The projects we like

Home
BLOG
POST

June 4, 2025

by Simone Renzi / June 4, 2025

This post is also available in: Italiano (Italian)

For one of our clients, we are developing a device to be installed inside cold rooms and refrigerated trucks, with the task of constantly monitoring temperature and humidity to ensure that the cold chain is maintained.

A project that, on paper, might seem quite simple: a microcontroller, a sensor to detect temperature and humidity, is a Python script that sends data to the cloud at regular intervals.
Perfect no? Not really! If we implemented it that way, it would be very short-lived. As they say. It would last “from Christmas to Boxing Day.”

Why? The reason is quickly stated: electronics and humidity have never gotten along well, and in industrial cold rooms temperatures can easily drop below -20°C. Quite a chill in short…

Table of Contents

First-impact solutions?

The first concrete action was to search the market for hardware components designed specifically for hostile environments, capable of operating over a particularly wide thermal range.

One of the most interesting options is the Compute Module 4 Extended Temperature version, recently introduced by the Raspberry Foundation. This module is certified to operate in a range from -40°C up to +85°C.
When the upper threshold is exceeded, an automatic clock reduction mechanism kicks in, which is useful for containing the thermal rise and keeping the system stable, but sacrificing processing speed.

Problem already solved then?

In part, yes, we have addressed and solved the issue of temperature, but that of humidity still remains open. To protect electronics effectively from condensation, seepage, and external agents, one more measure is needed.

The solution adopted is to immerse the entire circuitry in an electrical insulating gel, more specifically a low-viscosity two-component polyurethane gel designed to seal and protect junction boxes, connectors and PCBs from water, moisture, dust and other critical environmental factors.

At first glance, it seems like an ultimate solution: just buy the right two-component, apply it, and the moisture problem would seem to be solved.
But even here, the reality is quite different…

Solved one problem, another problem arises again

The moment we solved the problem of humidity, the problem of thermal management came up again in a new guise.

That’s right, because the insulating gel, in addition to shielding from moisture, is also an excellent thermal insulator. This means that once the electronics are immersed in the gel, heat exchange with the outside becomes extremely inefficient.

Even if the device is in a -20°C environment, the CPU temperature still tends to rise rapidly because the heat generated cannot dissipate effectively.

To prevent the system from overheating and thermal throttling, an additional engineering element must be introduced.

The passive heat sink

To effectively solve the problem of thermal dissipation within the insulating gel, it is necessary to integrate a specially designed passive heat sink.

This heatsink should be mounted directly in contact with the SoC, coupled by thermal conductive paste to ensure optimal heat transfer; partially immersed in the gel, but with a surface in direct contact with the external environment, so as to promote heat transfer to the cold air of the cold room.

This solution allows us to transfer the heat generated by the SoC outward, bypassing the insulating effect of the gel. But before proceeding, it is essential to ask ourselves some key engineering questions, which are essential to properly size the system:

At what temperature do we wish to maintain the SoC during operation? The goal is to avoid both condensation (which occurs at too low a temperature) and exceeding the critical threshold of 85°C, which would activate thermal throttling.
What should be the contact surface area between heatsink and CPU to ensure effective heat transfer without dropping too much in temperature? This value depends on the thermal power generated, the material used for the heatsink, and the temperature difference between the SoC and the external environment.
How much power does the SoC of a Compute Module 4 actually dissipate under full load? Based on our tests, the thermal dissipation at full load is about 6 watts. This figure is critical for correctly calculating the heatsink area needed to keep the system within the desired thermal limits.

Only after accurately answering these questions can we move on to the next step: the actual thermal calculation, which will allow us to correctly size the dissipating interface and ensure the stability of the device even under extreme environmental conditions.

Sizing of the surface in contact with the SoC

Although the Compute Module 4 Extended Temperature (CM4ET) is certified to operate in a range as low as -40 °C (-40 °F), it is advisable to prevent the die from operating at excessively low temperatures during full load to avert the risk of condensation forming in the immediate vicinity of the electronics.

When the CM4ET is under full load, it dissipates approximately 6 watts of thermal power. This heat must be transferred from the SoC die to the external environment, namely the cold air in the cold room.

But beware: balance is also needed here.

If the heat sink is too efficient, there is a risk of the die dropping below -20 °C, which should be avoided to prevent localized condensation phenomena.
On the other hand, if the heatsink is insufficient, that is, the heat produced by the CPU is more than the heatsink can dissipate, the temperature will rise gradually until it reaches 85 °C, the threshold above which thermal throttling comes into play.

