Recent advancements in modern radar technology have underscored the imperative need for high-quality frequency-swept signals [1,2]. Linearly chirped microwave waveforms (LCMWs), known for their expansive time-bandwidth product (TBWP), are particularly advantageous for achieving both high-range resolution and extended detection distances [3]. Various microwave photonics approaches have been explored, leveraging the inherent benefits of photonics, such as high-frequency operation, extended bandwidth, compactness, and immunity to electromagnetic interference. The techniques are implemented using photonic methods, including opto-electronic oscillation, optical frequency division, Kerr frequency comb oscillation, self-heterodyne detection, spectral shaping, and wavelength-to-time mapping [4–6]. Among these, the optoelectronic oscillator (OEO) stands out for its capacity to generate high-frequency, low-phase-noise, and tunable microwave signals, facilitated by its ultra-high-quality factor (Q) using a long optical fiber delay line for achieving long energy storage time in the loop [7–9].

Fourier domain mode-locked optoelectronic oscillators (FDML OEOs) have emerged as a promising solution for broadband LCMW signal generation [10]. By rapidly tuning a frequency scanning filter and synchronizing the scanning period to the cavity round-trip time, FDML OEOs enable the coexistence and phase-locking of all modes within the loop, eliminating the need for mode buildup time and resulting in rapid frequency scanning signals with low phase noise. The key component in an FDML OEO is the rapidly scanning narrow bandpass filter [11,12], typically implemented with microwave photonic filters (MPFs) involving phase-to-intensity modulation (PM-IM) conversion. Common implementations of MPFs include fiber Bragg grating filters or narrow bandpass filters through stimulated Brillouin scattering, which require bulky and expensive tunable laser sources. These configurations are often plagued by complex structures, low energy efficiency due to optoelectronic conversions, and poor frequency scanning repeatability due to laser frequency drift. To address these challenges, we have proposed and demonstrated an FDML OEO using a low-cost, compact diode-tuned BPF [13]. This BPF leverages a diode-tuned phase shifter in a loss-compensated ring resonator to enable rapid passband frequency tuning with low power consumption. However, this approach is limited by its relatively narrow sweep bandwidth due to the small free spectral range of the microwave ring resonator.

For applications demanding low cost, compact size, and low power consumption, photonic integrated circuits (PICs) offer a highly favorable OEO implementation. Recent research and development efforts have focused on integrating OEOs onto a chip to achieve miniaturization, cost efficiency, and enhanced system stability [14–16]. For example, hybrid-integrated wideband tunable OEOs have been developed, combining lasers, modulators on silicon, and compact fiber loops with yttrium iron garnet filters for frequency tuning [17]. As to the integrated FDML-OEO, a hybrid FDML-OEO was demonstrated using a thermally tunable high-Q micro-ring resonator (MRR), a phase modulator, and a photodetector (PD) [15]. A bandwidth as wide as 6 GHz and a TBWP as large as 1.96×106 are realized by tuning the microheater power. To accommodate the slow speed of the thermally tuned MRR, a long length coil of 5.02 km is used in the OEO loop, corresponding to a scanning period of 25.1 μs. Similarly, Ref. [18] also used a thermally tunable MRR to realize an MPF, with an even longer single-mode fiber of 10 km to accommodate the slow response time of the MRR. Long et al. presented an architecture for generating coherent frequency combs and low-noise microwaves using an OEO integrated with a Kerr microcomb [19]. The OEO microcomb synergizes a microcomb of 10.7 GHz repetition rate and an X-band microwave with phase noise of −97/−126/−130dBc/Hz at a 1/10/100 kHz frequency offset, yet does not require active locking, additional lasers, and external radio frequency (RF) or microwave sources.

In the past decade, the TFLN platform has emerged as a promising platform for future electro-optic integrated devices. TFLN boasts a remarkably high electro-optic coefficient, which renders it highly suitable for making high-performance electro-optic modulators. Moreover, its low optical loss enables the creation of high-Q optical resonators. Recently, in a previous publication, we realized two different OEOs operating at the Ka-band, with phase noises lower than those of the X-band OEOs on SOI and InP platforms [16,20]. One is a fixed frequency OEO at 30 GHz realized by integrating a Mach–Zehnder modulator (MZM) with an add-drop MRR [16], and the other is a tunable frequency OEO at 20–35 GHz realized by integrating a PM with a notch MRR.

