Miniature Freezer

A Project of building a working miniature fridge/freezer

Similar projects worth following
‘Miniature fridge’ focusing on Optimizing Multiple Layered Thermoelectric Modules for maximum efficiency and cooling performance.

  • Introduction

This project was to make a tiny freezer that is yet big enough to be functional. Here I would like to share my experiences in making miniature-sized (93x96x116mm) functional refrigerator, specifically optimizing variables like, different TEC size, stacking, voltage and current adjustments, minimizing heat dissipation and maximizing cooling performance simultaneously.

There are lots of methods of cooling and variety of cooling systems known and be chosen from. In order to satisfy condition that it would be compacted in a very small space (and cost effectiveness was also a factor), I decided to use thermoelectric module cooler (TEC) for my cooling apparatus. TECs are simply solid state heat pumps. They move heat from one side to the other, causing one side hot whereas the other side cold. TECs do not have a fix or rated temperature when applied power, which means that TEC modules could be stacked to achieve lower temperatures.

  • Initial Tests

The desirable cooling system should be as small as possible. For this reason, I started with the smallest heat sink yet big enough to remove heat effectively from a TEC surface, and gradually increased in size. The thermal conductivity of the thermal paste I used for the primary tests were 1.42W/m-k and for the 4th 9.2W/m-k. All the test were done in a room of 23~25 degrees C.

 1. The first test was performed with 45x60x24mm heat sink, 3030 12V 4A TEC, and 40x10 fan. I was able to achieve the maximum of -7 degrees C at 12V 2.5A. However, when more than 2.5A was applied, the cooling was rising in temperature, which was probably caused by the capacity limit of the small heat sink. To overcome the capacity limit, a cube enclosure (60x50x50mm) with 6mm thick foam was added to contain the cool air inside of the cube. But that failed to reach my targeted temperature which is below freezing. 

1-1 3030 TEC test

1-2 with chamber

 2. Second test was performed with 90x60x24mm heat sink and two 40x10 fan, which is double the size in length of the 45x60x25, and 3030 TEC 12V 2.6A. With this setting I could get almost double the temperature, that is -13C, but still that was not enough to produce cooling. I realized that signal or mono layer of TEC was not able to cool the cube enclosure below freezing temperature enough to freeze water or ice cream stored inside because of the limited sized itself.

2-1 3030 TEC test

2-2 With chamber

 3. Third test was performed with two layered TECs, which was 100x95x24 heat sink, 4040 TEC 12V 4A for the bottom layer and 3030 TEC 12V 4A on the top of the bottom layer, 92x38 fan was used.

With the setting of 4040 TEC 12V 2.6A and 3030 TEC 5V 1A, I was able to get -23C, roughly 36W for both TECs and a value of -23C. I designed a small enclosure (60x50x50mm) to measure the chamber temperature. With the same test setup mentioned above, I added a 30x30x10 heat sink to the cold side of the 3030 TEC. Despite the surface temperature of -23C, the lowest chamber temp I was able to achieve was 4 degrees lower than ambient temperature. Soon I realized it was due to the lack of circulating fan inside the chamber for churning the cooled air. After adding a 25x25x10 fan on top of the cold side heat sink, I was able to get a lowest temperature of -5C, which was much lower than my expectation.   However, other problem came up with the size of the heat sink. The heat sink of 100x95x24 that I used was too big to be fitted into the miniature-sized frame of the fridge. Therefore, I had to optimize the size of a heat sink for it to satisfy the cooling capacity by trial and error method.

3-1  4040, 3030 TEC test

3-2 With chamber

4. I got three different heat sinks 92x92x30, 77x77x30, and 85x85x44. After some tests, 77x77x30 heat sink was proved to be the exact form factor for my design, namely 77x77x30 heat sink with 80x10 fan. I did a test with the same setup as the 3rd test (4040 TEC on bottom,...

Read more »

document - 31.05 kB - 02/28/2021 at 10:26


presentation - 4.69 MB - 02/28/2021 at 04:49


presentation - 2.99 MB - 02/28/2021 at 04:49


  • Fridge assembly

    Jung Hoon Lee02/28/2021 at 10:54 0 comments

    • 3D Designing

               I started designing the heat sink, followed by other components. The design process took a while because I needed it to be very small, had to be a 3D printable, modular enough to be assembled with ease, and the most importantly had to designed for maximum performance, which in my case was minimizing any temperature rise In the cold chamber.

    • Electronics            

               I wanted a display to monitor the system information and temperature. A 0.96 inch OLED display and Arduino Nano was used. For my project, “the smaller the better” I had two sensors to choose from, which were AHT10 (digital) and KY013 (analog) sensors. I wired the AHT10 to an Arduino Nano via I2C, printing values to the serial monitor to see if I could get correct measurements. It was a perfect choice, until I realized that the AHT10 breakout module had SMD components that could mal-functioning under freezing temperature. Moreover, AHT10 was design to measure space/chamber temperature, not to measure surface temperature. The only sensor left on my list was the KY013 analog sensor. Same as before, I connected the KY013 sensor to an Arduino Nano. Using serial print, I was able to get reliable temperature reading. This temperature sensor is a NTC thermistor, which is changing its resistance value when applied different temperatures. According to the specifications of the thermistor, it has a temperature range of -55C to 125C, which is well above my temperature range. After making sure the sensor is working properly, I needed two sensors for the cold chamber and the hot side of the heat sink. At the same time the cold sensor should display the real time values using the OLED display. I’ve wired the two sensors, OLED and an Arduino Nano. After uploading the finished code to the Nano, I had a problem that the two temperature values displayed on the OLED was off by a lot. After few days of troubleshooting, I finally found out that the I2C connection with the OLED and the Arduino Nano A4 and A5 pins was causing the problem. I switched to the SPI version of the OLED, by which I could solve the problem. I wasn’t sure why the I2C connection was causing the problem, but I assumed it was due to the I2C bus on the analog pins interfering with the analog read or it could be a library issue. I later added a “warning” prompt when the hot side heat sink temperature rises above 50C.

    • TEC control

               I needed a different voltage and current to drive the two TEC modules. PWM control would be the easiest to drive the TECs but, after some searching on the web, I found out that TECs operate its max efficiency, when given constant voltage and constant current. However, I wasn’t able to find a controller that could keep output constant V, A. I could find some controllers that are specialized in controlling TECs, but the price was couple of hundred dollars. I instead thought about using an adjustable voltage and current buck converter with MCP4725 DACs to convert digital signal from an Arduino to analog voltage and feed it to the pins where the adjustable potentiometer for adjusting V, A are connected. But soon realized that I was going too far with the electronics, it just made things more complex. I decided to reserve this for my second revision of this project. Using my DMM I set both voltage and current for the two TEC modules, 4040 at 12V 2.6A, 3030 at 6V 1A. After I checked everything working properly, I started to assemble the fridge. The wires for the electronics were a challenge to fit inside the case, but other than that everything came together, with minimum design changes. I used a 12V 5A power supply for the power. Powered on, after few minutes, the cold chamber went down to -6.6C at 24C ambient temperature. To make sure it’s a “real...

    Read more »

View project log

Enjoy this project?



Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates