Thermal management of electronics means keeping the electronic components inside an electronic device below the temperature their datasheets allow, in the hottest air they will actually see. The methods form a ladder: a heat sink cooled by natural convection, then fans, then liquid cooling, then refrigeration or thermoelectric coolers, with heat pipes and vapor chambers helping move heat away to wherever it gets rejected. The practical rule is to use the simplest cooling solution that fits your thermal budget. Fans cover a large part of that range. We make fans, and this guide to electronics thermal management also shows where they stop being enough.
Start Thermal Management With the Thermal Budget, Not the Cooling Method
Effective thermal management starts before you choose a heat sink, a fan or a cold plate: work out how much temperature rise you can afford. Call it the thermal budget: the temperature a component is allowed to reach, minus the temperature of the air around it on the worst day. Every part of the cooling system spends from that budget.
1. Find the right number on the datasheet. Semiconductor datasheets often list more than one temperature limit. Texas Instruments defines the absolute maximum junction temperature as “the temperature beyond which damage occurs to the device.” The recommended operating junction temperature is the one at which the device runs continuously at its designated performance over its designed lifetime. Design to the recommended figure, not the absolute maximum. Other components, such as capacitors and resistors, carry their own temperature ratings.
2. Use the air temperature around the component, not the room. The ambient term in a junction temperature estimate refers to the local air touching the component and its board. Inside an enclosure that air is usually warmer than the room, sometimes by a lot, because the electronic system inside it generates heat. An article in Electronics Cooling uses a desktop PC as the example: the ambient for a component on the board isn’t the roughly 20°C room, but a higher temperature caused by the heat inside the case. TI makes a related point: different standards place “ambient” in different spots (the air entering the box, the air under the device, the air stream ahead of the board), so settle which one you mean before any system-level analysis.
3. Subtract. Target temperature minus worst-case local air temperature is the whole budget. The heat sink, the thermal interface material and the airflow each take a share of it.
As an illustration only: if a part’s recommended operating limit were 105°C and the air around it could reach 55°C on a hot day, the cooling path from that part to the air would have 50°C to work with. Your parts will have their own figures.
Two cautions before moving on:
- Don’t estimate junction temperature from θJA alone. The common shortcut adds room temperature to power multiplied by θJA, the junction-to-air thermal resistance from the datasheet. But θJA is measured on a standard JEDEC test board in a still-air box, and PCB design, pad size and altitude all change it. TI’s guidance is that it’s best used to compare packages. The Electronics Cooling authors, mechanical and thermal engineers at an aerospace company, recommend a detailed thermal simulation, a conduction or computational fluid dynamics (CFD) model, instead, or the junction-to-board resistance when the board temperature is known.
- Leave margin early in the design process. That’s what derating is. A UK defence reliability manual (GR-77, 2012 issue) defines it as “a policy of deliberately under stressing components in order to provide increased reliability.” The same manual notes that failure rates typically change by a factor of 1.1 to 2.0 for a 15°C change in temperature, depending on the component type.
If your thermal management question is about a whole control cabinet rather than one component, see why PLC cabinets overheat. Component derating is also different from the correction a fan itself needs at high temperature or altitude, which is covered in fan performance at altitude and temperature.
The Cooling Ladder: What Each Method Does and Where It Runs Out
Pick the lowest rung that keeps your parts inside the thermal budget. The rungs are ordered by where the heat finally goes. Heat leaves a component by conduction into whatever it touches, then gets handed to a fluid, usually air and sometimes liquid, so conduction and convection do the work between them. That final handoff is where most of the difference in capacity comes from. A university heat-transfer course gives these typical ranges for the convection heat transfer coefficient: 1–25 W/m²·°C for natural convection in gases, 25–250 for forced convection in gases, 50–20,000 for forced convection in liquids, and 1,000–100,000 for boiling and condensation. They’re ranges, not design values, and they overlap; what changes most is the upper end, from 25 for still air to 250 with a fan and 20,000 with pumped liquid.
| Method | What it does | Typical h (W/m²·°C) | Where it runs out | What it costs you | Go deeper |
|---|---|---|---|---|---|
| Heat sink + natural convection | Conducts heat from the heat source into fins made of high thermal conductivity metal; air moves by buoyancy alone | 1–25 | When the fin area needed no longer fits, or the local air is too warm | Size and weight; nothing to wear out | PCB cooling fan selection |
| Forced air (fans) | Pushes air across fins and components, replacing warm air with cooler air so the fins dissipate heat faster | 25–250 | It can bring parts toward the temperature of the incoming air, never below it | Noise, dust, a moving part with a finite life, some power | Sections below |
| Liquid cooling (cold plates) | Coolant carries heat from a plate on the device to a heat exchanger elsewhere; liquids’ higher thermal conductivity and heat capacity raise the coefficient | 50–20,000 | Coolant can’t run below the dew point without condensation risk, and the heat still has to be rejected somewhere | Pumps, plumbing, more design work | Server rack cooling |
| Thermoelectric (Peltier) coolers and refrigeration | Pump heat, and can cool below ambient | Not a convection method | The hot side has to reject the heat removed plus the cooler’s own input power | Extra power consumption; a larger heat sink on the hot side | Enclosure air conditioners vs fans |
In the first row, the contact between part and heat sink is easy to overlook. The surfaces of a component and a heat sink may look flat, but tool marks and tiny imperfections keep them from touching fully, and the gaps fill with air, which has low thermal conductivity. A thermal interface material, such as a grease or thermal pad, fills those microscopic gaps with a material whose thermal properties beat air. The heat sink also only helps if it can hand its heat to air that isn’t already hot.
