The figure above shows an analog signal chain commonly found in acquisition systems. On the left side of the figure is a sensor that collects various analog signals, followed by an amplification circuit. The amplified and conditioned signal is sent to an ADC to realize the conversion from analog signal to digital signal. The converted digital signal output is sent to a processor, such as MSP430, for subsequent processing. In addition, a complete system will also include the power system. When examining the system noise, every part of the system, including the power supply, and even the reasonable design of the processor will play a very good role in suppressing the system noise. In this section, we mainly discuss noise around the front and back of the data converter, which is the circled part in the figure. Power supply noise suppression can refer to the power supply section.
In the real world, everything we hear or see is simulated. Various types of sensors make it easy for us to convert these analog signals into analog voltage or current signals, so that the data converter can convert the analog signal into a digital signal, so that the processor can perform subsequent processing. The figure above shows a load cell circuit. The LCL-816G load unit on the left is a resistance bridge sensor, which needs power excitation. The power supply circuit part provides 5V DC voltage. When the 5V excitation voltage is applied to the sensor, its maximum load is 900g, and the full-scale output is ±10mV at this time differential signal. The output signal is amplified by the instrumentation amplifier. The instrumentation amplifier in the circuit is composed of two operational amplifiers (A1 and A2 in the figure) and peripheral resistors. Among them, the resistance values are:
The differential amplification in the figure is a classic amplifier circuit composed of dual op amps. It is advisable to choose OPA2337, a dual-channel CMOS op amp powered by a single power supply. After calculation, it can be found that the gain of the amplifier circuit is about 153V/V. This gain range is within the full-scale amplification range of the instrumentation amplifier, which will not cause saturation, and also meets the range of AD conversion. Through SPI or other interfaces, the processor can obtain the result of ADC conversion for subsequent calibration and other processing. The ADC conversion is initially selected as ADS7829, a 12-bit SAR data converter.
After the device selection is completed, the layout design is carried out, as shown in the figure below, the left side is the front view, and the right side is the rear wiring layout.
Let's first see how such a system performs. We might as well detect the DC characteristics of this system, that is, the pressure sensor is in the state of no pressure, so theoretically speaking, a fixed DC amount will enter the circuit. For the ADC, the converted result should be a fixed value. If we make a graph of the conversion result, the horizontal axis is the value of the converted digital signal, and the vertical axis is the number of occurrences of the corresponding value. The ideal result should be a peak at a certain point and only at that point. What did the actual result look like? As can be seen from Figure 6-5, the real results are quite different from our imagination. The figure below shows the data received directly after ADC sampling. We collected a total of 1024 data points. As can be seen from the figure, there are actually a total of 44 different data points. How to understand the impact of these 43 noise points on the system accuracy? From the figure below, we can see that theoretically there should only be one sampled value, but now the sampled value appears within a certain range. This range is the uncertainty range of the sampling result, and its size is 44, which is converted to 6.5 in binary. , which means that for this 12-bit system, 6.5 bits are inaccurate.
Let's look back at such a system design. For a weighing system, people are concerned with the range and accuracy of weighing, which is the requirement of system design. For the above-mentioned sensor with a full scale of 900g, if we require a minimum detection accuracy of 0.3g, we can calculate the number of digits N of the ADC selected, at least satisfying
In this way, we get that the resolution of the ADC should be at least 11.55 bits, and the 12-bit SAR ADC can already meet the system requirements. However, the actual accuracy obtained is far lower than the theoretical accuracy of the ADC, which shows the impact of noise on the system accuracy.