Device for simulating dissipation of thermal power by a board supporting an electronic component

I claim:

1. A board simulator comprising:

a thermoelectric cooler having a first side and a second side opposite said first side, said thermoelectric cooler having a power terminal and a ground terminal, said thermoelectric cooler having the property of using electrical power received between said power terminal and said ground terminal to cause said first side to be at a first temperature different from a second temperature of said second side;

a first temperature sensor thermally coupled to said first side, said first temperature sensor being capable of indicating a measurement of said first temperature;

a heat sink coupled to said second side, said heat sink having the property of dissipating heat from said second side; and

a second temperature sensor located to determine a temperature at a point between said thermoelectric cooler and said heat sink.

2. The board simulator of claim 1, wherein a difference in measurements of said first temperature sensor and of said second temperature sensor divided by thermal power passing through said thermoelectric cooler determines the thermal resistivity of said board simulator for a predetermined amount of electrical power supplied to said thermoelectric cooler.

3. The board simulator of claim 1 further comprising:

a thermally conductive substrate having a first side and a second side opposite said first side, said first side of said thermally conductive substrate being attached to said first side of said thermoelectric cooler and said second side of said thermally conductive surface being attached to said first temperature sensor.

4. The board simulator of claim 3 wherein said thermally conductive substrate comprises a diamond-kovar-diamond sandwich.

5. The board simulator of claim 1 further comprising:

a thermally conductive substrate having a first side and a second side opposite said first side, said second side of said thermoelectric cooler being attached to said first side of said thermally conductive substrate and said heat sink being attached to said second side of said thermally conductive substrate.

6. A board simulator comprising:

a thermoelectric cooler having a first side and second side opposite said first side, said thermoelectric cooler having a power terminal and a ground terminal, said thermoelectric cooler having the property of using electrical power received between said power terminal and said ground terminal to cause said first side to be at a first temperature different from a second temperature of said second side;

a first temperature sensor thermally coupled to said first side, said first temperature sensor being capable of indicating a measurement of said first temperature;

a heat sink coupled to said second side, said heat sink having the property of dissipating heat from said second side;

a first thermally conductive substrate attached to said first side and a second thermally conductive substrate attached to said second side, said first temperature sensor being attached to said first thermally conductive substrate; and

a second temperature sensor being attached to said second thermally conductive substrate.

7. The board simulator of claim 6 further comprising:

thermal insulation at least substantially surrounding said thermoelectric cooler, to thereby ensure that substantially all heat from said first side flows through said thermoelectric cooler.

8. The board simulator of claim 6 further comprising:

a coupon formed of a non-thermally conductive material, said coupon being coupled to said first side of said thermoelectric cooler; and

a device mounted on said coupon.

9. The board simulation of claim 8 further comprising:

thermal insulation at least substantially surrounding said coupon and said thermoelectric cooler to thereby ensure that substantially all heat from said coupon flows through said thermoelectric cooler.

10. The board simulator of claim 1 further comprising:

a third temperature sensor located in an ambient fluid surrounding said heat sink;

wherein said third temperature sensor is at a first distance from said first side of said thermoelectric cooler and at a second distance from said heat sink, said second distance being smaller than said first distance.

11. The board simulator of claim 1 further comprising:

a thermally conductive substrate having a first side and a second side opposite said first side, said first side of said thermoelectric cooler being attached to said second side of said thermally conductive substrate; and

a test component mounted directly on said first side of said thermally conductive substrate.

12. The board simulator of claim 1 further comprising a thermally insulative material at least substantially surrounding said thermoelectric cooler, to thereby ensure that substantially all heat from said first side flows through said thermoelectric cooler.

13. The board simulator of claim 1 further comprising:

a coupon formed of a non-thermally conductive material, said coupon being coupled to said first side of said thermoelectric cooler; and

a device mounted on said coupon.

