Heat dissipation design of FPGA and DSP signal processing system

introduction

As system performance continues to improve, system power consumption also increases. How to effectively dissipate heat in the system and control the system temperature to meet the normal working conditions of the chip has become a very difficult problem. Air cooling technology is usually used to dissipate heat in the system. When using air cooling technology, the heat dissipation efficiency should be considered. Generally, it can be achieved by using better thermal conductive materials and increasing the heat dissipation area, but this will increase the system cost and volume, so the optimal combination point must be selected. In addition, the direction of heat propagation should be fully considered so that while it is propagated to the outside world in the best possible path, it can ensure that the heat is away from those devices that are easily affected by temperature. Now, some companies have also launched auxiliary tools for system heat dissipation design, which greatly improves the reliability of system design.

1 System Structure

This system uses FPGA as the data acquisition and control center of the high-performance real-time signal processing system, and two DSPs as the data processing center. It mainly includes four functional modules: data acquisition module, FPGA data control module, DSP processing module and communication module. The system structure block diagram is shown in Figure 1.

The system uses an external 5 V regulated power supply as the main power supply; uses a 50 MHz external crystal input, and completes frequency division and multiplication inside the FPGA. There are two reset methods: power-on reset and manual reset. Inside the FPGA, a power-on reset signal is automatically generated through a counter, and then the signal and the reset signal provided by MAX811 are passed through the AND gate to generate a reset signal on the system board. This can ensure the power-on reset time and retain the manual reset feature of MAX811. [page]

2 System Power Consumption Estimation

The core part of this system is mainly composed of 1 FPGA (XC3S1500) and 2 DSPs (ADSP-TS201), which account for the main part of the system power consumption. Therefore, a rough estimate of this part of the power consumption should be made. At the same time, taking into account other devices on the board, the estimated result should be appropriately relaxed, and finally the specific design parameters of the power supply part are given.

(1) FPGA (XC3S1500) power consumption estimation

The XC3S1500 needs to provide three voltages for normal operation: 1.2 V core voltage, 2.5 V and 3.3 V I/O voltages. Its power consumption is estimated as shown in Table 1.

(2) DSP (ADSP-TS201) power consumption estimation

ADSP-TS201 needs to provide three voltages for normal operation: 1.2 V core voltage, 1.6 V on-chip DRAM voltage and 2.5 V I/O voltage. When ADSP-TS201 operates at 600 MHz, its power consumption is shown in Table 2.

3 Introduction to FLOPCB Thermal Design Software

FLOPCB is a software specially used for PCB heat dissipation design launched by Flomerics, a British company. Its interface after startup is shown in Figure 2.

The software has the following features:

◆Easy and quick establishment of PCB board-level temperature system model;

◆Intuitive and flexible result observation method;

◆The operation interface is simple and easy to use.

When designing heat dissipation, a heat dissipation solution for the system is provided by using FLOPCB. [page]

4 System heat dissipation design

It is not difficult to see from the power consumption estimation results in Table 1 and Table 2 that ADSP-TS201 and XC3S1500 are the parts that generate the most heat in the system and can be regarded as the heat source of the system. In FLOPCB, the temperature model 1 of the system PCB can be drawn, as shown in Figure 3.

In Model 1, no heat dissipation device has been added, and the simulation results are shown in FIG4 .

As can be seen from Figure 4, the temperature near ADSP-TS201 reaches about 75°C, which is very close to the normal operating temperature of ADSP-TS201, and the temperature around XC3S1500 also reaches 42.2°C. When a heat sink of 30 mm (L) × 30 mm (W) × 15 mm (H) is used, temperature model 2 can be constructed, as shown in Figure 5. The simulation results are shown in Figure 6.

Comparing Figure 4 with Figure 6, it is not difficult to see that the temperature around ADSP-TS201 has dropped to about 55°C, and the temperature around XC3S1500 has also dropped by 4.2°C. It can be seen that the heat dissipation performance of the system has been effectively improved by adding a heat sink, achieving the purpose of system heat dissipation.

Conclusion

This article mainly introduces the heat dissipation design method of general high-performance real-time signal processing system. Based on the system power consumption estimation, some software-assisted design is used to determine the device parameters and provide a heat dissipation solution for the core part of the system. When designing the system heat dissipation solution, the FLOPCB thermal analysis software is used to assist in the analysis, and combined with the system's own heat dissipation characteristics, a reference heat dissipation solution suitable for this system application is given. After actual verification, this solution does have a good heat dissipation effect.