To obtain a LCMW signal with a high chirp rate, we present an integrated FDML OEO enabled by a high-speed electrically tunable MRR on the thin-film lithium niobate platform for the required PM-IM conversion, eliminating the need for expensive tunable lasers. In this FDML OEO loop, the electrically tunable MRR serves as a pivotal component for selecting the oscillation frequency via the PM-IM conversion process. Owing to the high electro-optic coefficient and low optical loss characteristics of the TFLN, the electrically tuned MRR exhibits higher tuning speed and higher Q-values compared to the thermally tuned MRR, which are critical for achieving high performance Fourier domain mode locking. The MRR’s resonance wavelength can be adjusted by altering the controlling signal applied to it, thereby tuning the MPF’s frequency. Experimental validation has confirmed the TFLN-based MRR’s excellent response characteristics when subjected to a sawtooth wave controlling signal at a frequency of 1.5 MHz. Our experiments have successfully generated a K-band LCMW signal with a frequency range from 17.7 GHz to 26.2 GHz and adjustable sweep bandwidths from 3.85 GHz to 8.5 GHz, with the capability of fine-tuning the center frequency and bandwidth of the generated LCMW signals. We have also evaluated the phase noise performance of the integrated FDML OEO at various frequencies, demonstrating a low phase noise of -106dBc/Hz at 10 kHz across oscillation frequencies of 20 GHz, 21 GHz, 22 GHz, 23 GHz, 24 GHz, and 25 GHz. Compared to FDML OEO configurations involving TLS, our approach offers significant advantages, including low cost, rapid tuning speed, high-frequency output, excellent linearity, and good signal repeatability.

Figure 1(a) presents the schematic of the on-chip integrated FDML OEO, which encompasses a phase modulator, an electrically tuned MRR on the TFLN platform, and a PD for forming an MPF through PM-IM conversion [21]. A continuous wave from the laser diode with its polarization aligned using a polarization controller is coupled into the phase modulator via butt-coupling, resulting in the generation of double sidebands with a π phase difference. This phase-modulated optical signal is then directed into the electrically tuned MRR, acting as an optical notch filter to suppress one of the first-order sidebands of the phase-modulated signal. In our TFLN-based FDML OEO, the PM features a low half-wave voltage and a large modulation bandwidth, which help to relax the required gain of the low-noise amplifier (LNA), resulting in reduced power consumption and cost. In addition, the higher the Q factor of the MRR, the better the selectivity of the filter, which in turn leads to the superior phase noise performance of the OEO [22]. Subsequently, the signal is amplified by an erbium-doped fiber amplifier (EDFA) and transmitted through a long delay optical fiber before being converted into an electrical signal by the PD. The long delay optical fiber is composed of a length of standard single-mode fiber (SMF-28) and a length of dispersion compensating fiber to minimize the total dispersion and hence the phase noise contribution from the laser diode (LD) frequency fluctuations. This electrical signal is then amplified by an LNA and reinjected into the PM, forming a closed-loop configuration. The EDFA and LNA are used to provide sufficient gain for the oscillating signal in the OEO loop, and the MPF is used for mode selection. Self-sustained oscillation can be established from noise when the optoelectronic oscillator fulfills Barkhausen conditions [23]. Figure 1(b) illustrates the operational principle of the PM-IM conversion-enabled MPF. The oscillation frequency of the OEO is governed by the frequency difference between the optical carrier and the nearest resonance peak of the MRR, which is modulated by a periodic voltage signal applied to the MRR’s electrodes. Fourier domain mode-locking is achieved when the OEO loop delay aligns with an integer multiple of the period of the driving signal applied to the MRR. The delay length of the loop can be measured by a network analyzer. Fourier domain mode-locking is achieved when the OEO loop delay aligns with an integer multiple of the period of the driving signal applied to the MRR, thereby enabling the generation of wideband LCMWs via Fourier domain mode-locking.

