Structural design and experiment of double-terminal matched capacitor voltage divider

Author： GOZ Electric Time：2024-07-15 09:30:39 Read：12

According to the analysis, the key to the engineering design of the improved two-terminal matching capacitor voltage divider is the high-voltage arm capacitor C1, the low-voltage arm capacitor C2, and the structural design of the matching resistor R2 and capacitor C3 at the end of the cable. For the starting resistor R1, since the resistance value is selected to be 60~75Ω, the influence of the distributed inductance and capacitance of the ordinary coaxial structure is relatively small for signals below 100MHz. C1 and C2 of the capacitor voltage divider adopt the traditional "umbrella" structure, and no external parallel compensation capacitor is used. Copper foil is used to form the structural capacitor. According to the simulation of the equivalent circuit containing stray parameters, it can be seen that the inductance of the ground capacitor C2 of the voltage divider must be effectively controlled. Under normal circumstances, the middle electrode of the capacitor voltage divider basically does not change the electric field distribution from the inner conductor to the outer conductor, that is, the electric field always passes through the probe surface vertically. Therefore, the structural inductance of C1 and C2 is extremely small (if the edge effect of the electrode is ignored, it can be regarded as non-inductive), and the voltage divider responds well. For the matching resistor R2 and capacitor C3 at the end of the cable, qualitative analysis shows that if the structural inductance of C3 reaches the order of 10nH, the frequency characteristics of the capacitor voltage divider will be poor when the frequency reaches the order of 10 MHz or above, and the structural inductance needs to be controlled. Using a pinless coaxial structure capacitor and connecting multiple resistors in parallel is an effective way to control the inductance. Therefore, this capacitor voltage divider adopts a coaxial structure design, C3 uses a through-hole capacitor, and R2 uses 6 300Ω resistors in parallel and welded on a printed circuit board. The structure is shown in the figure. In order to accurately obtain the comparison waveform of the square wave response of the voltage divider, avoid the influence of the distortion of the signal feed on the analysis of the experimental results, and also facilitate the experimental measurement of the frequency response characteristics of the capacitor voltage divider, the capacitor voltage divider is installed in a calibration room with an impedance of 50 Ω. The impedance of the calibration room matches the network analyzer and the square wave signal source, and the frequency response meets the requirements. According to the analysis, if a capacitor voltage divider with exactly the same structure is made on other (practical application) transmission lines, the response characteristics of the voltage divider will remain unchanged.

The voltage divider experiment includes square wave response experiment and amplitude-frequency characteristic experiment. In the square wave experiment, the input end of the calibration room is connected to the square wave pulse signal generator, and the output end is measured by an oscilloscope through an attenuator; the output of the capacitive voltage divider is recorded by another channel of the oscilloscope. In the experiment, the amplitude of the square wave source is about 2kV, the leading edge is about 10ns, and the pulse width is about 1μs. The amplitude-frequency characteristic is to connect port 1 of the vector network analyzer as input to one end of the calibration room; connect the standard 50Ω equipped with the vector network analyzer as load to the other end; connect the output of the load at the end of the capacitive voltage divider cable to the high-impedance probe (bandwidth 300kHz~3GHz) equipped with the network analyzer, and its output is connected to port 2 of the network analyzer; the measured S21 parameter is the amplitude-frequency characteristic curve of the capacitive voltage divider. In order to verify the influence of C2 and the voltage divider response characteristics and the selection of cable length, an experiment of changing C2 was carried out. Due to the limitation of the calibration room diameter, the area of C2 cannot be very large, so only the verification experiments of C2≈1.4nF and C2≈2nF were carried out. The results of the square wave response experiment and the amplitude-frequency characteristic experiment are shown in Figures 9 and 10, respectively. For the convenience of comparison, the waveforms are also normalized.

When C2≈1.4 nF is designed, when a coaxial cable with an electrical length of 50 ns is selected (R1=93 Ω, C3=2 nF is designed), the overshoot of the square wave response waveform δ≈7%, which is basically consistent with the theoretical expectation shown in Figure 6; when a cable with an electrical length of 25 ns is used, the square wave response of the voltage divider has basically no overshoot, which meets the test requirements. When C2≈2nF, when a coaxial cable with an electrical length of 50ns is selected (R1=75Ω, C3=2nF), the overshoot of the square wave response waveform δ＜2%, which also meets the test requirements. As can be seen from Figure 9: C2≈1.4 nF, the amplitude-frequency characteristic of the voltage divider with a cable of 50 ns electrical length is slightly uneven in the mid-frequency band near 10 MHz (corresponding to the overshoot of the square wave response waveform); however, the voltage divider with C2≈2nF has a more serious gain in the high-frequency band above 300MHz, mainly due to the spurious parameters of C2. That is, when C2 increases, the spurious parameters of C2 also increase, while the capacitive reactance of C2 decreases, which leads to the deterioration of the high-frequency characteristics of the voltage divider. If the test signal contains a higher-frequency interference signal, the output waveform of the voltage divider is prone to superimposing high-frequency interference. Therefore, from the perspective of engineering design, even if the improved double-end matching method is adopted, the increase in the length of the test cable is still limited by the high-frequency response capability.

Conclusion

The improved double-end matching capacitor voltage divider can solve the problem of flat-top fading when measuring square wave signals with a pulse width of μs. The simulation calculation shows that: although the transient voltage divider ratio of the traditional two-terminal matching method is the same as the steady-state voltage divider ratio, the amplitude-frequency characteristics of the mid-frequency band are not flat, and the square wave response waveform has overshoot and other phenomena. When the low-voltage arm capacitance is 2nF, for a cable with an electrical length of 50ns, the overshoot of the square wave response waveform of the traditional two-terminal matching capacitor voltage divider exceeds 10%. The overshoot problem can be reduced by optimizing the design of the starting resistance and the corresponding end capacitance. The experimental results show that for a voltage divider with a low-voltage arm capacitance of 2nF and a cable with an electrical length of 50ns, the starting resistance is designed to be 75Ω, and the terminal uses an improved two-terminal matching method with a 50Ω series 2nF capacitor. The test waveform has no flat-top fading, and the overshoot is less than 2%.