3.1 Signal Conditioning Circuit
Signal conditioning can be specifically divided into two aspects: one is the effective suppression of noise interference; the other is to condition the output signal power.
3.1.1 Low-pass filter
There is no low-pass filter inside the AD9858, so the swept frequency signal output by the DAC inevitably contains high-frequency noise. This noise can be divided into two categories: one is the step waveform component and its higher harmonics caused by the DAC digital-to-analog conversion, and the other is the internal system clock of the AD9858 and its higher harmonics.
Therefore, a low-pass filter is required at the signal output port to suppress high-frequency interference.
Filters can be divided into active and passive filters. The design of active filters introduces active components—integrated operational amplifiers. Since operational amplifiers have nearly ideal characteristics and can omit inductors, frequency response characteristics close to theoretical predictions can be obtained and the volume can be reduced (especially at low frequencies, the inductance of passive filters is large). However, due to the limitation of the bandwidth of the operational amplifier, active filters are generally limited to frequency bands within a few hundred kHz.
Passive filters are designed using discrete components [23] and have a relatively wide frequency range, so they are generally used in high-frequency designs. Since the internal clock of the AD9858 is 800MHz and the maximum frequency of the output signal is
300MHz, the two filters in Figure 3-2 need to be designed as passive low-pass filters.
Low-pass filters can be divided into Butterworth filters, Chebyshev filters, Bessel filters, and elliptic filters. A comparison of the amplitude-frequency characteristics of various filters is shown in Figure 3-3.
The passband and stopband of the Butterworth low-pass filter are both flat, but its transition band is too flat; the passband of the Chebyshev low-pass filter is equiripple jitter, the stopband is flat, and the transition band Butterworth is slightly steeper; the Bessel low-pass filter is just the opposite of the Chebyshev low-pass, with a flat passband and an equiripple jitter stopband; the passband and stopband of the elliptic low-pass filter are both jitter, but its transition band drops rapidly and is very narrow.
In this system, in order to minimize the interference of the AD9858 internal system clock 800MHz when the output signal frequency is up to 300MHz, an elliptic filter with a narrow transition band characteristic is used, and a three-stage elliptic low-pass filter is used. According to the system requirements, the frequency of the output signal can reach 300MHz, and its passband is set to 350MHz. The three-stage filter has a transition band with a faster drop speed, which can effectively filter out high-frequency interference above 400MHz. The two filters in Figure 3-2 are both low-pass filters. Their designs can be exactly the same, but their functions are different: LPF1 is used to filter out the internal output high-frequency noise of AD9858, and LPF2 is to minimize the impact of the internal high-frequency noise brought by multiple different devices on the output signal. Considering that the actual design of the elliptic filter is different from the theoretical analysis, this design adopts a method that combines theoretical analysis with the actual spectrum characteristics of the measured filter. In the actual debugging, the component values of the filter are gradually changed to achieve the best spectrum characteristics.
First, according to the theoretical design method of the low-pass elliptic filter, the original three-stage elliptic low-pass filter is designed by looking up the normalized filter table, and the component values are preliminarily determined. As shown in Figure 3-4, the theoretically designed elliptic low-pass filter.
Secondly, the theoretical filter is improved. Since the characteristics of resistors, capacitors, and inductors are not ideal, especially at high frequencies, they need to be improved. In practice, a PCB test board is made and the filter is soldered. The soldered filter is measured with a frequency characteristic tester. If the amplitude-frequency characteristic is not good, the capacitor value is adjusted and the amplitude-frequency characteristic of the filter is measured again. This is repeated until the filter's characteristics meet the requirements. After experimental adjustment, the circuit parameters and characteristic curves of the three-stage elliptical low-pass filter are shown in Figure 3-5.
3.1.2 Power Control Circuit
If the power adjustable range of the output signal of the frequency sweep signal source is required to be greater than the power range of the output signal of AD9858, the power of the output signal needs to be regulated. This system uses a controllable gain amplifier to regulate the signal power.
As shown in Figure 3-2, a fixed broadband operational amplifier AD8009 and an RF power amplifier RF2317 are added behind the controllable gain amplifier AD8369 [24]. This is mainly to further increase the power of the signal to a higher power range. The AD8009 also has the function of converting the differential signal into a single-ended signal. The
AD8369 is a digital controllable gain amplifier with an adjustable amplification range of 45dB and a step size of 3dB. The output signal power can be adjusted.
The bandwidth of the AD8369 is 600MHz, with differential input and output, a 4-bit parallel or 3-terminal serial data interface, a differential input and output resistance of 200, and a single-port to ground equivalent resistance of 100. The digitally controlled gain multiple is related to the peripheral circuit of the AD8369. In this system, the input port of the digital controllable gain amplifier is the output signal of the filter LPF1. In order to match AD8369, a 100 Ω resistor is connected across the output end of the filter. This 100 Ω resistor serves as the input impedance of AD8369. The output port of AD8369 is a differential to single-ended signal conversion circuit composed of a broadband operational amplifier AD8009. The design of the entire circuit is shown in Figure 3-6. [page]
In Figure 3-6, the four 1nF capacitors of AD8369 are DC blocking capacitors, whose function is to prevent DC from being output from the port of AD8369.