14. A board simulator comprising:

a thermoelectric cooler having a first side and second side opposite said first side, said thermoelectric cooler having a power terminal and a ground terminal, said thermoelectric cooler having the property of using electrical power received between said power terminal and said ground terminal to cause said first side to be at a first temperature different from a second temperature of said second side;

a first temperature sensor thermally coupled to said first side, said first temperature sensor being capable of indicating a measurement of said first temperature;

a heat sink coupled to said second side, said heat sink having the property of dissipating heat from said second side;

a thermally conductive substrate located between said thermoelectric cooler and said heat sink; and

a second temperature sensor attached to said thermally conductive substrate.

15. The board simulator of claim 14 further comprising:

thermal insulation at least substantially surrounding said thermoelectric cooler, to thereby ensure that substantially all heat from a first side flows through said thermoelectric cooler.

16. The board simulator of claim 14 further comprising:

a coupon formed of a non-thermally conductive material, said coupon being coupled to said first side of said thermoelectric cooler; and

a device mounted on said coupon.

17. The board simulator of claim 16 further comprising:

thermal insulation at least substantially surrounding said coupon and said thermoelectric cooler, to thereby ensure that substantially all heat from said coupon flows through said thermoelectric cooler.

18. A board simulator comprising:

a thermoelectric cooler having a first side and second side opposite said first side, said thermoelectric cooler having the property of using electrical power to cause said first side to be at a first temperature different from a second temperature of said second side;

a first temperature sensor thermally coupled to said first side;

a heat sink coupled to said second side, said heat sink being in contact with an ambient fluid in all portions of said heat sink except for a region coupled to said second side; and

a second temperature sensor located in an ambient fluid surrounding said heat sink;

wherein said second temperature sensor is at a first distance from said first side of said thermoelectric cooler and said second temperature sensor is at a second distance from said heat sink, said second distance being smaller than said first distance.

19. The board simulator of claim 18 further comprising:

thermal insulation at least substantially surrounding said thermoelectric cooler, to thereby ensure that substantially all heat from said first side flows through said thermoelectric cooler.

20. The board simulator of claim 18 further comprising:

a coupon formed of a non-thermally conductive material, said coupon being coupled to said first side of said thermoelectric cooler; and

a device mounted on said coupon.

21. The board simulation of claim 18 further comprising:

thermal insulation at least substantially surrounding said coupon and said thermoelectric cooler to thereby ensure that substantially all heat from said coupon flows through said thermoelectric cooler.

22. The board simulator of claim 18 wherein:

the heat sink has at least a finned structure in contact with the ambient fluid, the finned structure dissipating heat received from said second side to the ambient fluid; and

the second temperature sensor is closer to said finned structure than to said thermoelectric cooler.

23. A board simulator comprising:

a thermoelectric cooler having a first side and second side opposite said first side, said thermoelectric cooler having the property of using electrical power to cause said first side to be at a first temperature different from a second temperature of said second side;

a first temperature sensor thermally coupled to said first side;

a heat sink coupled to said second side, said heat sink being in contact with an ambient fluid in all portions of said heat sink except for a region coupled to said second side; and

a second temperature sensor thermally coupled to said second side.

24. The board simulator of claim 23, wherein a difference in measurements of said first temperature sensor and of said second temperature sensor divided by thermal power passing through said thermoelectric cooler determines the thermal resistivity of said board simulator for a predetermined amount of electrical power supplied to said thermoelectric cooler.

25. The board simulator of claim 23 further comprising a thermally insulative material at least substantially surrounding said thermoelectric cooler.

26. The board simulator of claim 23 further comprising:

a coupon formed of a non-thermally conductive material, said coupon being coupled to said first side of said thermoelectric cooler; and

a device mounted on said coupon.

27. The board simulator of claim 23 further comprising:

a third temperature sensor located in an ambient fluid surrounding said heat sink;

wherein said third temperature sensor is at a first distance from said first side of said thermoelectric cooler and at a second distance from said heat sink, said second distance being smaller than said first distance.

