Clifton Laboratories 7236 Clifton Road  Clifton VA 20124 tel: (703) 830 0368 fax: (703) 830 0711

E-mail: Jack.Smith@cliftonlaboratories.com

Heat Sink Measurements

I ran a series of quick tests on five heatsinks for TO220 tab-mount transistors today. The heat source is a Caddock MP930 film resistor, packaged in a TO220 style mount. The MP930 used has a nominal value of 10 Ω, 1% and is rated at 30 watts when mounted on an appropriate heat sink. The MP930's data sheet may be found here.

I applied DC power across the resistor with an HP 6205C dual DC power supply, paralleling the two outputs to obtain additional current. The voltage across the test resistor is measured with an HP 3468A digital multimeter. The MP930 test resistor measured 9.990 Ω using the HP 3468A in 4-wire resistance mode. Power is computed as E²/R, and I assumed the MP930's resistance remained constant over the test temperature range. (This is a reasonable assumption as the MP930's worst-case temperature coefficient is +80 PPM/°C. For a 100 °C temperature increase, the total resistance change is 0.8%.)

I monitored the temperature of the device and heat sink with a Fluke 62 infrared thermometer, and recorded the highest temperature reading. The maximum temperature reading was found on the MP930's surface, near the mounting hole. The data is plotted using Origin 7.5 and the data fitted with a linear curve fit to extract the thermal resistance.

In the illustration, Ta, Ts, Tc and Tj are:
Ta—Ambient (air) temperature
Ts—Heat sink temperature
Tc—Case temperature (the TO220 device's outer surface in our test arrangement)
Tj—Junction temperature, i.e., the temperature of the semiconductor junction, or in this case, the thin film resistive element, inside the case.

The three thermal resistances are:
Rjc—resistance from junction to case
Rcs—resistance from case to heat sink
Rsa—resistance from heat sink to ambient
 

Rcs is a function of the size of the device, the type of thermal compound used between the device and the heat sink and the thickness of the compound layer. For the TO220 device and the thermal compound I used (Dow Corning No. 4 silicone), I estimate Rcs is between 1 and 2 °C/watt.

If we use a heat sink with Rsa of, say, 12.7 °C/watt and if we apply 3.162 volts across the MP930, and if the ambient temperature is 70 °F, we may compute the various temperatures. 70°F corresponds to 21.1 °C.

First, the power dissipated in the MP930 is (3.162)²/10 = 1.0 watts.

The temperature drops are thus:

• Drop from junction to case: 1.0 watts * 4.17 °C/watt = 4.17 °C.
• Drop from case to heatsink: 1.0 watts * 1.5 °C/watt = 1.5 °C (taking the midpoint of the estimated Rcs
• Drop from heatsink to ambient: 1.0 watts * 12..7 °C/watt = 12.7 °C.

We may now compute the various temperatures:

• Ambient temperature, we have said is 21.1 °C
• Heatsink temperature is 21.1 + 12.7 °C or 33.8 °C
• Case temperature is 33.8 + 1.5 °C or 35.3 °C
• Internal junction temperature is 35.3 + 4.17 °C = 39.5 °C, or 103 °F

If we needed, for example, the heat sink rating needed to keep the junction temperature at 120 °C, with an ambient temperature of 100 °F (approx 38°C) and with 10 watts power dissipation, we would use the approach outlined above, but solving for Rsa.

Tcase = 120 °C - 10.0 watts * 4.17 °C/watt = 78.3 °C at case
Theatsink = 78.3 °C - 10 watts * 1.5 °C/watt = 63.3 °C at heat sink
Rsa = (63.3 °C - 38 °C) / 10 watts = 2.53 °C/watt

Thus, our target requirements dictate a heat sink with a thermal resistance of approximately 2.5 °C/watt. This is a large heat sink, by the way. We might consider relaxing the junction temperature limit (the MP930 is rated to 150 °C, after all) and also using a higher quality thermal compound to get by with a smaller heat sink.

The figure below shows the five heat sinks tested. In addition, I ran a test without any heat sink.
 

[2] With Dow Corning No. 4 silicon grease between MP930 and heat sink.

[3] With AAVID Thermalcote II heat sink compound between MP930 and heat sink.

The plot below shows the temperature versus power data for the six tests. Also shown, but difficult to see at the reduced scale version, are linear fits to the measured data. The slope of linear fit ΔT/ΔP (with T in °C, of course) provides Rsa for each measured heat sink.

I received a tube of AAVID Thermalcote II heat sink compound today and re-ran the Rotron / Comair heat sink test with it. The results are quite impressive. For a given maximum temperature, using Thermalcote II permits about 25% greater power dissipation.

I've had the tube of DC No. 4 for at least 35 years, so it's not surprising that better heat sink joint compounds have been developed and, I believe, DC No. 4 was intended more for electrical insulation and water displacement properties than for a heat sink compound.

I purchased several surplus heat sinks from All Electronics http://www.allelectronics.com/ intended for Pentium-type integrated circuit cooling with the thought of using them to cool TO-220 case type power MOSFET transistors and TO-220 style power resistors. The particular heat sink is described as:

AVC # AV-112C80FBL03. 12 Vdc, 0.1A, 60 mm ball bearing fan attached to an aluminum heatsink. Overall dimensions: 3.14" x 2.50" x 2.48" high. Bottom, mating, surface of heatsink is 3.04" x 1.85" Mating surface is prepped with a 0.8" square of thermal compound. Fan has three 8" leads with 3-pin female connector (0.1" spacing). UL, CSA, CE, TUV.
CAT# CF-286

I removed the mounting clip and drilled and tapped several 4-40 mounting holes in the heat sink's bottom surface. (Of course, if you drill & tap the bottom surface, remove the fan assembly first!) The photo below shows the heat sink with two Caddock MR930 resistors and two BUZ11 power MOSFETs installed. I used AAVID's Thermalcote II compound between the tabs and heat sink surface. BUZ11's connect the drain to  the tab, so the heat sink is also part of the drain circuit and the drains are in parallel.

The fan power connections are, by the way, positive to red, negative to black (12 volts). The white lead is an open-collector tachometer sensor, providing a speed-related pulse width for fan failure detection.
 

One question that I've assumed an answer to is what is the case-to-heat sink thermal resistance, R_θCS The commonly quoted value for a TO-220 metal tab transistor, employing decent quality thermal compound, is 0.5 °C/watt.

To measure R_θCS I added two Fluke 80PK-1 Type K bead thermocouple temperature sensors to an IRF510 MOSFET transistor mounted on the Comair/Rotron 411626B02500 heatsink used in my earlier studies. One attaches to the top of the IRF510's tab to measure case temperature and the second attaches to the heat sink near the IRF510, as shown below. The thermocouples are read with a Fluke 189 digital voltmeter on temperature scale, using a Fluke 80AK Type K to DMM adapter.
 

Based on these measurements, the standard quoted 0.5 °C/watt is conservative. Since the measured data involves subtracting two temperatures that are not that far apart at low power, the higher power readings are more trustworthy than those at lower power levels.
 

These error bands are large compared with the difference between the case and heatsink temperatures, which again suggests some caution should be exercised in using the measured 0.41 °C/watt value.