Empirical evaluation techniques:
Experiments were performed on prototype hardware to characterize detailed part dimensions, assembly tolerances and assembly forces. With regards to part characterization, investigations included non-contact laser scanning measurements of part details and assemblies, part flatness measurements during subassembly construction, TIM gap measurements, PCB board strain measured through assembly operations, each completed during global heat spreader attachment. In addition, during this assembly assessment the applied forces on VTMs were measured. The heat spreader characterization was completed using a laser based scanning system. Analysis software was used to compare laser scanning data to the 3D model. Figure 10 shows a sample measurement on an initial prototype heat spreader demonstrating about 1.9 mm flatness variance across the entire surface. In practice, this part will bend during assembly to conform to the PCB assembly (i.e., stiffener to board, with mounted VTM components).
Figure 10. Heat Spreader Flatness (sample)
Implemented as a design feature, the heat spreader incorporated fixed stand-offs (i.e., limited travel screw attach points) to establish the required TIM gaps. The stand-off dimensional height tolerance to individual pockets where TIM is applied is closely held during final machining of the heat spreader. These stand-offs make contact to the PCB surface when assembled. The PCB is also attached to a larger stiffener. Flatness data and VTM component height date were collected on six PCB subassemblies with the VTM component height being demonstrated to be within established tolerance limits.
Following the aforementioned characterization, an effort was completed to understand the gap variation, a parameter critical for control of heat transfer (thermal resistance), forces exerted on VTMs through TIM during assembly and volume of TIM to dispense during manufacturing. Several methods were employed and compared to characterize the gap. One method used putty assembled into the gap which was subsequently measured upon disassembly by laser scanning. The method, shown in Figure 11, is a side view of planes formed by the component and putty surface. As shown in Figure 12, the data gathered for various VTM sites indicate gaps for this set of initial parts that were on the low end of designed tolerance allowance and therefore provided an ideal worst case test vehicle for measurement of forces during assembly. Note, a capacitive bond line testing technique was also employed to quantify these gaps thereby correlating the putty gap measurements [2],
Figure 11. Putty Gap Laser Scan (sample)
Figure 12. VTM TIM Gap Measurement vs. Design Specification
With the gap understood, two techniques were used to quantify the mechanical loads imparted to the VTMs. The first technique employed a pressure sensitive film [3]. Given the range of expected loads, this work was completed using the Extreme Low film (7.2-28 psi usage range). As required, two sheets were cut to size and deployed at each VTM site and stand-off locations. The two-sheet type of pressure sensitive film is composed of two polyester bases, one being coated with a layer of micro-encapsulated color forming material and the other containing a layer of the color-developing material. Figure 13 shows sample results of the PSF study. The deeper coloration represents areas of higher pressure (therefore force) and the color variation from VTM site to VTM site represents the load variation across the assembly. These results were compared to a known load vs. coloration calibration study and yielded reasonable correlation to the predicted loads indicated in Figure 8.
Figure 13. Pressure Sensitive Film Study (sample)
In order to fully understand both the dynamic load magnitude and load distribution during assembly, a force sensor employing pressure sensitive ink was used [4]. The sensor in this case is created on two films with pressure sensitive ink artwork forming a matrix of sensor elements. Each element is sampled and recorded. Note, pressure sensitive ink sensors need to be carefully calibrated for use with selected substrates (i.e., using a material’s test load frame and small capacity load cell). Specific to this application, the sensors were calibrated using a discrete VTM and dispensed TIM under various compression rates (i.e., TIM strain rates). Once calibrated, theses sensors were mounted in-situ to monitor forces during heat spreader assembly. Note, care was taken to ensure sensors did not block free flow of TIM paste as this would result in high hydrostatic pressures where the paste is dammed. During loading, the sensor element sums the data acquired for a given area of interest and plotted in time (Figure 14). As shown, a peak force of about six (6) pounds was observed on the VTM, with this force quickly decaying to two (2) pounds in approximately two (2) minutes. Correspondingly, a colorized pressure contour map of individual sensor elements shows areas of higher pressure on the VTM at the time of peak loading. As shown in Figure 15, a band of highest pressure (14 psi) is represented by the deep orange color, a band of green represents a pressure of 7 psi while the average load is shown in the upper right corner (i.e., six lb).
Figure 14. Individual VTM Site; Applied Force vs. Time
Figure 15.1-Scan® Colorized Pressure Map
Expanding beyond the single site noted above, Figure 16 depicts an example of four individual VTM sites being simultaneously mapped. In each case, peak and residual loads are recorded. These, again, correlated well with the predicted results noted in Figure 8.
Figure 16. Multi-VTM Site; Applied Force vs. Time
Summary and Conclusions
A new cooling solution for a new high-end, energy-efficient server has been developed and verified. In early conceptual phases of the design, detailed material characterization of candidate TIMs was performed and response models developed. The material characteristics were a critical input to FEM analysis to verify the design point and to start to develop manufacturing tooling and assembly processes. Later in the development process, prototype hardware was used with PSF and pressure sensitive ink sensors to verify the design, the material response and the assembly process on form factor hardware. The combination of these early analysis techniques was the development of an aggressive cooling solution without delaying the product introduction.