The SENB signal is used to select the parallel or serial data port. When connected to a high level of 3.3V, the data is input in serial mode. When connected to ground, the data is input in parallel mode. In this system, data is input in parallel mode, and this pin is grounded in the circuit design.
The DENB signal is a data write control signal. In serial mode, the low level is valid, and in parallel mode, the falling edge latches the data.
In the case of Figure 3-6, due to the existence of internal input and output impedances of AD8369, its digitally controllable gain multiple will be related to the external impedance of the device. For the input end of AD8369, its internal differential input impedance is 200, and the external equivalent impedance is 100. The attenuation introduced by the input circuit is shown in formula (3-7).
For the output of AD8369, when performing AC analysis, the equivalent resistance of the circuit connected to the output of AD8369 can be approximated as Rf and Rs in series, and the differential output impedance inside the device is 200, so the attenuation introduced by the output circuit is shown in formula (3-8).
Considering the influence of the input impedance of the AD8009 subsequent circuit, the actual equivalent input impedance of AD8009 is always smaller than the value of Rf and Rs in series, so the actual attenuation of the AD8369 output circuit is slightly smaller than the above calculated value (about -9.5dB).
Since the internal amplification range of the device is +3dB~+48dB, the actual amplification range of the AD8369 circuit is -10dB~+35dB. The relationship between the four-digit control amount and the amplification factor is shown in Table 3-7.
In Figure 3-6, the broadband operational amplifier AD8009 is used as a conversion circuit for converting differential signals to single-ended signals, and it has an amplification function, which can amplify the differential signal by 2 times. When the amplification factor of AD8009 is 2, its large signal bandwidth is 400MHz, but it is limited by the working voltage of +5V. When the 50 load is matched, the maximum power of the output signal is 15dBm. It cannot meet the maximum requirement for the output signal power, and an RF amplifier device is needed to further increase the output power of the signal.
3.4 RF Amplifier Circuit
There are two points to consider when designing RF amplifier circuits: 1. The power of the signal meets the requirements; 2. The noise of the output signal is suppressed.
Whether the power can meet the requirements is mainly determined by the RF device. Low-pass filters are used to suppress noise.
. First, the position of the filter is selected. When the filter is placed after the RF device, the output impedance of the frequency sweep signal source is required to be 50. However, the output impedance of the LC filter is different at different frequencies, and the impedance matching cannot be guaranteed, so filtering is performed at the input end of the RF device. This low-pass filter also uses elliptical filtering, which is LPF2 in Figure 3-2. Its design is exactly the same as LPF1 and will not be repeated.
The RF device in this system uses RF2317, which has a bandwidth of up to 3.0GHz, input and output impedances of 75, a fixed amplification of 15dB for small signals, and an output signal power of up to +24dBm. It is powered by a single 9V or 12V power supply. Since the output impedance of RF2317 is 75, in order to make the output impedance of the frequency sweep signal source 50, a 150 resistor needs to be connected in parallel to the output end of RF2317.
The 150Ω and 75Ω resistors are connected in parallel to ensure that the internal resistance of the signal source is 50Ω, but this reduces the output power of the signal because the signal power is shared by the 150Ω resistor and the load. The 150Ω resistor consumes 1/4 of the RF2317 output power, so the maximum signal power added to the network is 24dBm×3/4=18dBm. The RF amplifier circuit is shown in Figure 3-7.
In the figure, 1nF DC blocking capacitors are added to the input and output ports of RF2317 because the device contains a DC voltage higher than 5V at the input port and a DC component of about 8V at the output port. To prevent the DC flow from affecting the front and rear stages or even damaging the device, a DC blocking capacitor is necessary. The 471uH inductor has the opposite function to the DC blocking capacitor. The inductor can suppress the AC signal from being shunted by the 12V power supply, while not affecting the DC bias voltage required for the RF2317 to work. When the frequency of the AC signal is 100kHz, the impedance of the inductor is 2×π×10 5×471×10 -6, that is, 296, and the equivalent impedance of the output port is 37.5, which is the parallel value of 150 and 50. 11% of the AC signal flows away from the power supply loop. At this time, the signal power added to the load network will decrease and cannot reach the maximum output power of the signal. Only when the frequency of the AC signal increases, the equivalent impedance of the inductor increases, and then the AC signal lost from the power supply loop can be ignored.
In Figure 3-7, five 51 resistors are connected in parallel to form a 10 current limiting resistor, which protects the device. When the internal current consumption of RF2317 is too large, the voltage drop on the 10 resistor increases, and the voltage applied to the device decreases accordingly. The operating point of RF2317 decreases and it cannot work at full speed, thereby reducing the operating current. The reason for using five 51 resistors in parallel is that the maximum operating current of RF2317 can reach 200mA, then the power consumed by the 10 resistor is 0.4W, and the power that the chip resistor can withstand is less than 0.1W, so multiple 51 resistors are selected in parallel and their equivalent impedance is about 10, and the power consumption of each 51 is only 0.08W. In the actual circuit, since the inductor is a winding inductor, its internal resistance is relatively large, and the measured value is close to 20, it can play a role in current limiting, so the five 51 resistors in parallel are omitted.
Amplifier Design Practice 125 Questions Answered
京公网安备 11010802033920号