Mathematically, consider the input optical signal to the system, Ein(t)=E0exp[j(ωct+φc)], where E0 is the amplitude, ωc is the angular frequency, and φc is the initial phase. The RF input signal, denoted as (t)=VRF(ωRFt), drives the PM. The resulting phase-modulated signal can be given under a low modulation depth condition given by [24], $$\begin{gathered} EPM(t)=E0exp(jϕc){J0(β)exp(jωct) \\ +J1(β)exp{j[(ωRF+ωc)t+π2]} \\ −J1(β)exp{−j[(ωRF−ωc)t+π2]}}, \end{gathered}$$(1)where β=πVRF/Vπ is the modulation depth, Vπ is the half-wave voltage of the phase modulator, and Jn (n=0, 1) is the Bessel function of the first kind. It can be seen that the output of the phase modulator comprises the carrier and two first-order sidebands. When the phase-modulated optical signal is launched into a PD, the current is generated by this intensity-modulated signal as it passes through the photodetector, which can be expressed as i=ℜPcJ0(β)J1(β)H(ωc)cos(ωRFt+π2)[H(ω1)−H(ω−1)],(2a)$$\begin{gathered} H(ω1)=r(ω1)cos[θ(ω1)], \\ H(ω−1)=r(ω−1)cos[θ(ω−1)], \end{gathered}$$(2b)where ℜ is the sensitivity of the PD, Pc=Ec2 is the power of the carrier, H(ωi) (i=1 or −1) is the transfer function of the MRR [25], r(ωi) is the power transmission coefficient, θ(ωi) is the phase transmission coefficient, ωc is the angular frequency of the optical carrier wave, ω−1=ωc−ωRF is the angular frequency of the −1st-order sideband, and ω1=ωc+ωRF is the angular frequency of the +1st-order sideband. When no filter is present, H(ω1)=H(ω−1), resulting in i=0. When one of the phase modulation sidebands is suppressed by falling into a notch of the MRR, H(ω1)≠H(ω−1). The perfect cancellation is voided such that the phase modulated signal is converted into an RF signal by the PD, with a frequency equal to the spacing between the carrier and the notch. As shown in Fig. 1(b), an optical notch filter is mapped into a microwave bandpass filter, and the center frequency of the realized MPF can be tuned by tuning the resonance wavelength of the MRR. The center frequency of the notch varies linearly with the scanning voltage applied to the MRR, and then the output of the PD can be given by i∝ℜPcJ0(β)J1(β)H(ωc)[H(ω1)−H(ω−1)]⁢cos(2πfmt+πkt2+π2),(3)where fm is the frequency of the microwave signal, and k is the chirp rate. When the round-trip time of the OEO matches an integer multiple of the repetition time of the driving signal applied to the MRR, Fourier domain mode-locking is achieved. When mode-locking is achieved, each longitudinal mode in the OEO is synchronized with the MPF such that the MPF appears to be stationary with respect to each mode. Consequently, all longitudinal modes are always present and do not have to rebuild themselves when the filter is tuned, enabling the generation of wideband LCMWs via Fourier domain mode-locking. The periodic microwave waveform results in discrete and equally spaced frequency modes in the Fourier domain. All the modes are stably oscillating together and are mode-locked with a fixed phase relationship among them. Consequently, we can adjust the turning frequency step and scanning period by altering the round-trip time of the OEO loop and leveraging harmonic mode locking [13]. By combining Eq. (2b), it is evident from Eq. (3) that noise in the driving voltage signal can influence the OEO loop gain via the MRR transmittance and phase. This, in turn, affects the phase noise of the OEO output signal. Therefore, it is necessary to use a low-noise driving voltage signal to mitigate its detrimental impact on the phase noise performance of the OEO system.

As the key component in the system, the characterization of the electrically tuned MRR is essential for evaluating the spectral response of the OEO. The MRR is near critically coupled but still under-coupled to strike a balance between a high Q-value and a sufficient extinction ratio. Figure 2(a) shows the measured transmission spectrum of the MRR. As we can see, the free spectral range (FSR) and the extinction ratio (ER) of the MRR are 0.568 nm (corresponding to 71 GHz) and 9 dB, respectively. The notch has a 3-dB bandwidth of 272 MHz at 1550.273 nm, which corresponds to a Q-factor of 7.1×105. By applying DC voltages of 0–10 V to the MRR, as shown in Fig. 2(b), the notch of MRR can be shifted from 1550.274 nm to 1550.217 nm, corresponding to a voltage tuning efficiency of 5.5 pm/V or 0.688 GHz/V. The 3-dB bandwidth of the MPF is measured to be 272 MHz in the wavelength tuning range. From Fig. 2(c), we can see that the notch of the MRR has good linearity with the input voltage from 1550.274 nm to 1550.217 nm.