28. The board simulator of claim 23 wherein:

the heat sink has at least a finned structure in contact with the ambient fluid, the finned structure dissipating heat received from said second side to the ambient fluid; and

the second temperature sensor is closer to said finned structure than to said thermoelectric cooler.

Description Submit all comments and votes

FIELD OF THE INVENTION

This invention relates to an apparatus and method for determining the temperature of a junction of a semiconductor die in an electronic component by measuring the temperature rise of a board on which the component is mounted, and using the board's temperature rise to determine the junction temperature.

BACKGROUND OF THE INVENTION

The amount of energy transmitted by an electronic component into a board on which the component is mounted depends on a number of factors including, for example, the materials used in the component's package, construction of the package, materials used in the board, and construction of the board. Similarly, the energy transmitted by the component into air adjacent to the component is also a function of a number of variables, such as the air's velocity and the construction of the component's package. Calculation of the amount of energy transmitted by a component into the air and into the board is extremely complex and time consuming.

However, such calculation is helpful to determine the performance of the component in a system. One parameter useful in calculating the destination of the energy emitted from a package is the thermal resistivity of the package. The thermal resistivity of an electronic component's package is typically measured with the package mounted on a board, the board having either ground and power planes, or only conductive traces on the top and bottom side of the board. The thermal resistivity can also be measured with the package suspended inside a box.

SUMMARY OF THE INVENTION

A board simulator in accordance with the invention is used to simulate a board (henceforth a "target board") on which an electronic component is to be operated, and to thereby estimate the energy transferred by the electronic component and the junction temperature of a semiconductor die inside the electronic component. The board simulator includes a thermoelectric cooler (such as a peltier device), a first thermally conductive substrate coupled to a first side of the thermoelectric cooler and a heatsink coupled to a second side opposite the first side. The thermoelectric cooler regulates the amount of heat transferred from the first side to the second side of the thermoelectric cooler, in dependence on electric power (also called "peltier power" or "bias power") supplied to the thermoelectric cooler. The first thermally conductive substrate acts as an isothermal cold plate that ensures uniform temperature across the first side of the thermoelectric cooler. The heatsink transfers heat away from the second side of the thermoelectric cooler into an ambient fluid (such as air, or water) that surrounds the heatsink. A second thermally conductive substrate can be optionally included in the board simulator, sandwiched between the second side and the heatsink.

A test component (described below) similar (in one case identical) to the electronic component can be mounted directly on the first thermally conductive substrate, if the first thermally conductive substrate has substantially the same area as the footprint of the electronic component. Alternatively, the test component can be mounted on an optional part called a "coupon" that is included in the heatflow controller and that is thermally coupled by the first thermally conductive substrate to the thermoelectric cooler first side. The coupon can be formed of any material (such as a plastic core of a printed circuit board, copper or aluminum), has known dimensions (e.g. width, thickness and length), and preferably has a known thermal conductivity (to estimate the amount of heat passing through the material for a given temperature drop across the material). In one embodiment, the coupon has the same area as the footprint of the electronic component to be operated on the target board. However, the coupon can have an area larger than the footprint, as long as the area outside the footprint is insulated.

A board simulator in accordance with this invention includes insulation that is wrapped around the optional coupon if present, and around the first thermally conductive substrate and the thermoelectric cooler. The insulation ensures that all heat received from terminals (e.g. leads) of the electronic component is transferred through the optional coupon if present, the first thermally conductive substrate, the thermoelectric cooler and thereafter through the heatsink into the ambient fluid.

A board simulator as described herein allows measurement of a junction temperature of a test component mounted on the board simulator. The board simulator acts as a thermal equivalent of the target board, and therefore allows the junction temperature of an electronic component mounted on the target board to be determined empirically, from the junction temperature of the test component.

