FrostyTech Mk.II Synthetic Thermal Test Platforms - Intel LGA775 & AMD K8 Versions Inclusive - Thermal, Acoustic, Material Test Parameters & General Evaluation Methodologies.

As the field of computer heatsink testing with synthetic thermal loads continues to expand, the equipment used to support these tests must also evolve. FrostyTech.com has developed the Mk.II Synthetic Thermal Test Platforms to meet the needs of an extended array of CPU socket formfactors, and Thermal Design Power (TDP) values.

The Mk.II Platform enables the evaluation of computer heatsinks separate from other influencing factors within the PC system and chassis.

Synthetic vs. Real World

The point of synthetic thermal testing is to remove the variations in heat loads generated by a CPU, as heat production is interconnected with its moment-by-moment processing work load. From our experience, synthetic thermal testing provides a more accurate and reproducible sets of results than testing with actual processors running in a system environment.

The challenge then for synthetic thermal testing is to replicate 'the CPU' a heatsink is designed to cool, both in thermal load and contact surface area.

FrostyTech Mk.II Synthetic Thermal Test Platform, configured for AMD K8 heatsinks.

The 38x38mm Copper Interface Die for K8 heatsinks is driven by a 125W heater cartridge while a type-K thermocouple measures temperature readings.

Mk.II Synthetic Thermal Test Platform - AMD K8 Version

In its AMD K8 configuration, FrostyTech's Mk.II Platform delivers a 125W heat load to socket 754, socket 939, socket 940 and socket AM2/AM2+/AM3 compatible AMD Athlon64/FX/X2, Sempron & Opteron class heatsinks by way of a 38mm x 38mm copper interface die. Set within the center of the copper die is a type-K thermocouple for taking temperature measurements once equilibrium is reached.

Dimensions of the Integrated Heat Spreader (IHS) for socket 754/939/940/AM2 class AMD processors is identical to that of the copper interface die, this allows the FrostyTech Mk.II Platform to mimic the surface area in contact with the base of a heatsink accurately, and generate a flow of heat over 1444mm² as would be the case with an actual CPU.

The 125W heat load represents the upper limits of thermal solutions designed for the socket 754/939/940/AM2 form factor, so it functions as a good basis for stressing each subject heatsink.

To maintain consistent heat output, a Sencore PR57 Variable Isolation Transformer is used to supply power to 120V Chromalox cartridge heaters. Once equilibrium is reached at the copper interface die, an Omega HH501DK Type-K digital thermocouple thermometer records the temperature results.

All heatsinks are mounted to the interface die with good quality silicon-based thermal compound applied in a consistent manner. Original thermal interface materials and pads are removed prior to testing.

Omega HH501DK Type-K Thermometer with Kapton encapsulated thermocouple.

The results of each heatsink under evaluation with both high (12V) and low (~5V) fan-speed settings (where applicable), are compared to a set of reference heatsinks for that socket formfactor. This list is constantly updated as reference models near EOL and new models added.

Information not expressly stated in the thermal test results chart includes ambient temperatures and the number of fans per heatsink. All values are quoted in degrees Celsius. Heatsinks which maintain the lowest rise over ambient temperature with the least amount of noise generation are considered best in our books.

Mk.II Synthetic Thermal Test Platform - Intel LGA775 Version

In its other configuration, FrostyTech's Mk.II Platform delivers a 150W and 85W heat load to socket 775 compatible Intel Pentium 4/D/Extreme Edition, Celeron, Core 2 Duo & Core 2 Quad class heatsinks by way of a 30mm x 30mm copper interface die. Set within the center of the copper die is a Kapton encapsulated type-K thermocouple for taking temperature measurements once equilibrium is reached.