To achieve FDML OEO, it is essential to match the MPF periodicity with the OEO loop’s round-trip time. A round-trip time of 0.5 μs (equivalent to a 100-meter fiber delay) requires a bandwidth of approximately 2 MHz. Using an electrically tuned MRR, a fast tunable bandpass MPF was implemented via PM-IM conversion. We measured the MRR’s response to a sawtooth wave driving signal, as shown in Figs. 3(a) and 3(b). A continuous-wave narrow-linewidth laser was sent into the MRR with its wavelength within the MRR’s free spectral range and was tuned by the sawtooth wave at different frequencies. The output was detected by a low-noise APD, producing a voltage-time curve [Fig. 3(b)]. This curve helps characterize the MRR’s voltage response by analyzing the 3 dB temporal width of the APD output signal as the sawtooth wave frequency changes. The 3 dB temporal width Tf (defined as the time required to scan across the MRR’s 3 dB bandwidth Δf3dB) can be expressed as Tf∝Δf3dBdf/dt,(4)where df/dt is the frequency scanning rate of the MRR. Equation (4) shows that Tf is inversely proportional to the frequency scanning rate (df/dt).

An experiment based on the setup illustrated in Fig. 3(b) was conducted to delineate the response characteristics of the MRR to the sawtooth wave driving signals at different frequencies. A continuous-light-wave centered at a wavelength of 1550.32 nm was emitted from a laser (Pure Photonics PPCL500) with a linewidth of 10 kHz. The optical signal, after traversing the MRR, was detected by an InGaAs avalanche photodetector (APD KY-PRM-200M-I-FC) and an oscilloscope (Rohde & Schwarz RTB2004) with a bandwidth of 300 MHz and a sampling rate of 2.5 GSa/s. Concurrently, the MRR was driven by sawtooth wave signals at various frequencies within a voltage range of −1.5V to 5.5 V and was generated by an arbitrary function generator (AFG, Tektronics AFG3051C).

By increasing the sawtooth wave frequency (i.e., decreasing period) while maintaining a constant range, the frequency scanning rate (df/dt) of the MRR resonance will also increase. Figure 3(c) shows the 3 dB temporal width (Tf) of the APD output voltage responses at different frequencies. The red curve represents the simulated results with the values of Δf3dB and df/dt at 272 MHz and k×4816×f[k×688MHz/V×7V×f, where k=2 is a coefficient related to the frequency scanning rate, and f is the sawtooth wave frequency], respectively. The blue curve is the fitted results based on the measured data. As can be seen, at up to 2.5 MHz the measured data agrees well with the theoretical curve, allowing for accurate calculations of their temporal widths. However, beyond 2.5 MHz, the deviation between the experimental data and the simulation curves increases significantly. It can be assumed that the MRR exhibits an ideal frequency response for the sawtooth wave signals within the bandwidth of 2.5 MHz. Figures 3(d-i) and 3(d-ii) show the measured temporal waveform of the APD output signal at a sawtooth frequency of 1 MHz, where the 3 dB temporal width Tf was about 0.025 μs. Figures 3(e-i) and 3(e-ii) show the measured temporal waveforms of the APD output signal (blue lines) and the sawtooth wave output from the AFG applied to the MRR (red lines) at a frequency of 3 MHz. As can be observed, as the frequency of the driving signal applied to the MRR is increased from 1.0 to 3.0 MHz, the 3 dB temporal width Tf has decreased from the value of 0.025 μs to 0.012 μs, and the voltage at the notch point increased from 0.92 V to 1.5 V. At this notch point, the resonant wavelength of the MRR matches the wavelength of the incident laser. Note that owing to the limitation of the modulation bandwidth of the electrically tuned MRR, one can observe a slight response of the APD output voltage signal at the corner of the falling edge in Fig. 3(d-i). In Fig. 3(e-i), the notch point is closer to the middle of the falling edge to experience a higher frequency scanning rate, which exceeds the operating frequency of the electrically tuned MRR. Therefore, the response of the APD output voltage signal at the falling edge cannot be observed. The maximum frequency-scanning speed of the loop filter is governed by the design of the MRR electrode and ultimately the electro-optic effect of TFLN. Furthermore, the scanning speed can be raised by incorporating a traveling-wave electrode into the MRR, which will reduce the Vπ, thereby improving the measured electrical tuning efficiency.