Heat pipes and vapor chambers aren’t a rung above fans. They move heat rather than reject it. A heat pipe uses an evaporating and condensing working fluid to carry heat from the source to a heat sink somewhere else, giving it a far higher thermal conductivity, in effect, than solid copper; notebook CPUs, with no room for a large heat sink on the chip, became an early high-volume use. A vapor chamber spreads heat sideways from a concentrated source. In a webinar published by Electronics Cooling, the presenter’s guidance was to use them when the design is conduction-limited, and reach for a vapor chamber when height is limited, power density is high or airflow is low. In most electronics, the end of either one is still a set of fins and some air. A 1996 Electronics Cooling article on heat pipe heat sinks says their resistance is “predominately controlled by the available fin volume and air flow.” In thermal management systems they pair with natural convection or with fans. They don’t replace the air side.
Do Your Electronic Components Need a Fan?
You need a fan when a heat sink cooled only by natural convection can’t keep your electronic components inside the thermal budget at the worst-case air temperature. Work through it in this order:
- Heat load. The total, and which components run hottest. A single hot part can decide the whole product design.
- Worst-case air around those components. Use step 2 of the thermal budget, not the room temperature.
- Budget check. Can passive solutions, such as a heat sink alone, dissipate that amount of heat within the budget, in the space you have? If yes, you may not need a fan.
- Constraints. Noise limits, dust, a sealed enclosure, years without maintenance, vibration and strict power requirements can rule out a fan even when the heat load says you need one. Then look at passive systems, such as conduction to the enclosure wall or heat pipes, or at sealed thermal solutions.
- Your equipment. Start with the scenario guide that matches (listed below); the details still vary case by case.
We deliberately don’t give a wattage cutoff here. The rules of thumb you’ll find online for passive versus active solutions disagree with each other, and none of them knows your enclosure or your ambient.
For guidance on specific electronic equipment, see micro fans for small devices, power electronics in inverter cabinets, EV charger cooling, and our electrical enclosure cooling guide.
How Forced Air Cooling Does Its Job
A fan carries excess heat away by replacing the warm air next to hot surfaces with cooler air. That lifts the heat transfer coefficient into the forced-convection range, roughly an order of magnitude above natural convection, and it lowers the local air temperature from step 2 of the budget. A heat sink spreads heat across a larger area; the fan helps transfer thermal energy away from it.
Whether a particular fan improves thermal performance depends on three things.
Enough airflow for the heat and the temperature rise you allow. Start from the temperature you need and work back to the airflow. An equipment builder in South America asked us for a fan that would bring their control enclosure down by about 5°C. They asked for a result, not a model. When we ran their heat load against that target, the fan class they had shortlisted came up short, so we suggested the next size up in the same mounting family. The calculation itself is in our CFM calculation guide.
Enough pressure for the resistance in the path. Filters, grilles, fin stacks and tight board spacing all push back on the airflow, so a fan delivers less installed than its free-air rating. Reading the pressure side is covered in fan static pressure explained and how to read a fan curve. Whether an axial fan or a blower suits the job is compared in axial fan vs blower. For compact electronics, the usual candidates are DC brushless fans for straight-through airflow and centrifugal blowers where the path is tight or needs more pressure.
Air that actually reaches the heat. Intake and exhaust shouldn’t short-circuit, and air should pass the heat sources on its way out. Filters keep dust out but add resistance, and whether the fan pushes air in or pulls it out changes where the dust collects. See enclosure fan and filter placement.
Fans don’t have to run flat out, either. Speed control lets a fan run slower at light load and speed up as temperature rises; the options are compared in PWM vs voltage fan control.
Where Fans Run Out: Four Signals
A fan can only push the air around a component toward the temperature of the air coming in; beyond that point, thermal management needs another rung. Any of these four signals means it’s time to look at the next rung.
- The incoming air is already at or near your target temperature. A fan can remove the extra rise that builds up inside an enclosure, but it can’t bring a component below the temperature of the incoming air. When the gap between inlet air and your target is nearly gone, extra airflow buys very little.
- You need components below ambient. That takes something that actively pumps heat, such as refrigeration or a thermoelectric cooler. A fan can’t cool below ambient, as our comparison of enclosure air conditioners and fans explains.
- The enclosure has to be sealed. With no air exchange, an open-loop fan has nothing to work with. That’s where heat exchangers and enclosure air conditioners come in (same guide).
- The environment is too hot for the fan itself. A fan is a component with limits of its own. Fan datasheets list an operating temperature range; check it against the worst-case thermal conditions the fan will see. For how temperature affects fan bearings and life, see sleeve vs ball bearing fans and high-temperature axial fans.