FrostyTech Mk.II Synthetic Thermal Test Platform, configured for Intel LGA775 heatsinks. The 30x30mm Copper Interface Die is driven by an internal heater cartridge while a type-K thermocouple measures temperature readings

The dimensions of the Integrated Heat Spreader (IHS) used on these Intel processor families is identical in size to the copper interface die, in effect mimicking the surface area in contact with the base of an Intel LGA775 heatsink accurately. Depending on the class of CPU heatsink, a representative amount of heat is conducted through 900mm² surface area would be the case with an actual CPU.

The 150W heat load represents the upper limits of thermal solutions designed for the socket 775 Intel Pentium D / Extreme Edition & Core 2 Quad (multi-core) class heatsinks.

The 85W heat load represents the average limits of thermal solutions designed for Intel Core 2 Duo / Pentium 4 / Celeron class heatsinks among others. The Mk.II functions as a good basis for stressing each subject heatsink.

To maintain consistent heat output, a Sencore PR57 Variable Isolation Transformer is used to supply power to 120V Chromalox cartridge heaters. Once equilibrium is reached at the copper interface die, an Omega HH501DK Type-K digital thermocouple thermometer records the temperature results.

All heatsinks are mounted to the interface die with good quality silicon-based thermal compound applied in a consistent manner. Original thermal interface materials and pads are removed prior to testing. The results of each heatsink under evaluation with both high (12V) and low (~5V) fan-speed settings (where applicable), are compared to a set of reference heatsinks for the socket formfactor. The remainder of the testing parameters are identical to those described above.

Noise Isolating Enclosure

To record the noise generated by a particular fan and heatsink combination, FrostyTech uses a sound isolating enclosure lined with 2" of foam that is sealed from the outside environment during the recording process. The thick layer of foam helps to ensure that the sound recorded is only that generated by the heatsink/fan inside.

A monophonic microphone is positioned about 6" from the center of the enclosure so it is out of the direct air flow path generated by the fan. A solid state digital recorder is used record the sound picked up by this nondirectional mic. The digital recorder has a frequency range of approximately 500Hz to 3500Hz.

The purpose of the enclosure is to record an accurate representation of the noise generated by a heatsink and fan combination - separate from any other external computer noises like hard drives or power supplies. This recording provides an indication of the pitch and frequency of the noise generated by heatsink.

Sound Level Measurements and Acoustic Samples

To measure the level of sound produced by a heatsink in decibels, an Omega HHSL1 sound meter is used. The HHSL1 is highly accurate between a range of 35-130 dB. To measure the level of noise produced, a heatsink is placed in the Sound Isolating Enclosure and the sound meter positioned about 12" off to one side. The highest average level of noise is recorded.

For the acoustic heatsink tests, we record the noise signature of a particular fan and heatsink combo and measure the sound levels at both high (12V) and low (~5V) fan speeds where applicable.

Omega HHSL1 Sound Meter; range 35-120 dB.	Sound Isolating Enclosure for recording acoustic samples. A monophonic mic records actual fan sounds.

Surface Roughness Comparator

The base of a heatsink plays an important role in how well the cooling apparatus interfaces with the processor. Poor surface roughness will affect a good heatsink just as much as a good surface finish will improve thermal conductivity. In an effort to put a more quantitative spin on the comments we provide about base finishes, FrostyTech uses a Surface Roughness Comparator which offers a concise cross section of common machine surface finishes. Not every heatsink base finish will fall into the envelope of this gage, but it does offer a very handy set of representations.

This commercially available gage has 22 machined surfaces from 2 to 500 microinches; Lapped (2, 4, and 8 µ" RA), Ground (8, 16, 32, and 63 µ" RA), Blanchard Ground (16 and 32 µ" RA), Shape Turned (32, 63, 125, 250, and 500 µ" RA), Profiled (63, 125, 250, and 500 µ" RA), and Milled (63, 125, 250, and 500 µ" RA).

That about covers FrostyTech's test methodologies and equipment. Results from the Mk.II Synthetic Test Platform are not comparable with those of the Mk.1.5 Platform.