Via Temperature
For many years, the current carrying capacity of vias in printed circuit boards has rarely been discussed. I believe this is due to the lack of practical methods for measuring or predicting the temperature of circuit vias. People generally assume that the temperature of a via is determined by the current flowing through it, and then base their circuit designs on this principle. Circuit board designers typically employ the following three strategies to address via dimensions:
- For simplicity, avoid letting large currents flow through circuit vias; route high-current traces entirely on a single layer;
- Design via dimensions according to the IPC-2152 standard;
- Use some standard via current values and calculate the minimum number of vias needed to carry the total current based on the overall current in amperes.
The cross-sectional area of the copper plating on a via should be larger than the cross-sectional area of the copper trace it connects to. If the cross-sectional area of a single via is smaller than the connected trace, multiple vias are required to handle the same current flowing through the trace.
Simulation Results
Thermal Simulation Conditions: Consider the heat dissipation model of the circuit board shown below. There are a pair of traces on both the top and bottom sides of the board.
● Trace dimensions:
Length (L): 76mm (3.0 inches)
Width (W): 0.69mm (27 mils)
Thickness (T): 1.5 Oz (52.5 micrometers)
This pair of traces is connected through a via in the middle.
● Via dimensions:
Diameter (D): 0.26mm (10 mils)
Thickness (C): 1.0 Oz (30 micrometers)
The cross-sectional area of the via’s copper plating is roughly the same as that of the trace. The width of the circuit board is a few millimeters larger than the width of the trace. The insulation board thickness between the top and bottom layers of the circuit board is 1.6mm (63 mils). A thermocouple model is placed at the midpoint of the trace. The specific reason for placing the thermocouple at the midpoint of the trace will be explained later.
Simulation Results
If a current of 4.75A is applied to the circuit above, according to the heat dissipation model, the temperature of the trace without a via is 72.8°C. If a via is added, the temperature of the trace remains 72.8°C, but the internal temperature of the via is 70.1°C.
At this point, you may have the following questions:
- Why is the temperature of the circuit via lower than the temperature of the trace?
- If a current of 6.65A is applied and the temperature at the center of the trace, measured with a thermocouple model, reaches 114.2°C, is the temperature in the middle of the via higher, the same, or lower than the trace temperature?
- If we increase the width of the trace to 5mm (200 mils), keep the via dimensions unchanged, and apply a larger current, such as 8.55A, the TRM simulation result shows that the temperature in the middle of the trace is 44.8°C. In that case, what is the temperature of the via? (Note: An 8.55A current would melt a 27 mil-wide trace within one second).
The simulation model shows that, with the same copper layer cross-sectional area, the temperature of the via is lower than that of the trace. The surprising reason for this is also given in IPC2152: the temperature of inner layer traces in the circuit board is lower than that of the traces on the top and bottom layers directly exposed to the air! This is because the thermal conductivity of the circuit board is better than that of air.
As a result, the heat dissipation conditions of the via are better than those of the trace, cooling more quickly and maintaining a lower temperature. The illustration below shows the temperature distribution of the circuit board after simulation. From the solder joint to the via, the highest temperature on the trace occurs at the midpoint. This is why the thermocouple measurement point is placed at the center of the trace.
If the current is increased to 6.65A, the trace temperature rises to 114.2°C, and the temperature of the via remains lower than the temperature of the trace. Simulation results show that the temperature of the via is 108.2°C. From this, we can draw the following conclusions:
When the cross-sectional area of the via’s copper plating is the same as the connected trace’s copper area, and the same current is applied, the temperature of the via is generally lower than the temperature of the trace.
What if the trace width is increased and the current is increased, but the via dimensions remain unchanged? The increased current will, of course, cause the temperature of the via to rise. However, the temperature of the via does not increase significantly more than the temperature of the trace. Why is this?
This is because the length of the via is shorter compared to the width of the trace, so the trace actually acts as a heat sink for the via. Using temperature simulation software, we can find that the temperature of the trace at this point is 44.8°C, while the temperature of the via is 48.1°C, only 3.3°C higher. Although the temperature of the via is higher than the trace temperature, it is not significantly higher. Especially with a current of 8.55A, it is sufficient to melt a trace with the same cross-sectional area as the via.
Conclusion
From the simulation results discussed above, the traditional understanding is incorrect! Current does not determine the temperature of vias but determines the temperature of traces. As long as the trace width is suitable for the current flowing through it, a standard via can handle large currents between the top and bottom layers of the circuit board. The main reasons include:
- The heat dissipation conditions of vias are better than those of surface traces on the circuit board;
- Wide traces can provide additional heat dissipation for vias;
Therefore, in general, the temperature of vias does not significantly exceed that of traces.
Of course, this does not mean we can ignore more cautious design approaches for vias. However, it does indeed show that vias do not need to have the same cross-sectional area as traces, as we might have imagined.
Experimental Verification
Every conclusion needs to withstand scrutiny, right? So here it comes. Hohannes and I knew that we needed to build a real circuit board to verify the conclusions above. Prototron, a circuit board manufacturer, generously provided us with test circuit boards.
Earlier, we tested the temperature of two different trace widths under different currents using simulation software. The data from the simulations are provided in the table below.
In the experimental circuit, we used the same currents to measure different traces. The temperature measurements were made using precise thermocouples. The specific values can be seen in the table below:
Upon examining the actual experimental data, we can draw two particularly significant conclusions:
1. Firstly, when a current of 6.6A flows through a 27mil wide trace, it causes the via temperature to reach 109℃. However, when a current of 8.6A flows through a 200mil wide trace, the same via produces a temperature of 44.5℃. This fully demonstrates that the current only determines the trace temperature, while the via temperature is determined by the trace temperature.
2. Secondly, the measured temperature is very close to the simulated temperature in terms of numerical values. This gives us sufficient reason to believe that thermal simulation is a viable method for predicting temperatures in complex environments.