High power density, with heat concentrated in a very small area, is a related limit. Even good airflow can’t carry heat out of a hot spot that the heat sink can’t spread. Vapor chambers spread that kind of hot spot; when the heat still can’t be rejected to air within the budget, liquid cooling with cold plates is the next rung. For rack-level liquid cooling, see our server rack cooling guide.
A Case Where the Fan Was the Limit
In 2022, a European manufacturer of UPS equipment asked us for intake and exhaust fans for cabinets that run hot and run continuously. The requirements: a maximum working temperature of 70°C, L10 bearing life specified at both 50°C and 70°C, dielectric strength testing, and more airflow than our catalog fan delivered.
Their own measurements on samples didn’t match our published performance curve, and we acknowledged that directly rather than defending the catalog number. We then tried a new impeller material to push speed and airflow higher. The redesigned samples still fell short of the airflow target. We told the customer so, and that we were still looking for a better approach, rather than claiming a result we hadn’t reached.
A requirement like 70°C also puts the fan’s own ratings in play, which is signal four above.
If your thermal design is close to any of these edges, send us the numbers. We’ll tell you whether a fan will hold, and if it won’t, which direction to look.
Frequently Asked Questions
What temperature is too hot for electronics?
There isn’t one number for every electronic device. Each component’s datasheet sets its own limits, and the one to design to is the recommended operating temperature. Ambient ratings vary too. Texas Instruments, for its own industrial-grade parts, specifies operation in ambient air between –40°C and 85°C, and adds that system-level thermal design is still needed so the devices’ maximum operating temperatures aren’t exceeded. TI also notes that no industry standard defines what “industrial temperature” means.
How do you calculate junction temperature?
The common hand formula adds ambient temperature to power multiplied by θJA. It’s often misleading, because θJA depends on the test board and environment it was measured in, not just the package. TI’s application note gives better options when you can measure: junction temperature from the case-top temperature plus power multiplied by ΨJT, or from the board temperature plus power multiplied by ΨJB. TI adds that ΨJT shouldn’t be used when a heat sink is fitted. For the final estimate, the Electronics Cooling authors recommend a detailed conduction or CFD model instead.
What does temperature derating mean?
Derating means running a component deliberately below its rated stress so it lasts longer: lower power, voltage or temperature than its maximum rating. For semiconductors it’s generally power that gets derated, to control junction temperature. In electronic thermal management it’s the component’s margin; this is different from derating a fan’s performance at high temperature or altitude, and from the conductor derating tables in electrical wiring codes.
What are the downsides of thermoelectric cooling?
A thermoelectric cooler moves heat rather than removing it, and it adds its own input power to the load. The hot side has to reject the heat pumped from the cold side plus the electrical power going in, so it needs a heat sink, often with a fan, sized for more heat dissipation than the original load. The cold side can also pick up condensation in humid conditions, which proper sealing around the module can alleviate.
What are the downsides of liquid cooling?
The coolant can’t run below the dew point of the surrounding air without risking condensation on parts in contact with the electronics, so the dew point sets a practical lower limit on coolant temperature. The heat also still has to go somewhere: a cold plate moves it to a heat exchanger that rejects it to air or to facility water, for example through the dry coolers used in data centers. And pumps and plumbing add parts to design and maintain.
How cold is too cold for electronics?
Datasheets list a minimum operating temperature as well as a maximum. TI’s industrial-grade parts, for example, have a minimum operating temperature of –40°C. Fans have a minimum operating temperature on their datasheets too, so check both the electronics and the fan against the coldest start-up condition.
Send Us Your Numbers
If a fan is part of the thermal management solution for your electronic device or enclosure, send us the heat load, your target temperature or allowable rise, the worst-case ambient, the space available and the supply voltage. Our engineers will check the fan selection against those numbers. We reply within 1 business day.
Start a fan selection inquiry or email andy@proventofan.com.
If you’re already comparing DC fans, our DC cooling fan selection guide covers the next step.
References
- Texas Instruments. Semiconductor and IC Package Thermal Metrics, Application Note SPRA953D, 2024.
- Montanez, A. L., and Lozano, A. “Thermal Facts and Fairy Tales: Misusing Theta JA to Calculate Junction Temperature.” Electronics Cooling, June 2026.
- Purdue University. ME 418 Heat Transfer lecture notes.
- Applied R&M Manual for Defence Systems (GR-77, 2012 issue), Part C, Chapter 7: Derating. Hosted by the Safety and Reliability Society.
- Godfrey, S. “An Introduction to Thermoelectric Coolers.” Electronics Cooling, September 1996.
- Garner, S. D. “Heat Pipes for Electronics Cooling Applications.” Electronics Cooling, September 1996.
- Meyer, G. “Heat Pipes and Vapor Chambers – Useful Guidelines for Heat Sink Implementation.” Electronics Cooling, June 2021.
- Simons, R. E. “Estimating Dew Point Temperature for Water Cooling Applications.” Electronics Cooling, May 2008.
- Laird. Thermal Interface Materials.
- Ohmite. “The Role of Thermal Interface Materials and How They Affect Performance,” 2024.
