Tesla Model 3 Battery Pack Modules Made Like Machine Gun Belt

Tesla battery production


Remember when Elon Musk said batteries would come out of the Tesla Gigafactory as fast as “bullets from a machine gun”?

The hinted (and newly discovered) automation technique is somewhat similar. It would be simplified as: “making modules as fast as bullets on a machine gun belt.”

Yes, a machine gun belt. Each bullet is attached to the belt but in this case, the battery cells are the bullets and the machine gun belt is the cooling ribbon that Tesla uses to cool the pack.


Schematic of Tesla Model S Battery Module

In Model S, a ribbon-shaped cooling tube snakes its way through the pack to cool the cells (ref). In Model S and X packs, the cooling snake is not physically attached to the cells. In the Tesla Model 3 battery module, the cells are physically attached to the cooling ribbon.

How do we know this?

An article describes Tesla’s funding for battery pack automation equipment. Apparently, the article which first appeared on Teslarati has since been taken down – presumably because of the sensitive internal material pertaining to Tesla.

The article mentions one major expenditure that stands out:


Perhaps this new battery pack automation technique — which was likely intended to speed up the process — was the main culprit in the Tesla Model 3 production bottleneck

Cell to Tube Attach (CTA): $30.7M

That’s new. The cells were never attached to the cooling tube in the Model S, but apparently, they are now in Model 3.

How exactly the cells are attached is not known for sure, though speculation suggests that the cells are glued to the cooling tube. Speculation also assumes that this gluing procedure is where the Tesla Model 3 production bottleneck is.

Gluing the cooling tube to the cooling ribbon accomplishes two things:

  1. It allows for better cell cooling because the heat transfer coefficient is higher.
  2. It allows for speedy assembly of the cells into a module as opposed to manually inserting the cells.

This completed machine gun belt is then snaked in an “S” shape to form the module.

It’s so simple and obvious I don’t understand why we didn’t think of it earlier.

*Fellow engineer Keith Ritter also contributed to this article

Categories: Battery Tech, Tesla

Tags: , , , , ,

Leave a Reply

37 Comments on "Tesla Model 3 Battery Pack Modules Made Like Machine Gun Belt"

newest oldest most voted

That one and a half minute video is literally the same 5 second video clip looped over bad music for a minute and a half. I now know the point where a static image is better than a video.

Now, if we can get 3 or more machine Gun Belts going here, we’ll be talking some very serious production…

It’s the same cell production video clip that has been around for many months. CNBC has used it for many “battery” stories and the like. Here is one sample use of it…from about a year ago.

Apples and oranges.

Weather cell is glued to cooling equipment or inserted into pack, motions required are roughly the same. Automation should be as efficient in both cases. Hence no change in possible assembly speed should possible.

But “as fast as machine gun” do make for a nice rhetorics.

I suspect the 2170 cell is designed to allow thermal extraction from the “bottom” of the cell for higher and more consistent thermal dissipation. That could greatly simplify and lighten the thermal transfer mechanism. I wonder how long before someone tears a Model 3 battery pack apart to see what the Tesla team has created?!

Sounds to me like the Most Practical Approach…

Is this really the way the battery cells are cooled?

Seems like there will be huge variability in the ability to control cell temperature across the pack, with cells at the beginning of the loop getting the best cooling/heating and cells at the end of the loop getting much slower cooling/heating response.

You’d think that, but liquid cooling is so effective that you just need to ensure your flow rate is fast enough for your loop length.

Unless somehow the flow is reversed periodically?

It doesn’t need to be if the rate is fast enough.
Just like a home heating system, with circulators is better then a gravity feed system.

Even nuclear reactors are liquid cooled. EVERYTHING that is liquid cooled has effectively the same issue. ICE engine too.
It’s a matter of magnitude. Increasing flow rate of the liquid reduces the temperature gradient beginning to end. Also the temperature differential between the liquid and the thing being cooled is then smaller. Rate of Heat transfer has several inputs. difference in temperature is one of those.