An experiment based on the setup shown in Fig. 1(a) was performed. A continuous optical carrier generated by a 1549.9135 nm narrow linewidth DFB laser with an output power of 13 dBm and a relative intensity noise (RIN) of −140dB/Hz at 10 kHz was injected into the lithium niobate photonic chip by means of butt-coupling after passing through a length of polarization maintaining (PM) fiber. The phase-modulated signal was then sent into the micro-ring resonator (MRR), effectively suppressing the upper first-order sideband. After passing through the MRR, the output optical signal was amplified by an EDFA (model AEDFA-PKT-DWDM-15-B-FA) to compensate for the on-chip insertion loss before transmitting through a 500-meter dispersion-compensating fiber (DCF). The PD used had a 3-dB bandwidth of 50 GHz and a responsivity of 0.65 A/W. An LNA with a gain of 42 dB was deployed as the electrical gain medium. An electrical coupler (EC) extracted the signal from the LNA for evaluation using various instruments, such as an electrical spectrum analyzer, a phase noise analyzer (Rohde & Schwarz FSWP26), and an oscilloscope (Keysight UXR0134A) with an analog bandwidth of 13 GHz and a sampling rate of 128 GSa/s. To address the bandwidth limitations of the oscilloscope, a mixer was used for down-conversion, allowing the resulting difference frequency signal to be analyzed by both the spectrum analyzer and the high-speed oscilloscope. A 16 GHz signal was injected into the local oscillator (LO) port of the mixer to produce a difference frequency by mixing with the signal to be tested.

A sawtooth wave signal with an amplitude range of −2V to 10 V (12VPP), generated by the arbitrary function generator (Rigol DG4202) and with a period of 2.64 μs (corresponding to a frequency of 380 kHz), was applied to the MRR’s electrodes. This matches the loop round-trip time for achieving the Fourier domain mode-locking operation. Figure 4(a) presents the measured spectrum of the generated LCMW, as measured by the electrical spectrum analyzer (Rohde & Schwarz FSWP26), spanning a frequency range from 17.7 GHz to 26.2 GHz (8.5 GHz) with a span of 10 GHz. Figure 4(b) provides an expanded view of the spectrum around 22.003 GHz with a span of 2 MHz. The frequency spacing between adjacent modes is 380.0 kHz and is consistent with the loop delay, and the mode powers are uniform. Figure 4(c) shows the optical spectrum of the LCMW obtained using an optical spectrum analyzer (OSA WaveAnalyzer 1500S). The spectral widths of the upper and lower first-order sidebands of the phase-modulated signal are approximately 0.068 nm, corresponding to a sweep width of 8.5 GHz. Additionally, due to the suppression of the upper 1st-order sideband (in frequency) by the MRR, the lower 1st-order sideband is 7 dB lower than the higher 1st-order sideband (in wavelength).

The time domain waveform is measured by a high-speed oscilloscope (Keysight UXR0134A), as shown in Fig. 5(a). The temporal period of the waveform is approximately 2.64 μs, matching the period of the control voltage applied to the MRR. The corresponding instantaneous frequency distribution of the LCMW is shown in Fig. 5(b), which is calculated by short-time Fourier transform (STFT) with MATLAB. This analysis confirms the 2.64 μs period of the LCMW and indicates a bandwidth of 8.5 GHz per period, consistent with the bandwidth depicted in Fig. 4(a). The TBWP of the LCMW is calculated to be 22,440. Additionally, Fig. 5(b) shows that the waveform of the demodulated microwave signal closely matches the waveform of the driving voltage. Due to the linear response of the MRR filter, the generated LCMW signals exhibit good linearity. By linearly fitting the time-frequency curve in one period, it can be observed in Fig. 5(c) that the sweep linearity R2=0.99948 and the sweep rate is 3.22 GHz/μs. To assess the pulse compression capability of the generated LCMW, its autocorrelation was calculated, as shown in Fig. 5(d). The full-width at half-maximum (FWHM) of the result is approximately 0.08 ns. This corresponds to a pulse compression ratio of 33,000, which is the ratio of the temporal period of the waveform (2.64 μs) to the FWHM (0.08 ns) [26].

To enhance versatility for various applications, key parameters like bandwidth and center frequency are adjusted. The center frequency and scanning bandwidth of the LCMWs are tuned by changing the peak-to-peak voltage of the driving signal applied to the MRR. Figure 6(a) shows the spectra of the LCMWs with scanning ranges of 3.85 GHz, 4.50 GHz, 6.00 GHz, 7.50 GHz, and 8.50 GHz, corresponding to driving signal values of 5.0VPP, 6.0VPP, 8.0VPP, 10.0VPP, and 12.0VPP. The corresponding chirp rates are 1.46 GHz/μs, 1.70 GHz/μs, 2.27 GHz/μs, 2.84 GHz/μs, and 3.22 GHz/μs, respectively. The frequency of the MPF is the frequency difference between the optical carrier and the nearest resonance. The maximum bandwidth, theoretically up to 35.5 GHz (0.28 nm), is determined by half of the MRR’s FSR.