Also, the board simulator can be used to measure thermal resistivity of a target board. In this context, thermal resistivity of any two points is the difference in temperatures between the two points divided by the thermal power passing between the two points. The board simulator includes a first temperature sensor and a second temperature sensor, to determine the respective temperatures on the first side and second side of the thermoelectric cooler. The second temperature sensor can be directly coupled to the second side of the thermoelectric cooler, or alternatively the second temperature sensor can be suspended in the ambient fluid.

In a first calibration step of the invention, the thermal resistivity of the board simulator is determined from the difference in temperatures of the two temperature sensors and a known amount of thermal power generated in a test component mounted on the board simulator. If the determined thermal resistivity is higher than that of a target board, the thermoelectric cooler is operated to cool the board simulator. Alternatively, if the thermal resistivity is lower than that of a target board the thermoelectric cooler is reverse biased and operated to act as a heater to increase the thermal resistance of the board simulator. So the first calibration step is used to measure the board simulator's thermal resistivities for different amounts of power (also called "bias power" or "peltier power") supplied to the thermoelectric cooler.

A test component that is attached to a board simulator is substantially similar to the electronic component that is mounted on the target board. The test component includes a heating element to generate thermal power equal to the thermal power known to be generated during operation of the electronic component. The test component has a package and terminals that are respectively identical to the package and terminals of the electronic component. The test component also has a temperature sensor to measure the temperature (hereinafter "junction" temperature) adjacent to the heating element.

The heating element and the temperature sensor can be formed adjacent to each other in a single integrated circuit die (hereinafter, "thermal die") included in the test component. If the electronic component is to be attached directly to a target board (e.g. by a flip chip method), the test component is also directly attached (e.g. by the same flip chip method) to the board simulator.

During a second calibration step, while the known amount of thermal power is generated by the heating element, the junction temperature is measured at the temperature sensor, for different values of the board simulator's thermal resistivity and different values of the bias power. The measurements in the two calibration steps described above can be used to empirically determine the junction temperature of an electronic component operated on the target board.

Specifically, a user, knowing the thermal resistivity of a target board can look up the board simulator measurements (in table form, in graph form or in a formula) of the first calibration step to empirically determine the bias power needed by the board simulator to mimic the target board. Then the user can use the determined bias power to determine the junction temperature of the test component from the board simulator measurements of the second calibration step. The test component's junction temperature empirically estimates the electronic component's junction temperature, according to the principles of this invention, and is used, e.g. by a system designer as described below.

The user can determine the target board's thermal resistivity in a number of ways. In one embodiment, the user attaches a test component to the target board to generate the known thermal power and thereby mimic conditions during operation of the electronic component on the target board. In an alternative embodiment, the user attaches a heater wire and a first thermocouple adjacent to each other on one side of the target board, suspends a second thermocouple in ambient air and measures the difference in temperatures of the two thermocouples while generating a known amount of thermal power in the heater wire. In the alternative embodiment, the rise in temperature of a target board at a known distance from the first thermocouple (for example, to mimic a neighbor electronic component) can be determined by mounting and operating a second heater wire at the known distance.

A board simulator as described herein provides an easy way to empirically determine the junction temperature as well as the amount of thermal power conducted through the terminals (such as leads) of the electronic component. Use of a board simulator eliminates bulky plumbing needed to measure the actual temperature rise in a prior art water cooled cold plate. Electrically controlling the heat passing through a thermoelectric cooler as described herein is easier and eliminates the prior art steps of measuring fluid flow and computing the amount of power dissipated by the fluid flow, with the inaccuracies inherent in such measurements and computations.

Also, for a given application, a system designer can pick a target board having a thermal resistivity low enough to meet a predetermined maximum junction temperature, e.g. a target board which dissipates a predetermined minimum power to be generated by operation of the electronic component. Also, heat loss from an electronic component into air surrounding the component can be determined as the difference between a known thermal power generated by the component and an empirically determined thermal power transferred to the target board.