It will have an effect, and if you could meassure the temperature on the first and last cell – there would be a difference. It that difference matter or not, is another question. Tesla can compensate for this by either have a cooler liquid, or let the liquid flow faster. There will still be a difference, but at least the last cell will be cooler. Being made in a desert like environment, I’m sure Tesla has compensated for this and made the right adjustment. With thermistors at different locations inside the pack they can let the car adjust this in real time. I’m sure their battery expertise keeps the battery temperature in the sweet spot. They could also have used a separate circuit from the cars AC, and aircool the unit. But you have to think about flow control, and enough space around the cells. Tesla packs are crammed with cells. I’m sure the engineers have tried several options, before they found the one they use. I’ve seen military batteries that are submerged in a liquid (oil of some kind, even silicone oil)and the liquid is cooled. That is a very good solution too, but adds weight, needs some kind… Read more »

That doesn´t explain, why they used Samsung cells in Australia:
If it was the same cooling technique – same problem.
If it was the old cooling technique they should have enough of their own cells (because the bottleneck is the glue not the cells)

Shipping cost.

A couple reasons they might have used Samsung cells in Australia:

1. GF switched from storage cells to Model 3 cells in May. They had built up stock to fill PowerWall/Powerpack orders, but didn’t have enough stock for this large order.

2. Samsung cells had the specs this project needed (a big chunk of it needs 1C discharge vs. the 0.25C of typical Powerpacks).

“That doesn´t explain, why they used Samsung cells in Australia:
If it was the same cooling technique – same problem.
If it was the old cooling technique they should have enough of their own cells (because the bottleneck is the glue not the cells)”

Old 18650 cells = assembly line in Japan.

New 2170 cells = assembly line at Gigafactory One in Nevada.

Presumably the new assembly lines at Gigafactory One are set up to handle only the larger 2170 cells, not the older 18650 cells. Shipping 18650 cells to Gigafactory One for assembly makes no sense if they are the wrong size, regardless if they’re made by Panasonic or Samsung.

It’s more reasonable to assume that Panasonic has not yet been able to set up their Gigafactory One production lines to handle different chemistries. It’s almost certain that Tesla’s stationary storage packs, the PowerWalls and PowerPacks, use cells with a different chemistry than the ones they put into their EVs. If that is so, then it appears Tesla is still using 18650 cells in the PowerWalls and the PowerPacks it shipped to Australia.

I believe read a post from an inside-GF source on the Panasonic side that Panasonic was so far ahead of Tesla with the Model 3 cell production a couple of months ago that they made a temporary change of a line to run some powerwall cells to keep busy. That was a one-shot deal. They are focused on Model 3 cells and will be back to making those exclusively going forward unless they install a new dedicated line for powerwall cells.

That’s a pretty slow machine gun

(⌐■_■) Trollnonymous

Something tells me that’s neither Tesla nor Panasonic video…….lol

Keith Ritter here To clarify our understanding of the Model 3 battery module design assembly process – what I have heard from an inside source is that the cooling ribbon tubes are rigid, not flexible, shaped like corrugated roofing, and run the length of the module. One tube for every 2 rows of cells. Each module has 14 rows of cells, so there are 7 tubes per module. Assembly steps as follows: Step 1 – cells are glued on both sides of the tube for the full length of the module. The cell/tube assembly is called a “bandolier” in the factory, hence the reference to a machine gun. Step 2 – After the glue dries, the bandoliers are stacked one by one into the module tray until there are 7 bandoliers and 14 rows of cells. It appears to me that every bandolier must be slightly different different than the others, as the combinations of “positive” up and “positive” down cells varies for each row. This adds to the production complexity, but is inherent with the cylindrical cell-based module architecture Tesla continues to use. Step 3 – with all bandoliers in-place and cells properly aligned, current collectors are placed over… Read more »

Just a SWAG, but I’d guess whatever tooling/fixture they were using to hold the assembly didn’t have a tight enough tolerance for the robot consistency. You could add a vision system, but that would eat into cycle time. Again.. just a SWAG w/out any knowledge on the matter.

(⌐■_■) Trollnonymous

Anyone else think gluing the cells is kind of not normal?

Talk about a pain in the a$s for a pack rework!

I agree. You’d think that, since they build the cells, each cell enclosure could include a clip or two that would snap into a groove or ridge extruded on the cooling line. Then only thermal paste would be needed for proper thermal transfer.

Tesla purchased automation specialist company Perbix in November to work on “fixing” the battery pack assembly automation.

(⌐■_■) Trollnonymous

“Tesla some time to convert or build more bandolier and module assembly lines to ramp up module production. Therein their recent guidance for a slow ramp-up through Q2 2018.”

So that verifies my post on another IEV’s thread on Jan and Feb ramp up numbers from some peeps I spoke to at Fremont.