Therefore, the goal is to keep the SoC in a safe and stable thermal range. A target temperature of about -5 °C is reasonable, considering that the outdoor environment can be as low as -22 °C.

To achieve this balance, we must correctly size the contact surface of the heat sink. The analysis is based on stationary heat conduction, modeled by the following equation:

$Q = \frac{k \cdot A \cdot \Delta T}{L}$

Where:

$Q$ is the heat output transferred (in watts),
$k$ is the thermal conductivity of the material (in W/m-K),
$A$ is the cross-sectional area (in m²),
$L$ is the thickness of the material (in m),
\Delta $T$ is the temperature difference between the two sides (in K).

Solving the equation for area A:

$A = \frac{P \cdot L}{k \cdot \Delta T}$

This formula allows us to calculate theminimum area required so that, when dissipating power $P$ , the temperature difference between the die and the environment is equal to \Delta $T.$

Entering known values:

$P = 6\,W$
$L = 0{,}00675\,m$
$k = 205\,W/m-K$ (conductivity of pure aluminum)
\Delta $T = 17\,K$ (i.e., -5 °C to -22 °C)

we get:

$A = \frac{6 \cdot 0{,}00675}{205 \cdot 17} = 1{,}162 \cdot 10^{-5}\cdot{m}^2 = \mathbf{11{,}62,mm^2}$

This is the theoretical minimum surface area of the heatsink in direct contact with the SoC required to keep the temperature within the desired limits.

Of course, this value is ideal and calculated under perfectly static conditions; in practice, it is advisable to consider a margin of safety, both due to mechanical tolerances and the effect of intermediate layers (such as insulating gel or plastic interfaces).

Realization of the 3D model of the heatsink

Due to contractual constraints with the client, it is not possible to show pictures or diagrams of the project. However, I can provide some useful information on the final sizing.

At the end of the modeling and manufacturing process, the useful contact area of the heatsink was about 13 mm². A value slightly higher than the calculated theoretical minimum, docuted to mechanical limitations related to CNC fabrication.

However, this area is within an acceptable margin, sufficient to prevent the temperature of the SoC from falling too far below the desired values, thus preventing condensation.

To further optimize heat transfer, the heatsink was placed at the bottom of the enclosure, with most of its surface area below the plastic of the enclosure. In practice, only a small part is in direct contact with the cold air of the cold room being separated from it by a 2-mm layer of plastic.

It should always be remembered then that there is a substantial difference between theoretical calculations and actual behavior in operation: the former serve to give a solid design basis, but it is only with direct experimentation that the actual reliability of the solution is verified. In our case, within the expected margins, the design proved effective.

Test

To validate the design, I performed a home experimental test, placing the entire device inside the home freezer, set to the lowest available temperature. Unfortunately, it was not possible to reach the -22°C of the industrial cells but only -17°C, but the test still provided very useful data to evaluate the thermal behavior under real conditions.

The text was structured over a duration of 60 minutes, with sampling every second and after idle stabilization.
During the test we monitored:

The temperature of the SoC
The temperature inside the enclosure via an I2C sensor.
The temperature on the external surface of the heatsink, measured by a thermocouple placed between the heatsink and the enclosure plastic.

After about 30 minutes, all temperatures had already stabilized.

During idle, the maximum CPU temperature reached 12.5°C. At this point, once stabilization was noted, I started the stress test and ran the stopwatch for the next 60 minutes.

At the end of the test under load, the measured values were as follows:

Maximum CPU temperature: 38.7°C
Temperature inside the enclosure: -1.3°C
Temperature on the external surface of the exchanger: -11.4°C

Conclusion

The data obtained confirm that the CPU operates within a safe thermal range, the operating temperature does not fall below critical levels for condensation, the electronics are protected even under extreme environmental conditions, and the system proves stable and balanced with effective passive thermal dissipation management. Thus, the chosen configuration is suitable for continuous operation, even in harsh refrigerated environments, without risk of failure, condensation, or thermal throttling.

These are the projects that we are really passionate about. The ones that force you to think, to challenge yourself, to look for solutions outside the box. Projects that can’t be solved with an imported library or a standard sensor, but require ingenuity, experimentation, and multidisciplinarity spanning electronics, physics, thermodynamics, and mechanical design.

It is precisely these challenges that leave you, each time, with a richer technical and cultural background that ends up coming in handy in the most unexpected contexts.
And in the end, when everything works as it should, the satisfaction is double!

Sei interessato ai nostri servizi di consulenza?

RENOR & Partners