you lose an additional 1/4512th of range, and (unless the cell is in the same row) no additional loss of peak power. I bet they don’t even report this. The cells are like pixels in your smartphone or HDTV — if you read the fine print in the warranty, the manufacturers consider a few “off” pixels normal, and will only replace the screen if there are more than 5 or 10 dead ones.


Fairly off topic, but is there any chance you or someone can write an article on the Rimac battery cooling in the Koenigsegg Regera?

“Rimac developed a complex Fully Flooded Thermal Management System and a fully machined structural housing”
( http://www.rimac-automobili.com/en/latest/news/rimac-helps-bring-world-s-most-powerful-production-car-to-reality/ )

I assume this means coolant flows physically around the cells instead of being in a flattened tube or plate. I assume this is the optimum cooling solution, but a bit complex to do with some significant potential downsides. I have seen no mention of this concept in articles about battery cooling.

There are several existing or upcoming battery makers employing this design. Kreisel Electric has been touting it for years.

One downside might be higher weight due to more coolant. The major problem though seems to be that the special non-conducting cooling liquid needed for that is super expensive.

This concept is not new, only different from what most battery assemblies are done for cylindrical cells. Actually, it is a twin to how regular lead-acid cells are attached in series because both methods use the approach of having both positive (anode) and negative (cathode) cell electrodes on the same side or end. The cylindrical cells have a very large cathode which is the entire casing, and that casing makes a ring (or edge, as described above) around the anode, so attaching a wire on the ring allows electrical contacts on the same end as the anode.Thus, only one side has all the current collectors. As I said, this is different, but not unique nor “radical”.

“InsideEV” should have added a diagram of how each individual cell is wired for laypersons to understand. I am an EE with 45 years of experience so I do understand how Tesla and Panasonic did this.

Actually, having both current collectors on one side is the standard design for prismatic canned Li-Ion cells as well. (And for some pouch cells too.) This is a completely different situation though, as in those cells, the casing does *not* double as electric contact. That’s only the case for cylindrical cells — and I’m indeed not aware of any other design using this bonding approach with cylindrical cells.

Ha ha. This will make inspection and testing easier. Especially when on the car. Also, encase the collector in silicone or rubber liquid applied then once cured you could emerge the whole thing directly in contact with a fluid coolent.