It’s possible that the PCB for the graphics board itself serves a structural purpose, but a quick and dirty approach would just get a pouch battery the same size as the GPU board, leave the GPU heatsink in place, and squeeze in a charge controller close to the interposer.
At least for prototyping- seeing if a battery would even charge and discharge in this space
To give another way of explaining it:
Heatsink is not the same as heatpipe. The heatsinks for the dGPU are in the very back of the expansion module, directly adjacent to the fans. The heatpipes are over the actual GPU chip and lead to the heatsinks.
Now where the misunderstanding happened: @Chr was talking about only “replacing” the heatsinks, which would give them the full length of the expansion module to work with. You thought they wanted to use the space where the GPU and heatpipes are.
A final product should use all available space imo. If you’re going about designing and manufacturing a battery module, you might as well go as big as you can (while sticking to regulations and limits, like the 100Wh for airplanes)
I’ve not misunderstood anything. Large parts of the expansion module are inside the footprint of the laptop, which needs space for the fans.
The full width is only available if you give the battery module a massive caboose, like old ThinkPads with a 9-cell.
If framework sold a fat bottom battery for the 16, I’d buy it in a heartbeat.
I’m curious how hard it would be to add an interoperator cable to the current battery header and then 3D print your own battery shell or adapter and include the necessary control circuitry in the bay you construct to allow hot swapping. Make it as big as you like, materials withstanding. IMO, it would be best to make the bay effectively an adapter for a secondary “mainstream” battery.
I’m Looking at the massive pile of old Dell Latitudes I have that shared basically the same battery for generations. With those I can use the bigger 97wh battery that hung off the back (I don’t recall the size on the even bigger one that lifted it up off the table) and then add a 2nd battery via the proprietary “e-port” for the dock in the bottom - another 97wh. And then I could still swap the hot swappable Slimline bay for a third 30wh battery. That’s THREE hot swappable OEM batteries on one Dell E6430 with an i7, 16GB DDR3, Nvidia 5200M, SSD, Wifi6+BT5.1 and various I/O ports. 224wh of juice…ahhhh, the freedom! THIS is what I’m looking to replace with a Framework. There’s obviously a market if Dell has been making these things for generations.
I too miss the days of the big honkin’ add-on battery options. Light weight? Heck no. But I don’t buy 17" laptops to be light weight, and now it’ll run for DAYS.
Anyone considered how it’ll affect the weight distribution? Is it going to make your Framework handle like a 1970s Porsche 911 with all that weight out the back?
Pretty close. Theoretically five 18650s should be 91% as wide as the laptop.
The largest 18650 cells I can find appear to be just over 12 Wh each, so five off those would be 60 Wh. A 71% improvement to battery life would make a big difference.
I will note that I personally would prefer a design that uses the normal expansion bay shell (not the larger graphics module shell) for the charging PCB and has a separate hot-swappable part that clips on the back (using the opening on the back of the shell for the connection). This would have a similar footprint to the graphics module based idea but would allow for it to be easily removed without needing to swap expansion bag modules. That would be more expensive, but I’d happily pay a couple hundred for that in a heartbeat.
Is that 12Wh each at 3.3v? I’d be guessing the 5 batteries need to be in serial to accommodate 15v (via 16.5v total across the 5 in series), which to my understanding means the only thing would only provide 12Wh. Another 5 cells in series could then provide another 12Wh, assuming those spec values are right and my math is to be believed.
You’re mixing up Ah and Wh.
Wh = Ah × V
The highest capacity 18650 cells I see are 3.55 Ah and 3.6 V, which calculates to 12.78 Wh.
If you wire cells in series the voltages add together while the amperage stays the same. So 5 of those cells would result in a battery that has effectively 3.55 Ah (same as each individual cell) and 18 v (same as all cells combined) which would calculate to 63.9 Wh.
If you wired the cells in parallel you would also reach 63.9 Wh, however through 17.75 Ah (same as all cells combined) and 3.6 v (each individual cell).
So regardless of how you wire it the total capacity (in Wh) of the pack would be the sum of the capacities (in Wh) of each individual cell.
Edit: For example the Framework’s main battery pack has four cells, each with rated for 5.491 Ah and 3.87 V rated voltage, which calculates to 21.25 Wh. Add those four together in series and you get a battery pack with an effective Ah of of 5.491 Ah and effective voltage of 15.48 v. That calculates to 85 Wh, which is the same as if you just summed the 21.25 Wh of each of the four cells.
Ah ok, thanks for clarifying. I’m just a software engineer, so I’ll just leave the spicy computer juice for you guys to handle 
That really is a ton of capacity then, I’d slap it in my rig without a second thought. Granted, rectifying the controls and state of charge logic between two batteries is a bit tricky, but nothing that hasn’t been done (imperfectly) before.
Just be really ghetto and make it like a power bank, then run a cable from the back to a usb-c on the side
Yeah that would be really ghetto. But I think would defeat the purpose of a Expansion Bay Battery. You would lose one USB-C port. You could just use a PD Powerbank instead but…
Tbh I’ve considered velcroing a battery to the bottom and doing this. For some reason it just doesn’t feel the same though.
i would not have it right in front of the exhaust ports. I would say it would be better to have the battery be most of the green space into the blue area behind it. the BMS can be a very small thing about the size of where it connects to the PCIe connector.
Those would probably be blocked off if this gets made. Although even if they weren’t it wouldn’t be a huge issue (would help keep the batteries cool) as the rear exhaust is purely for the GPU, and if the GPU is removed then the exhaust is just room temperature air.
I went and looked at the step file for the expansion bay; it actually would be pretty easy to design an enclosure for the stock bay. For the past couple of days, I have been working on remodeling the enclosure in SolidWorks. Basically, re-assembling the parts with mates instead of everything being locked and uneditable. I am going to work on an adapter to house a battery. After that, it is a matter of having a custom board designed. I looked at the pin layout on the interposer, and it seems like there are dedicated pins for a battery module. I am no integrated circuit designer, so that will be the hard part. I have a photo of what I’m planning for it to look like.
We believe in you and bid you good luck on your quest 
Thanks man, after a bit of head-scratching It seems like I’ll be able to fit a 10-Wh battery with the stock slot (about 40min extra battery life per pack). But with a few adjustments to the slot, it is possible to fit a larger pack, which would be desirable. My intention with the pack is not to act as an extension to the laptop but as an external charging pack (in the sense that it will charge the internal battery rather than increase the perceived capacity). The pack will be hot-swappable, but I’m unsure if the majority want that. The capacity can be further increased if it isn’t a hot-swap pack. However, hot-swap would allow for multiple packs to be carried if need be.
I would love to hear some feedback on what people want in a battery add-on.
That’s less than a single 18650, you sure you can’t fit more than that?
