Optoelectronic generator with one PCSS

The optoelectronic generator operation is based on the frozen wave generator principle. It consists of a PCSS placed at one of the ends of a microstrip line, equivalent to a capacitive element (see Fig. 1). This microstrip line is charged via a resistor connected to a high voltage DC power source, for a relatively long time. The PCSS is then illuminated by infrared pulses provided by a laser source. The incident optical energy creates electron hole-pairs, greatly reducing the PCSS impedance and leading to an ultrafast release of the high voltage static field, confined to the microstrip line, in the form of a square electrical pulse. After some time, the recombination of the charge carriers naturally occurs in the PCSS and induces its electrical reopening. The duration of the electrical output signal t ₁ depends on the microstrip line length L, while the peak level V _A of the pulse depends on the applied bias voltage and optical energy E _A.

Fig. 1. Optoelectronic generator with one PCSS.

Experimental setup

PCSS used are the GP02-20 model sold by VISHAY. They are made of silicon and have a maximum bias voltage of 2 kV. The characterization set up makes it possible to determine the value of the on-state resistance R _on after a laser pulse, as well as the time t _on during which the PCSS remains electrically closed (the time before the beginning of the charge carrier recombination). The knowledge of R _on and t _on ensures the establishment of a correct PCSS model for simulations.

The optoelectronic generator depicted in Fig. 1 has been used to measure these parameters. The PCSS is illuminated by infrared pulses of 80 ps, delivered by an Ekspla PL2250 laser source with a wavelength of 1064 nm and a repetition rate of 1 Hz. A long 50 Ω coaxial cable of 100 m (type RG58CU, Ø4.95 mm, conductor 19/0.19 mm, NVP 0.66), terminating in a circuit open, is added at the end of the microstrip line of the optoelectronic generator (see Fig. 2(a)). With this cable, the reflected wave takes a long time to return to the PCSS. This procedure illustrates the recombination phenomenon and a relationship between the bias voltage, the optical energy of the laser, and the effect on the R _on and t _on parameters has been established. The R _PCSS impedance can be calculated using (1).

(1)$$R_{PCSS}={\lpar V_b - V_{out}\rpar \cdot Z_{load}\over V_{out}} - Z_{source}$$

Fig. 2. (a) Measurement bench to characterize the PCSS. (b) Evolution of R_PCSS after a laser pulse E_A of 10µJ at a bias voltage of 400 V.

V _b is the bias voltage, V _out is the measured voltage at the output of the SMA connector, Z _source is the impedance of the microstrip line of the optoelectronic generator (50 Ω) and Z _load is the impedance of the load (50 Ω of the oscilloscope). The evolution of the measurement of the resistance of the PCSS is plotted in Fig. 2(b) with a fixed bias voltage of 400 V. In the closed state, R _on is equal to 1 Ω. The time t _on when the PCSS remains electrically closed is a function of the amount of optical energy E _A emitted. For example, this gives the possibility of generating signals with a duration of 20 ns for 10 μJ.

Optoelectronic generator with two PCSS

It is also possible to design optoelectronic generators with two PCSS placed on either side of the microstrip line (see Fig. 3). An infrared pulse on both PCSS drops their impedance and allows the ultrafast release of the static high voltage field, confined on the microstrip line, in the form of a progressive wave and a regressive wave. The reflection of the regressive wave at the PCSS B placed near the ground induces an inversion of its polarity. A superposition of these two waves forms, at the output, a bipolar electric pulse. The duration of the electrical output signal t ₁ depends on the microstrip line length L, whereas the bias voltage and the optical energies E _A and E _B applied on the PCSS allow to respectively modulate the peak levels of the positive part V _A and the negative part V _B of the electric pulse. It is also possible to shift the time when the inter-alternating transition front t _A appears by acting on the trigger times T _{laser~pulse~A} and T _{laser~pulse~B} of the infrared pulses.

Fig. 3. Optoelectronic generator with two PCSS.

From a bipolar electrical pulse, and with the aim of tending toward a damped sinusoidal signal, the optoelectronic generator with two PCSS illustrated in Fig. 3 has been modified (see Fig. 4). Following PCSS A, a microstrip line associated with a stub, as presented in [Reference Zhang and Zhu17], introduces the resonance phenomenon. The bipolar electrical pulse is thus maintained over time to form a damped sinusoid.

Fig. 4. Damped sinusoids optoelectronic generator.

Figure 5 represents the realized model with Keysight ADS software of the damped sinusoids optoelectronic generator where the microstrip lines and the stub have been implemented. Both PCSS are considered switches with an on-state resistance of 1 Ω in the closed state, matching the impedance value of the components after a laser pulse (see Fig. 2(b)). Bias voltage applied to the microstrip line between the PCSS is modeled by a pulse source and the large value of the resistor R ₁ prevents from any reflection. The output signal, measured on the load resistor R ₂, is acquired once the microstrip line is fully charged and after closing the two switches. The substrate used is epoxy resin (FR-4) which has a thickness of 1.5 mm, a relative dielectric constant of 4.3 and a dielectric loss tangent of 0.02.

Fig. 5. Keysight ADS model of the damped sinusoids optoelectronic generator.

Simulations

To create a spectrum over a wide frequency band [300 MHz–3 GHz], and to obtain at least an attenuation of 20 dB for the two selected rejected frequencies (900 MHz and 1.8 GHz which correspond to European GSM frequencies), the transient pulse is chosen as the sum of three damped sinusoids by an optimization algorithm [Reference Reineix, Négrier, Lalande, Couderc, Andrieu and Desrumaux15]. Center frequency, peak-to-peak magnitude, and Q factor of each sinusoid must be optimized in order to respect the desired spectral shape. The results of the algorithm are listed in Table 1.

