Frothingham Electronics Corporation Thermal Articles Part 2

Testing Thermal Response of Diodes, Chapter 2

THERMAL RESPONSE is a non-destructive electrical test for evaluating the quality of the bond between a diode junction and the package in which it is mounted. It is most commonly used to check the solder bond between the semiconductor "chip" and the package. Solder voids in this bond cause poor thermal conductivity leading to failure of the device due to over-heating of the junction. A moderate amount of voids has little or no effect on the electrical properties of the part, so attempts to detect them by changes in voltage drop generally fail.

The THERMAL RESPONSE test measures the increase in junction temperature due to a "heating pulse" of specified current and duration. The current and pulse width are chosen to heat the junction moderately and to allow time for heat to travel through the bond but not much further. A pulse that is too long will allow heat to travel into the leads or mounting surface. This results in a test that is sensitive to the thermal connection between diode and its test socket or heat sink. Conversely, a pulse that is too short will not allow heat to travel through the bond and so the result will not be effected by the quality of the bond.

Typical pulse widths are in the range of 10 to 50 milliseconds, although shorter or longer pulses may be used for some purposes. Once the pulse width is chosen you must choose an appropriate heating current (IH). The value is not critical but you must avoid two extremes.

At the low extreme the heating of the junction is so low that the DVF is very small and difficult to measure. At the high extreme the device is heated so much that it may be damaged or the K factor may change significantly making the results unreliable. A good choice would be a current that causes a rise in junction temperature of 40 to 50 degrees Celsius. This causes the DVF to be typically in the 70 to 100MV range (for a single Silicon Diffused junction) which is easy to measure and a junction temperature of 75 degrees will not damage the device or cause a significant change in K factor.

The usual first question asked by a new user is "how do you measure the junction temperature". The most common method is to make use of the known relationship between junction temperature and VF (Forward Voltage) at a low current.

The current is usually in the 1 to 10mA range and is referred to as "IM". This relationship is referred to as K or 1/K. K is the ratio of the change in temperature over the resulting change in VF. This might typically be 0.5 degrees/mV. Expressed as 1/K that would be 2.0mV/degree. The "K Factor" is usually reasonably constant over a lot of diodes of the same type.

It is common to measure K on a sample and then to use the average value for all of the parts in the lot. K is measured by reading VF at room temperature and at a moderately elevated temperature in an oven or hot bath. The test current (IM) should be the same as used for the THERMAL RESPONSE test.

K is not constant over a wide range of currents.

There are two variations on the THERMAL RESPONSE test. The tests are commonly referred to as DVF (Delta VF) and THETA. I will describe DVF first since it is the most fundamental of the two.

First a measurement is made of VF at IM and the reading is remembered.

Next a constant current pulse of the specified amplitude (IH) and duration (TH) is applied.

Finally, the current is set back to IM and after a brief delay (TMD) the VF is measured once more and then subtracted from the initial cold VF. This results in the test result which is DVF. It is not necessary to know K when doing a DVF test.

THETA is almost exactly the same electrical test except that the VF is also measured at the heating current IH. This allows us to calculate the heating power (VF x IH). We now have enough information to express the results in "Degrees per Watt". THETA in Degrees per Watt is K times DVF over VF times IH.

You may be wondering about that "brief delay" TMD after the heating pulse and before the final VF reading. Ideally TMD would be zero so that the junction would have no time to cool before the final "hot" VF measurement. There are practical limitations however, some related to the diode and some to the test equipment. Published specifications normally specify the TMD to be used. This is typically in tens of microseconds for small fast diodes to hundreds of microseconds for large slow diodes.

The tests described above produce a single reading for a set of test conditions that you have chosen. It is also very useful to make plots of Thermal Response vs TMD and Thermal Response vs TH. These plots, called "Cooling Plot" and "Heating Plot" respectively, are described fully in another paper by this writer.

Frothingham Electronics Corp. produces a number of testers that measure DVF and THETA as described. Software included can produce the plot mentioned. Please contact this writer if you have any questions.