BRIEF DESCRIPTIONS OF THE DRAWINGS

FIGS. 1A and 1B illustrate, in cross-sectional views, an apparatus including a test component attached to a board simulator during two calibration steps.

FIGS. 1C and 1D illustrate in an alternative embodiment, the apparatus of FIGS. 1A and 1B devoid of a coupon.

FIG. 1E illustrates, in a cross-sectional view, an apparatus similar to the apparatus of FIG. 1A.

FIG. 1F illustrates, in a cross-sectional view, an alternative embodiment of the apparatus of FIG. 1E.

FIG. 2 illustrates a thermal die on which are formed a heating element and a temperature sensor.

FIGS. 3A-3C illustrate three packages that enclose the thermal die of FIG. 2 in three variants of one embodiment.

FIG. 4 illustrates a model of thermal resistivities for the apparatuses of FIGS. 1A-1B.

FIGS. 5A-5E illustrate temperature rise Td-Tw from dissipation of a predetermined power by component 160 (FIG. 1B) for different bias power supplied to thermoelectric cooler 111.

FIGS. 6A-6D illustrate calibration graphs showing thermal resistivity (also called "lambda") vs. bias power (also called "peltier power") for various packages.

FIGS. 7A-7D illustrate measurement graphs showing junction temperature vs. peltier power for various packages.

FIGS. 8A and 8B illustrate in plan and cross-section views respectively, the use of a heater wire to measure thermal performance of a target board.

FIG. 8C illustrates measurement of temperature rise due to a neighboring electronic component.

FIGS. 9A-9C illustrate graphs of junction temperature Tj as a function of air flow rate for the apparatus of FIG. 1A for various target boards: aluminum, kovar low conductivity (FR4) printed circuit boards.

FIG. 10 illustrates a normal semiconductor die with a number of gates as the heating element and a diode as a temperature sensor for use as a test component in one embodiment.

DETAILED DESCRIPTION

The temperature of a junction inside an electronic component is determined empirically according to the principles of this invention by measuring the thermal resistivity of a target board on which the electronic component is to be mounted, and using the measured thermal resistivity to lookup a set of calibration measurements of a board simulator. The board simulator (also called "power dissipation control structure" and "heatflow controller") is calibrated with a test component for various thermal resistivities, and while dissipating various amounts of power (including the predetermined amount of power).

In one embodiment, a board simulator 110 (FIG. 1) is attached to a test component 120 that is similar to a predetermined electronic component to be mounted on a target board. Board simulator 110 includes a thermoelectric cooler 111, a heatsink 112 and a coupon 113. Coupon 113 is formed of the same material as the target board. Alternatively, coupon 113 can be formed of a non-thermally conductive material so that heat flowing to coupon 113 can be determined from the temperatures at the coupon's two sides, a first side 113a and a second side 113b.

In this embodiment, coupon 113 has an area approximately equal to the footprint F (FIG. 2) of package 121, e.g. an area enclosed between terminals 121A-121N (FIG. 1A). Also in this embodiment, coupon 113 is formed of FR4 in accordance with a well known industry standard specification, EIA/JESD 51-3, "Low Thermal Conductivity Test Board For Leaded Surface Mount Packages" (see Table 1 below). Coupon 113 can be formed of Kovar in a second embodiment (see Table 2 below) or of aluminum in a third embodiment (see Table 3 below).

Test component 120 (described below) is mounted on first side 113a by e.g. soldering onto traces in the normal manner for the FR4 coupon, or by thermal grease for the kovar and aluminum coupons. Second side 113b of coupon 113 is thermally coupled to a first side 111a of thermoelectric cooler 111, for example, by a first thermally conductive substrate 114 that can be formed, for example, of diamond, or as a sandwich of diamond/kovar/diamond. A second side 111b of thermoelectric cooler 111 is thermally coupled to a first side 112a of a heatsink 112, for example, by a second thermally conductive substrate 115 that can be formed of, for example, diamond. A second side 112b of heatsink 112 is exposed to the ambient air in region 141 surrounding board simulator 110. Heatsink 112 has a finned structure in contact with the ambient air, and dissipates heat received from first side 112a (which in turn receives heat from second side 111b of thermoelectric cooler 111 as discussed above).