Table 1. Parameters of the three damped sinusoids

Each of these signals will be generated by a different damped sinusoids optoelectronic generator. A transient optimization (10 ps time step) of the output shape of the Keysight ADS model in Fig. 5 has been used to calculate the length and impedance of the microstrip lines. Figure 6 shows the geometry of the three determined optoelectronic generators to obtain the signals of Table 1. The summation of the three spectra makes it possible to create the global spectrum containing the two rejected frequencies (see Fig. 6(d)). In simulation, the attenuation for the first frequency at 900 MHz is 20 dB while the attenuation for the second frequency at 1.8 GHz is 30 dB.

Fig. 6. Transient and spectral shapes of the output signal (a) of the first generator with f _c = 786 MHz; (b) of the second generator with f _c = 1.47 GHz; (c) of the third generator with f _c = 2.46 GHz; (d) simulated global spectrum with the two rejected frequencies and the associated transient shape.

Summation with a combiner

Figure 8 shows the measurement bench used to generate the global spectrum by summation of the three signals with a combiner (ZFRSC-4-842-S+). An Ekspla PL2250 laser source delivers 80 ps pulses with a repetition rate of 10 Hz in the infrared range (1064 nm). The laser beam is separated into two adjustable parts using a half wave plate (HWP ₁, HWP ₂, …, HWP ₅) coupled to a beam splitter cube (BSC ₁, BSC ₂, …, BSC ₅). The distribution of the optical energy between the beams is adjusted to obtain the same illumination of the six PCSS. Adjustable mirrors (M ₁, M ₂, …, M ₆) are arranged after the splitter cube to equalize the length of the different optical paths. Each damped sinusoids optoelectronic generator is powered by a voltage power supply and is connected to a four-way combiner; the fourth unused input has been connected to a 50 Ω termination. To avoid phase shift before the signals are summed, the three cables (C ₁, C ₂, and C ₃) are paired in phase at ± 2^°. For the acquisition of the output signal, a Barth attenuation chain of 68 dB and a Keysight DSOX92004Q Infiniium oscilloscope have been used. It operates in real time with an 8 GHz bandwidth and an acquisition time of 20 GSa/s.

Fig. 8. Measurement bench to generate the global spectrum with a combiner; M: Reflective Mirror; HWP: Half Wave Plate; BSC: Beam Splitter Cube; C: Cable.

The simulated and experimental results are compared in Fig. 9. The good superposition of the curves allows to validate the Keysight ADS model. On the frequency band [300 MHz–3 GHz], two rejected frequencies have been obtained with attenuation levels in line with expectations. The first rejected frequency at 900 MHz has an attenuation of 20 dB while the second at 1.8 GHz has an attenuation of 25 dB. A slight difference between the experimental and simulated center frequencies of each damped sinusoids optoelectronic generator is noticeable. The center frequency values are 703 MHz, 1.37 GHz, and 2.38 GHz instead of 786 MHz, 1.47 GHz, and 2.46 GHz. This is caused by the miniaturization of optoelectronic generators when the desired center frequency increases. The length of the PCSS pins then becomes less negligible and induces a frequency shift. The prospects for improvement will focus on establishing a better PCSS simulation model. It will then be possible to take into account the PCSS pins length, as well as the impedance difference caused by the welding of these components on the microstrip lines. This should help to solve the center frequency shift of each damped sinusoids optoelectronic generator.

Fig. 9. Comparison between simulated and measured spectra.

Summation with an antenna by generator

Figure 10 shows the measurement bench used to generate the global spectrum by summation of the radiations of the three signals with antennas. The measurement bench is identical to that shown in Fig. 8 except that each damped sinusoids optoelectronic generator is connected to an antenna. The UWB antenna chosen for this application is based on the work of Koshelev [Reference Koshelev, Buyanov, Andreev, Plisko and Sukhushin10] which combines a magnetic loop and the radiation mode of a traveling wave antenna (see Fig. 11(a)). With a height of 21 cm, a length of 30 cm and a width of 10 cm, this antenna has a very wide adaptation with a matching bandwidth at − 10 dB between 300 MHz and 3 GHz (see Fig. 11(b)). Figure 11(c) shows that the realized gain in the main radiation direction is weak for low frequencies under 1 GHz and moderated/high for frequencies above. The spacing between each antenna is 30 cm. For the acquisition, a Montena SFE3-5G derivative electric field sensor has been placed at a distance of 8 m from the antennas. To visualize the experimental spectrum, the output signal acquired by the electric field sensor corresponding to the derivative of the field is integrated; an FFT is then applied.

Fig. 10. Measurement bench to generate the global spectrum with antennas; M: Reflective Mirror; HWP: Half Wave Plate; BSC : Beam Splitter Cube; C: Cable.

Fig. 11. (a) Antenna; (b) reflection coefficient; (c) realized gain in the main direction.

The simulated and experimental results are compared in Fig. 12. Since the transfer function of the antennas has not been taken into account in the simulation, the geometry and the bias voltage of the damped sinusoids optoelectronic generators are no longer adapted to maintain the shape of the desired global spectrum. Rejected frequencies are less attenuated: 15 dB at 900 MHz and 11 dB at 1.8 GHz. To better match the shape the Keysight ADS simulation must consider the antenna characteristics (in particular the gain evolution) to include their effects in the optimization of generator geometry.

Fig. 12. Comparison between simulated and measured spectra.