Safe PCB Temperature: How Hot Is Too Hot?

The Question Behind the Question

When an engineer asks what a safe board temperature is, the useful answers are three different numbers. There is the temperature the board may reach without damaging itself, the temperature at which the components on it still meet their specifications, and the temperature the product standard or the safety approval allows a user to touch. A design can be safe by one measure and failing by another, and the confusion usually comes from treating them as one limit.

This note separates the three and gives the practical limits for each.

The Board’s Own Limits

Glass transition temperature. The laminate changes from a rigid to a rubbery state at this temperature, its expansion rate rises sharply, and its mechanical strength falls. Standard FR-4 sits between about 130 and 140 degrees Celsius and high Tg grades between 150 and 180. Operating the board at or above the transition temperature is not immediately destructive, but the mechanical and dimensional behaviour changes and the long term reliability suffers.

Decomposition temperature. Above this the resin begins to break down chemically and the material loses mass. It is well above the glass transition, typically above 300 degrees Celsius for FR-4, and it is relevant to the soldering process rather than to normal operation.

Continuous service temperature. The temperature the laminate manufacturer allows for long term use, which is usually well below the glass transition temperature. A common working limit for standard FR-4 is around 105 to 130 degrees Celsius, and the higher Tg grades extend it.

Flammability. The UL 94V-0 rating is a property of the material under a flame test, not a service temperature limit, but the product approval usually depends on it, and the rating has to be maintained through the assembly and the coating process.

The Components’ Limits

The board is usually not the weakest part of the assembly. Every component has a maximum operating temperature, and the smallest and most efficient components have the lowest ones.

Electrolytic capacitors. The most temperature sensitive parts on a typical board. Their life roughly halves for every ten degrees of temperature rise, so their temperature determines the life of the product far more often than the laminate does.

Semiconductors. A junction temperature limit, commonly 125 or 150 degrees Celsius, which is not the case temperature but the temperature inside the die. The case temperature is typically 20 to 40 degrees lower, and that gap is what the thermal path has to achieve.

Plastic packages. Moulding compounds have their own service limits, and a part that is operated near its maximum case temperature will see its plastic discolour and its leads and joints degrade over time.

Solder joints. The joint itself tolerates the temperature, but the thermal cycling between the cold and hot states creates the fatigue that eventually cracks it. The temperature swing matters as much as the absolute value.

The Standards’ Limits

Two constraints come from outside the design. The first is the safety requirement for accessible surfaces, which limits how hot the outside of the product may become and therefore how much heat the board may transfer to the enclosure. The second is the flammability and the material certification of the board itself, which is what the UL listing depends on.

For conductor sizing, the established practice is the IPC convention: the current a trace can carry is defined for a permitted temperature rise above ambient, commonly ten degrees Celsius for conservative designs and thirty degrees for designs where the rise is acceptable. The trace width is then calculated from the copper thickness and the permitted rise, with the board area and the air movement taken into account.

thermal imaging of a powered circuit board

Practical Limits by Class of Board

  • Consumer electronics, sealed plastic housing: board ambient commonly reaches 60 to 85 degrees Celsius, and the laminate is chosen to give margin above it. The limiting part is usually an electrolytic capacitor or the battery.
  • Industrial control inside a cabinet: 70 to 95 degrees Celsius at the board, with the electronics rated to 105 degrees to leave margin for a hot day and a blocked filter.
  • Automotive under the hood: 125 degrees Celsius at the board and above, which removes nearly all standard components from the design and requires a high Tg laminate, often with a metal core section.
  • Power conversion and lighting: the board beneath a power device may exceed 100 degrees Celsius locally even when the average is far lower, so the design is judged by the hottest point rather than the average.
  • Medical and laboratory instruments: usually moderate, but the sterilization cycle imposes temperatures that the board must survive in the unpowered state.

How to Measure It

Thermocouples. Attached to the component case and to the board surface with a small amount of adhesive, they are the standard method. Their limitation is thermal mass: a thermocouple on a tiny component draws heat away and reads lower than the true temperature.

Infrared thermal imaging. Fast and non contact, and it gives a picture of the whole board in seconds. It is unreliable on shiny surfaces, where the emissivity is low and the measured temperature is wrong, so the surface is coated or a known emissivity is applied.

Thermal test coupons. A resistor or a diode that is used as a temperature sensor, placed on the board during the design, gives a reading that can be monitored in production without contact.

Simulation. A thermal model of the board and the enclosure predicts the temperatures before the hardware exists. It is worth building early and then correcting against the measurement, because the corrected model is what allows the next product to be designed without a prototype.

The measurement has to be made in the real enclosure at the real ambient, with the product at full load, and it has to run until the temperatures stop rising. On a sealed product that can take several hours, and a ten minute bench test proves nothing. Our notes on PCBA testing describe the test methods, and our notes on quality management describe how the data is recorded.

thermocouple instrumentation on a PCB under load

Designing for Margin

Decide the ambient first. The internal ambient of the enclosure is a design input, not a result. If it is unknown, assume a value and verify it.

Derate the components. A part operated at half its rated current and well below its maximum temperature lives far longer than one operated at the limit. The derating rules in the applicable standard exist for that reason.

Spread the heat in copper. The copper under and around a hot component is the first heat sink. Wider traces, larger pours and thermal vias to the other side cost nothing but board area.

Move heat to the enclosure. A thermal pad or a metal backed section transfers heat to the chassis, which is often the only large surface available in a sealed product. Our notes on PCB design and layout cover the layout methods, and our notes on PCB manufacturing describe the metal core and thermally conductive constructions that support them.

Watch the temperature swing. A joint that cycles from minus 20 to plus 100 degrees Celsius will fatigue, even if neither extreme is outside a rating. The cycle count, not the absolute temperature, is what cracks solder joints.

Warning Signs in a Finished Product

  • Discoloured laminate around a component, which indicates sustained overheating and a resin that has begun to degrade.
  • Smell of hot resin on a board that has run for some hours, which usually means the laminate is above its long term service temperature.
  • Bulging capacitors, which follow from the electrolyte boiling and indicate a thermal design that has no margin.
  • Cracked solder joints at the corners and edges of a component, which are the signature of thermal cycling fatigue.
  • Delamination or blisters under a hot device, which come from absorbed moisture turning to steam.

FAQ

What is a safe temperature for a PCB? Around 105 to 130 degrees Celsius for continuous service on standard FR-4 and higher for high Tg grades, but the components, especially electrolytic capacitors, usually set the real limit well below the laminate.

What is the maximum temperature a PCB can survive? The laminate withstands the soldering process at 260 degrees Celsius for a limited time and its decomposition temperature is above 300 degrees, but neither is a service condition.

How hot is too hot for a component? Compare the measured case temperature with the rated maximum, allow for the junction to case difference, and derate according to the applicable standard rather than operating at the limit.

How should the temperature be measured? In the finished enclosure, at full load, until the readings stabilise, using thermocouples or a thermal camera with the emissivity accounted for.

Conclusion

Safe board temperature is three numbers: what the laminate tolerates, what the components tolerate and what the product standard permits. The laminate is usually the most forgiving of the three, and the electrolytic capacitor and the semiconductor junction are usually the least. Decide the internal ambient as a design input, spread the heat in copper, move it to the enclosure where you can, measure in the real product at full load, and judge the design by the hottest point and by the temperature swing rather than by the average.

Leave A Comment