Thermoelectric cooler 111 is enveloped in insulation 116. Insulation 116 ensures that substantially all the heat received by coupon 113 from test component 120 flows through thermoelectric cooler 111 to heatsink 112. In this embodiment insulation 116 includes rigid insulation 116a and soft insulation 116b. Rigid insulation 116a is attached to thermoelectric cooler 111, and provides structural support needed to mount board simulator 110 in, for example, a wind tunnel 140. Soft insulation 116b can be put all around coupon 113, thermally conductive substrate 114 and thermoelectric cooler 111 such that soft insulation 116b does not go beyond a plane passing through the first side 113a of coupon 113, to insure that insulation 116 does not disturb flow patterns in region 150 adjacent to test component 120.

In this particular embodiment, a first thermocouple 117 is mounted on a surface 113b of coupon 113. First thermocouple 117 is thermally coupled by thermally conductive substrate 114 to a first side 111a of thermoelectric cooler 111. A second thermocouple 118 is suspended in the ambient air surrounding heatsink 112. Thermocouples 117 and 118 are respectively used to monitor voltages Vd and Vw, that correspond respectively to device temperature Td and coolant temperature Tw. The difference Td-Tw is indicative of the thermal resistance of board simulator 110.

As illustrated in FIG. 1A, in this particular embodiment, second thermocouple 118 is at a first distance from first side 111a of cooler 111. Moreover, second thermocouple 118 is at a second distance from heatsink 112, wherein the second distance is smaller than the first distance. Therefore, the difference in temperatures sensed by first thermocouple 117 and by second thermocouple 118 is indicative of the thermal resistance of board simulator 110.

Insulation 116 has a thickness t of, e.g. 0.06 inch, a length l of, e.g. 4.5 inches and a width w (not shown) of, e.g. 3.25 inch. In three variants of this embodiment, coupon 113 is formed of a FR4 board, a kovar board, or an aluminum board, all available from Pycon, Inc., 3301 Leonard Court, Santa Clara, Calif. 95054.

Thermally conductive substrates 114 and 115 can be, for example, 0.6.times.0.6.times.0.008 inch diamond available from Norton Diamond Film, Goddard Road, Northboro, Mass. 01532. Thermoelectric cooler 111 can be a Peltier device, part number CP 1.0-71-08-L, available from MELCOR, Materials Electronic Products Corporation, 1040 Spruce Street, Trenton, N.J. 08648.

Rigid insulation 116a can be e.g. cardboard (made of wood pulp) of area 1 foot.times.1 foot and thickness 1/8 inch, while soft insolation 116b can be, e.g. plastic foam of area 6 inch.times.6 inch and thickness 0.5 inch both available from any hardware store, for example Orchard Supply Hardware, 720 W. San Carlos Street, San Jose, Calif. In another embodiment rigid insulation 116a and soft insulation 116b are encased in an aerogel (available from Aerogel Corporation) to form insulation 116. Insulation 116 can be formed of other materials such as fiber glass, cork, paper, plastic and vacuum in other embodiments.

Heatsink 112 can be, e.g. part number 2296B available from Thermalloy, Inc., 2021 West Valley View Lane, Dallas, Tex. 75234-8993. In one embodiment, all parts of board simulator 110 are attached by a thermal grease, such as THERMACOTE.TM. available from Thermacote, Inc., 2021 West View Lane, Dallas, Tex. 75234-8993. Thermocouples 117 and 118 can be, for example, part numbers EXPP-K-245 available from Omega Engineering, P.O. Box 2284, Stamford, Conn. 06906.

In one embodiment