What do GE, LG, Samsung and Whirlpool appliances have in common? They’re all examples of how Wi-Fi connectivity is rapidly becoming standard in refrigerators, stoves, washing machines and other large appliances. In fact, Wi-Fi is now such an important feature for consumers that retailers such as Best Buy and Home Depot enable shoppers to search for appliances that have it.
“Brands like Samsung are leading the charge with their Family Hub refrigerators, complete with touchscreen displays, internal cameras, and connectivity that lets you see what’s inside from your phone while grocery shopping,” says Albert Lee, Washington State’s largest independent appliance dealer. “Storing food smarter also means saving money. With Wi-Fi connectivity, that means getting receive alerts if the door is left open, if the temperature shifts, or if it’s time to replace the water filter. That means fewer spoiled items and fewer unexpected repair bills.”
This trend means appliance manufacturers must make Wi-Fi reliability and performance a top design consideration to stay competitive. One major factor is the electromagnetic radiation of antennas built inside what are essentially big metal boxes — not exactly an ideal environment from an RF perspective.
Additionally, some of these appliances have large electrical dimensions compared to the antenna integrated in them. An example is a refrigerator with a height, width and depth of 1.8 m, 0.9 m and 0.6 m, respectively, and the Taoglas FXP830 Wi-Fi antenna, which measures 0.0008 m, 0.042 m and 0.007 m.
This size disparity can substantially increase the simulation time even when applying numerically efficient solvers. As a result, efficient simulations are critical for both shortening product development cycles and meeting customer expectations for performance and reliability. Here’s how.
Using CST Microwave Studio Software/hardware Acceleration Techniques to Speed Simulations
One option is software based that uses a hybrid technique in CST Microwave Studio where the system is decomposed to two sub-domains referred to as source and platform, as shown in Figure 1.
Figure 1: Hybrid simulation [1].
Hybrid simulations provide freedom to select numerical algorithms for source and platform, which can effectively shorten the simulation time using minimal computational resources. However, this might not be sufficient for expediting product development cycle and meeting customer expectations, especially for wideband simulations where several frequency samples across the range need to be examined.
Figure 2 shows a generic model of the simulated refrigerator measuring 1.8 m, 0.9 m and 0.65 m, respectively. The Taoglas 2.4/5.8 GHz FXP830 antenna is located at the middle of the front panel for easy integration with the other electronics and PCBs there.
Note that PCBs and other fine details were not included in this generic model. The depth of 117 mm was considered for the front panel from the surface of the refrigerator. The FXP830 was used for performing simulations has an adhesive mount, which facilitates installation on the front panel. It also is flexible enough to accommodate curved surfaces.
Figure 2: A generic model of the refrigerator [2].
CST Microwave Studio’s hardware acceleration option uses graphics processing units (GPUs). In the first step, the system is simulated using the Transmission Line Method (TLM) in CST Microwave Studio without applying hardware acceleration using GPUs. Wideband simulation is considered with a frequency range from 2 GHz to 6 GHz.
Additionally, 10 farfield monitors at different frequencies are defined. The hexahedral meshcell count is about 126 million. The total simulation time for this setup is 11 hours and 10 minutes.
To investigate decreasing the simulation time through hardware, the hardware acceleration option in CST Microwave Studio was implemented. At the first step, one GPU was used to carry on the wideband simulation. Adding one GPU substantially decreased the simulation time from 11 hours and 10 minutes to 4 hours, 42 minutes and 26 seconds. In this setup, the TLM was used to solve the problem. The simulation time was further investigated by adding up to 4 GPUs.
Figure 3 compares the simulation times using different number of GPUs. Doubling the number of GPUs cuts the simulation time almost in half.
Figure 3: Simulation time versus the number of activated GPUs.
Performing simulations at a narrower frequency range or at a specific frequency instead of wideband simulation can further shorten the simulation time. For instance, running the simulations for the latter setup with four added GPUs for the frequency range from 2 GHz to 3 GHz and one farfield monitor at 2.4 GHz takes only 24 minutes and 2 seconds.
Figure 4 illustrates the 3D radiation pattern of the antenna in combination with the refrigerator. As predicted, the metallic bodies of the refrigerator substantially affect the radiation pattern of the antenna contributing to higher directivity and gain compared to free space.
Figure 4: The simulated 3D radiation pattern of Taoglas FXP830 Wi-Fi antenna mounted on the refrigerator.
The total efficiency and gain plots are illustrated in Figure 5.
Figure 5: (a) The simulated total efficiency and (b) gain.
The key takeaway is that using GPUs in combination with the transmission line method can considerably shorten the original simulation time. This can contribute to an efficient design and simulation of antennas for electrically large smart home appliances.
References
[1] Dassault Syst., Inc. (2024). [Online]. Available: https://www.cst.com/products/cstmws/.
Milad Mirzaee and Patrick Frank, “Efficient Antenna Design and Simulation for Electrically Large Smart Home Appliances,” in 2024 IEEE International Symposium on Antennas and Propagation and INC/USNC-URSI Radio Science Meeting (AP-S/INC-USNC-URSI), Florence, Italy, Jul. 14–19, 2024. ©2024 IEEE. Available: https://ieeexplore.ieee.org/document/10687192
