Just chill!

My wife and I bought a mini-fridge from Target a little over two years ago and have used it for sodas ever since. While I never did measure the actual temperature of the machine, I knew when it wasn't chilling as well as it had been. This week, I opened it up to get a look at things and see about changing the heatsink grease out.

The backstory to this is I took it apart soon after buying it to figure out how well-built it was as I've long known anything Made In China tends to be slapped together at high speed in mass quantities. I was right. The heat transfer grease used was the standard white silicone goo with zinc oxide particles in it. Not very good, really. I replaced it with some cheap copper grease which was better than the stock compound. It stayed like this for two years.

I had used the copper compound on a few computer systems and noticed that it tended to separate after a few months, screwing up the coverage. I figured I'd take care of it when I had something better.

What I ended up with was a kit from Arctic Silver that had their two cleaning fluids and ceramic-based compound. A compound rated for high-output CPU dies should be able to take on the range of temperatures from the cold and hot side of the Peltier device used as the cooler in the fridge.

The added bonus in all of that was that the compound was being put on six surfaces, making three layers of the stuff, with the central aluminum cold block being the major addition to the design.

So, on with the process.

I undid the screws on the back to access the assembly and saw there was dust on the housing that the filter screens missed. I knew it was time for a cleaning anyway. No surprise there.

What I didn't expect was what I found when I removed the chiller assembly. The cooling fins were completely caked with dirt at the inflow on one side, explaining why the unit wasn't working as well as it did when new. Since Peltier devices don't wear out and the fans were fine, it had to be something like that.

A vacuuming later, the fins, fans and casing were clean as new, or at least close to it. It was time to tear into the cooler. The assembly is better than most of the plug-in travel coolers I've cannibalized for their Peltier modules. Those usually have a Peltier module sandwiched between a heatsink and coldsink and leave it at that. The only one I didn't see with this assembly style was a cheap one I picked up for a couple of dollars once. It had a cold-block with a thermostat for the heating function bolted to it.

The gigantic heatsink of about ten inches square and three inches thick has the Peltier module resting against it. This is where the high-grade compound really shines. That's what it's made for. The Peltier has the cold-block touching the cold-side (naturally) and the other side of this block is pressed against the back of the coldsink (their term, not mine). Ther entire assembly is bolted together with styrofoam insulation surrounding the primarily cold areas and silicone caulk sealing it all up.

When I replaced the factory silicone heatsink grease with the copper compound, I broke the seal and never replaced it. Luckily, I had a few syringes of RTV sealant salvaged from the trash at work to use for that.

I used the Arcticlean products and paper towels to remove as much copper as I could from the surface to which the Ceramique compound was to be applied. I figured any that stayed around was lodged in pits in the surface and likely wouldn't budge. Applying the new compound with the edge of an old driver's license was straightforward. Some use credit cards for this part, but both work equally well. I had to use a lot more compound than what is required for a computer CPU, but that's to be expected for the surfaces I was joining.

I pressed the greased Peltier module to the freshly greased heatsink and held the Peltier in place while covering the cold side with the new compound. That done, I put down a layer of RTV sealant over the foam contact area of the heatsink, inserted the bottom-greased cold-block into the insulator and fit it to the top of the Peltier device. The plastic frame that held the insulator in place from the coldsink side needed sealing as this was where the factory seal had been placed. I ran a bead of RTV around the top of the insulator where the frame contacted and on the top of the frame where the coldsink contacted. Then, just to be sure, I ran a bead on the ouside of the frame, joining it to the coldsink. To finish it all off, I bolted the entire assembly back together and replaced the silicone caulking in the screw holes with RTV.

I buttoned it all back up and started the fridge. The temperature dropped faster than I've ever seen and the air coming off the heatsink was actually hot. I had done it. The temp stabilized at 41.2F and would not go any lower over the next two days. I noticed that the fans weren't turning as fast and figured it was the thermostat limiting me. Yep.

I looked at the circuit board for the fridge and saw that it had no controls for the thermostat. I figured the thermostat might have controls inside it, however. Yep. It had a resistor network and an 8.2K resistor in series with it all. I shorted the 8.2K and found that the resistive output of the network was 17.49K at the highest and 9.9K at the lowest versus 23K at the highest and 16.48K at the lowest before the mod. Since the top reading was just barely getting into the refrigeration range, I left it at that, figuring the bottom rating of 9.9K would never have reason to be used.

I set the fridge to work cooling down. It got to 40.9 and slowed down the cooling trend a bit. It went to 39.2 over one hour's time and is now sitting at 38.3F, two hours after that. I will check the temperature in the morning and give updates as this goes on. I plan to make the fridge chill down as far as possible for the conditions and then add a resistor to keep it that way when close to the maximum setting. I'm sure there comes a point where the temperature won't get any lower with the Peltier module powered continuously. I'll have to find where it will pulse with the best efficiency to allow a lower temperature if that's the case.

Still, I met my goal of getting the temp down into the 30s. Anything extra from here is a bonus. I'd like 35F as an ideal temp and I'm very close. I'll have to stock the fridge with gamer fuel and cold ones since the ones shall be colder than ever before.

For the record, I used a cheap bimetallic thermometer in the past to check the temperature and was able to get down to 45F. The digital wireless thermometer I used for the readings posted in this blog are pretty close, measuring within a degree of the cheap thermometer, the cheap one reading low.

The sending unit was placed on the wire shelf in the fridge so I have no benefit of cold air tending to sink. I will get a bottom reading soon as that is where the beverages will be stored.