Opencooler

A compact water chiller/heater that can function as a drop-in alternative to cooling mattresses like 8Sleep + other purposes (lab).

Description

This is an ongoing project to investigate how to keep water at an exact temperature in the most efficient (and quiet way), and also cost effective way, AND in a way that anyone can buy off-the-shelf components and build it entirely from scratch themselves. It should be smart, connected, and allow for build-out of features like dynamic schedules, both heating and cooling to precise amounts, and user metric tracking. It should be functional, and work vertically and horizontally and allow placement in a stacked configuration, or for example, beneath the bed. It needs to be able to be loaded with coolant easily, and be easy to quickly swap and exchange components for upgradeability and repairability. Also I want it to look cool, as this entire thing is also a practice in product design. Ideally all of this would balance out to being affordable as well.

Details

Inspiration for the project

I've had a lot of trouble sleeping through all of my life that I can remember. I'm at the point where I will do almost anything in order to feel well rested and to regain some of the energy I've lost from my insomnia. I think a lot of other people are in the same boat, but after some pretty extensive research, I came to the conclusion that there really isn't much tech out there for improving sleep, short of CPAP machines for sleep apnea, something that I'd ruled out as an issue (but might be your problem! You don't need to be overweight to suffer from sleep apnea, your quality of life may be drastically improved from CPAP).

I saw an ad.

One of the first things you will see when checking YouTube for "how to improve sleep" is Bryan Johnson's psychotic approach to sleep optimization (a millionaire who is dead set on becoming immortal). In the video, he is using a water cooled mattress topper by 8Sleep that records his "sleep score", a number that the bed cover, or the strip of sensors beneath it is recording and uploading to the cloud. This looks really interesting, until you realize is costs thousands of dollars AND a mandatory monthly subscription. There are a few others, but they don't have the same amount of features as the 8Sleep does, they all have issues or questionable practices, and none of them are open source.

Being me, I want to optimize my sleep, but I am also poor. Even if I could afford to buy the bed, I couldn't afford to pay the subscription, and frankly, even if I could, I wouldn't want to give a cent to such a predatory company. One thing they've done very well with though, is advertising. It seems like nearly every health influencer who is in some way interested in sleep tracking has received one of these beds for free, and swears by it. Now I can go into almost any makerspace and mention water cooled beds and SOMEONE in the room will immediately know what I'm talking about, usually followed by "aren't those super expensive?"

"NO!"

(if you build it yourself)

Let's tear it down

I started on this project about two years ago, where I asked on the r/8sleep subreddit for a teardown of the chiller unit the "pod". I did a quick breakdown on the github project page about the actual cost of the unit after finding all of the parts on Aliexpress and Alibaba with the help of a fellow redditor. The total cost for this $1000+ piece of equipment? About $300. And that's after accounting for things like shipping and adding a bit of headroom.

Nothing fancy inside

The Thermal Electric Coolers (TECs) they use for heating and cooling cost only around $4 each. While being very inefficient, they are more than sufficient for cooling a human, and are the quietest option. (I would still like to test a phase change system in the future, but those do not have the advantage of being solid state.) TECs I also learned can heat and cool on the same side, just by switching polarity. I immediately realized this would be important to keep the temperature at a precise level, so polarity switching would be required in any build.

It's feasible! Let's go! Oh, will it shock me?

So now I know it is feasible to build. Preface: going into this project, I knew almost nothing about electricity and electronics, but it still looked like a it was a relatively simple device, and I'm sure I can use a pi or arduino in place of their custom arm board. Looking at reddit, there quite a few others who thought the same, and some of them had even mentioned doing it, but nobody had shared how.

The prototype

At first I looked at relays as a way to implement the cooling fast, so I ordered some cheap aluminum waterblocks (40x80mm) off Aliexpress, and some thermal compound, and slapped some 12703 TECs and the cheapest CPU cooler ($20) I could find on top of it. I ordered a benchtop power supply connected everything together with some silicone tubing...

Files

shell-prototype.png

Portable Network Graphics (PNG) - 238.02 kB - 03/07/2026 at 14:11

Preview

Project Logs

Collapse

Final testing for V1

Pat • 6 hours ago • 0 comments

I am moving countries again. This is a final test assembly before I toss the bulky stuff and pack everything I own into a single suitcase. The test did not go as I was hoping, and there are still a lot of things to refine.

The chinese splitters I was using I believe had a short, which causes some magic smoke with the controllers. I did get some useful results though, I'll do a quick summary below.

While I thought at 50% power draw the TEC 12715's would be sufficient for cooling, I feel one side getting warm to the touch and the other is not noticeably chilling the water at all. I will have to do another test one at a time to make sure it was not the short causing this. Otherwise, I will switch back to the 12706's if the 12715's need considerably more power to effectively cool than I thought. The capacitors of the MD13s cytron boards was a bit hotter than I'd like. Increasing the power draw caused the splitter to melt and short I think.
The nice pump I ordered has to be in a specific sideways orientation or gravity fed by the reservoir to work effectively. The 3D printed reservoir had a leak with the front plate. It needs to be a more solid material or printed as one solid piece to avoid leaks. I made another reservoir with a screw top printed as one piece, and it did not leak badly. Horizontal placement of components makes this complicated.
I think it is not worth it to try and continue to iterate on the flat design. The tower is better physics-wise and will let me gravity feed the pump, which is far simpler user experience.
It sucks to route the wires
PSU above the water was important (I got shocked anyways)
Screw top does leak a bit, the 3D print design needs to be better optimized so you don't have to screw it so tight to get a pressure seal.
The 3d printed connector on the back leaked way too much, it needs to be coated in epoxy.
While the black hose looks cool, you aren't able to see the flowing water for testing. This messed me up a bit.
I think the mounting bracket for the heatsinks will need to be redone to better isolate the copper heat block. Even though it looks so sexy, I think I will have to kill it.

I should probably stop trying to build the entire thing around this heatsink design....

Switched from aluminum to copper because you shouldn't mix those I guess. The fittings are coated copper or brass.
Some issues with the pump speed not changing as expected. Runs at 100% which is too loud. I need to figure out how to run it lower. Circulation speed was pretty good.

Quick connectors worked well! Screen did not. Power switch works after switching the breaker x1. Black segment was an extension because it had to be longer to fit the separate pump because of the leaking reservoir from the last post. Black and white looks sick though, so happy accident.

So I think in conclusion, 8Sleep pod is pretty close to the optimal layout already. There's a reason they aren't laid horizontally. I think the V2 will be a cuboid. See you in a few months when I have (hopefully) a new workshop. I do not know where I am going yet, just that I am leaving in three weeks. Might be longer.

Finalizing reservoir design

Pat • 04/14/2026 at 21:05 • 0 comments

Design on the reservoir and pump placement continue. Brain not work

First foray into machining parts. I plan to use plastic for this reservoir, still exploring limitations on thickness, etc. Once dimensions are dialed in on the 3d printer I'll see how much it would cost.

Attempting to find the ideal orientation for the tank and the pump. I've invested in a popular PC water cooling pump, the D5 for testing. It's around $80, but lasts significantly longer than the Chinese pumps. (Rated 5 years instead of ~2). It is also much easier to swap and source for the average person.

Discussions