I must admit it bugged me that there is a lot of unused volume below the keyboard of the Pocket Reform. Even before I had received my unit I looked into finding some lithium pouch cells which offered more capacity than the 4000mAh cells from Eremit and which would fit below the keyboard. Investigating the available options made me realize how difficult sourcing these things is. I got so frustrated that I was close to accepting that there really are no strictly better cells that fit into the available space and which can just be bought off the shelf, that I started considering to 3D print myself a custom bottom plate which would let me fit a thicker battery under the keyboard. Does anybody know a reliable way to find lithium ion pouch cells satisfying certain size constraints? There is a common naming scheme but that does not help much unless you already know the exact dimensions. I was close to throwing the towel when I stumbled across the magic number: 7565121 (7.5 mm times 65 mm times 121 mm) I was unable to find these from a much better place, so I took the risk and got some from amazon seller “HXJN”: https://www.amazon.de/-/en/gp/product/AmazonKillsOurPlanet/B091YCRQXL/ This is how my Pocket battery compartment looks like right now:
So yes, they could be another 8 mm wider each but they are a precise fit in length and thickness. I had to solder JST PH2 plugs onto these. I used tape to hold the cells close to the temperature sensors. I should 3D print some new plastic inlays for the left and the right hand side. Lets compare the battery runtime with these cells against the data I collected in this post: Battery life tests with RK3588 - #15 by josch
Fullscreen YouTube with full display brightness: 02:26 h ->05:06 h (209%)
Actually watching a movie: 02:47 h → 05:53 (211%)
Trying to conserve as much energy as possible during use: 04:56 h → 10:06 h (204%)
Maximum CPU stress: 02:04 h → 04:21 h (210%)
sway: 04:59 h → 10:11 h (204%)
The cells are advertised as 8200 mAh which is 205% of the 4000 mAh Eremit batteries. This fits my runtime improvements which range from 204% to 211%.
So even when trying very hard (all cores at 100% and full display brightness) these cells give me more than 4 hours of runtime. Under normal use I can get more than 10 hours. I hope that the cells will not explode into my face one day.
Good question. I asked minute beforehand of course. The answer was: no. So I’m currently running the stock sysctl firmware. (not lpc firmware – only classic reform has an lpc chip on it, the pocket has the rp2040)
That’s awesome! I was going to try the same mod with EREMIT 1260100, I think it’s 10000 mAh each, and it’s great to know that no firmware updates are needed.
Thanks for this. I had bought some batteries with the same XY footprint as the originals but with double thickness, and was going to look into printing a “bump out” bottom plate to accommodate them. But yours is a much better solution, as it doesn’t add to the case thickness, and installation only requires removing the plastic battery brackets.
I found some nearly identical batteries on Amazon US that already had JST connectors soldered on, although I had to swap the connector polarity. (It is VERY IMPORTANT to check this if you are reading this thread and considering this mod!) Will use the other batteries I bought in other projects.
There is still some unused space on the sides of the batteries. I bet the case would accommodate 10000-12000 mAh cells, although I guess they would need to be custom made. I filled in the extra space with some anti static red foam I had saved from various packaging. Like you I used kapton tape to hold the cells in place.
Everything seems to work well. Currently the battery charge level seems to be off, I did pre-charge the batteries individually before installing them but it won’t seem to charge past 71%. Maybe it needs a few cycles to learn the correct capacity.
The specific ones I bought are now out of stock, but there seem to be several manufacturers selling there. Just search for “7565121 battery” and confirm the measurements. To switch the polarity on JST connectors, you can use a pin or other small object to gently lift the retaining tab on the connector, while carefully pulling on the wire. Eventually you will be able to pull out each wire and you can reinsert them with the correct polarity. Be careful not to pull too hard or you could pull the wire out of the crimped sleeve.
Also, after running it to flat a couple times, it seems like the battery gauge is adapting.
I went with the Liter Energy version of these 7565121 off Amazon. Looks like they are out of stock now. I swapped the pins on the JST to correct polarity. So far it is working well.
I used locktite mounting putty to hold them I hope it works.
I’ve experimented with bigger batteries for two weeks, I don’t think I like anything bigger than 8000 mAh, for following reasons:
This laptop is already heavy enough, adding an oversized battery and custom bottom plate makes it an almost 1.5 kg brick, I’m scared of dropping and breaking somebody’s foot with it.
Due to safety I can’t make the bottom plate thin, basically it’s another 10 mm of thickness for already thick boy. Maybe if I do a gasket instead of full plate it could be thinner, but I don’t have long screws to put it in place.
I think @josch found the best option for those who need capacity above all, but I’m still concerned about safety with the battery protected by one thin layer of PCB. I might look for 6000 mAh part that I could protect better, but for the time being I’ll probably stick to a powerbank, if I need to extend the battery life a bit.
Also I’ve remixed an original battery holder into something that could hold this big battery in place better than the tape, if somebody wants to go this route.
OpenSCAD code for the holder
bat_width = 9;
dist_between_holes = 115;
module holder(){
mirror([0,0,0]) union() {
difference() {
rotate([-90,0,0]) import("pref-assembly-20240112-BattHolder1-reexport.stl");
// countersinks
translate([-35.5, -8.5, -2.3])
cylinder(h = 1.2, r = 3.8/2, $fn = 20);
translate([17.7, -8.5, -2.3])
cylinder(h = 1.2, r = 3.8/2, $fn = 20);
// cut off a bit
translate([-39, -13, -4]) cube([60,2,4]);
// cut off the center part
translate([-37.7 + bat_width,-13,-4])cube([80,80,80]);
}
// add connection
translate([-37.7, -5, -3.5])cube([bat_width,dist_between_holes-17,2]);
}
}
holder();
translate([0,dist_between_holes-17,0])
mirror([0,-90,0])
holder();
I get away without any screws because x/y movement is constrained by the aluminium frame and the charging board and z movement is constrained because the bracket is squeezed between keyboard pcb and bottom lid.
The only odd thing is that the battery board does not quite sit in the center. So while one of the brackets needs to be 80.5 mm long, the one for the other side must only be 79.5 mm long.
I’ve just installed 2x 8000 mAh 7565121 batteries (I’ll print and fit joshc’s adapter tomorrow. Hele in with kapton tape currently. So far, so good. Thanks for all the information, just reporting in it’s gone well.
My 7565121s have turned up but I’m kind of on the fence about whether to install them or not.
It is a very tight fit, and I think I’d be fine with that if there were some way to easily check on the batteries without taking the back off. As it stands, I’m not totally confident in either the manufacturer or the space that the batteries have to swell when charging or that I’d notice quickly enough if they were doing the whole spicy-danger-pillow thing.
Part of me thinks a pressure sensor between the battery and the panel which can act as a killswitch would be good, but that’s not really a mission I want to go on.