Clarification on the cell orientations. Apparently I got it wrong. Model S modules have cells in both + up and + down, but the Model 3 modules have all cells assembled with + up. The top current collector that gathers the current from the + terminals of all cells wired in parallel then has a conductor to carry the + current down to the bottom neg. terminal current collector for the next cluster of parallel cells.

@HVACman (or is it Keith?)

Thanks a lot for the detailed description!

“It appears to me that every bandolier must be slightly different different than the others, as the combinations of ‘positive’ up and ‘positive’ down cells varies for each row.”

Is there some reason why there would need to be more than two types used alternately, a positive up bandolier stacked on top of a positive down bandolier, and vice versa?

Speculation I’ve seen online of the exact number of cells in Tesla’s various packs always assume a fixed number of cells multiplied by X. The fixed number of cells presumably, in this case, would be the number of cells glued to each bandolier.

Yes, aka Keith:). George and I collaborate on a regular basis. No secret there.

re: orientations. Up post I noted that I was “corrected” by my inside source about the cell orientations in the Model 3 pack. All cells have + up.

Re: Bandolier variations. Even with all orientations + up, there are several variations. First, to keep the cells staggered properly and help with the automated module assembly process, apparently some bandoliers require a “dummy” cell at the end. Second, Modules 1 and 4 have 23 sets of 42 cells in series, Modules 2 and 3 have 25 sets in series to get to a total of 96 cells in series, so these have different cooling tube lengths that must be accounted for.

Most EV battery packs of all manufacturers use 96 cells total in series to get to about 350-360 volts DC nominal operating voltage.

Hmmm, when Elon said that Gigafactory One would produce cells like bullets from a machine gun, I thought he was talking about speed, not that they were all attached to a belt! 😉 “The cells were never attached to the cooling tube in the Model S, but apparently, they are now in Model 3.” Well I dunno about “attached”, exactly, but in an extreme closeup photo of a group of cells from a Model S pack (see link below), it certainly looked to me like the ribbon-shaped tube was curved in an arc around part of each cell. Certainly it appeared to be in direct contact with the cells. Am I misinterpreting what I see? Are those sliver cylinders inside the copper-colored bands (presumably cooling loops) not the individual battery cells? “How exactly the cells are attached is not known for sure, though speculation suggests that the cells are glued to the cooling tube. Speculation also assumes that this gluing procedure is where the Tesla Model 3 production bottleneck is.” First I’ve seen of any such speculation. The previous speculation/rumors were that Tesla was having welding problems with the TM3 battery pack. “Gluing the cooling tube to the cooling ribbon… Read more »
Model S packs use a similar corrugated cooling tube, but slightly different. It is held in-place by friction and has flexible “fingers” that contact the cell case to pick up heat. It would appear the Model S tube is placed against the cells after the cells are set into the module pan. With the Model 3, the cells are first glued to the tube, then the bandolier tube/cell assembly is set into the pan. I think Tesla can get things more compact this way, with tighter clearances and less hand-work. My source is firm that the cells are glued in-place. Not snapped in. Not friction fit. One problem cited by my source is that there are sometimes problems with one or more cells touching each other or close-to-touching after being glued up and sandwiched into a module. This creates a short or arc-potential between cells, which causes the entire module to be rejected in QC. There is apparently currently (bad pun) a high scrap-rate, which really has to be expensive. Cells are not cheap. Also, there is the bigger fear that some may “pass” QC, but then during operation some cells are so close they shift slightly and can arc… Read more »

If all the cells are + up, then how does a cell touching an adjacent cell create a short? Since the can is connected to the negative side of the jelly roll, all adjacent cans would be at the same potential, at least among the 43 cells in parallel, right? So it would only be a problem when crossing a 43 cell boundary up to the next set of cells in the series stack. Not knowing the topology of each module, this is the only case I can think of where there would be a problem.

The Model S cooling tubes are friction fit only and peel right off the cells once you get down to them. Watch this module teardown video below that gets down to the cell level. About everything else is really glued together, though!


Those battery cases look like rimless ammo cases. If this EV thing ever fail to work out, they can always switch to ammo production.

Unfortunately for those of us in CA, we’re going to have to make our own ammo instead of buying “Tesla ammo” since buying ready-made could cost 10 time more than it was before 2018. Yup, all the criminals will pay to have all their ammo shipped to federally licensed dealers rather than buying them illegaly out of state or making their own. /s

Anyone know if Tesla might adopt the LG ‘pouch’ battery?


the video is for cell production which is done by panasonic, not pack production which is done by Tesla. I doubt pack production is anywhere near as fast.