Additionally, the center frequency of the generated microwave signal was tuned by controlling the DC offset voltage of the electrical driving signal applied to the MRR. As shown in Fig. 6(b), the central frequency was tuned from 18.55 GHz to 23.59 GHz with a step of 1.25 GHz. The tuning resolution is limited by the mode spacing of the OEO to 380.0 kHz in this experiment. The tuning range of the FDML OEO in the experiment is limited by the FSR of the MRR. Note that there are some spurious signals at frequencies higher than the main signal, specifically in the purple and green curves of Fig. 6(b), which are the down-converted signals resulting from the second harmonics of the local oscillator signal and the FDML-OEO output signal.

The phase noise of the generated microwave signal is measured using a phase noise analyzer (Rohde & Schwarz FSWP26) at fixed frequencies without scanning, as shown in Fig. 7(a). Figure 7(b) presents the measured phase noise at the 10 kHz frequency-offset for various microwave signal frequencies at 20 GHz, 21 GHz, 22 GHz, 23 GHz, 24 GHz, and 25 GHz, respectively, consistently around −106dBc/Hz. For comparison, the phase noise of a signal from a commercial microwave source (Keysight E8257D) is also measured and shown in Fig. 7(a). The phase noise of the signal from the FDML OEO at the high-frequency offset is lower than that from the microwave source but is much higher at the low-frequency offset. This increase in phase noise near the carrier frequency is mainly due to the environmental fluctuation affecting the OEO and flicker noises in the microwave components such as LNA. The fluctuations of the laser frequency and the noise of the voltage applied onto the notch MRR can also be converted to the phase noise of the OEO. We can employ a laser with low frequency jitter and relative intensity noise, as well as a low-noise driving signal for the MRR, to reduce the phase noise. Proper packaging to minimize the power fluctuation due to vibrations of the chip-fiber coupling and eliminating the EDFA can also help to reduce the phase noise close to the carrier frequency. It is observed that the SSB phase noise measurement exhibits several sharp peaks. These peaks are attributed to the beating between two adjacent modes, which occurs at a frequency of 380 kHz and its multiples. As a result, the phase noise performance deteriorates significantly at frequency offsets corresponding to these beating frequencies compared to that of the commercial microwave source. A multi-loop OEO can be an effective solution for removing unwanted modes and achieving significantly reduced phase noise at the beating frequencies [27].

Table 1 shows the comparison of our work with that of the previous photonic integrated FDML OEOs. It can be seen that our FDML OEO achieves a much higher chirp rate, thanks to the fast electrical tunability of the MRR, which enables Fourier domain mode-locking with a shorter fiber delay, facilitating future hybrid integration of the system. In addition, the scanning range of our TFLN-based FDML OEO is much wider than that based on SOI. An even wider frequency tuning range can be achieved with our TFLN PIC-based OEO if microwave components, such as an MRR with a larger VPP driving voltage signal and LNAs with wider bandwidths, are available in our laboratory.Table 1.

Comparison of Our Work with the Previous Photonic Integrated FDML OEOs

An advanced FDML OEO on a TFLN platform is demonstrated, featuring an on-chip phase modulator and an electrically tuned TFLN MRR notch filter. Leveraging the high-speed Pockels effect, the TFLN-based frequency-scanning MRR filter exhibits exceptional response characteristics. This setup generates a K-band LCMW signal with a frequency range of 17.7 GHz to 26.2 GHz, adjustable sweep bandwidths from 3.85 GHz to 8.5 GHz, and a chirp rate of up to 3.22 GHz/μs, showcasing a phase noise of −106dBc/Hz at a 10 kHz offset. The modulation bandwidth of the latest vector signal generator (Keysight M9484C) can only reach up to 2.5 GHz. The integrated FDML OEO offers cost-effective, rapidly tunable, high-frequency outputs with outstanding linearity, positioning it as a transformative technology for high-resolution radar and cutting-edge wireless communication applications. In the future, hybrid integration can be implemented by wire bonding our TFLN photonic chips with electronic chips, including amplifiers and other components, to microstrip lines. This approach will reduce losses and eliminate the need for an EDFA, thereby effectively lowering power consumption.