1. Introduction

Optical parametric oscillators (OPOs) are now well-established sources of coherent electromagnetic radiation. They are highly flexible with regard to both temporal and spectral coverage; the former extending all the way from ultra-short (femtosecond) pulse [1] to continuous-wave [2] generation and extending from the ultraviolet through the visible and near infrared to the mid-infrared, and more recently to the terahertz [3] spectral region. Very narrow linewidth devices are particularly suited to high resolution spectroscopic applications [4] and this mandates the use of sources in the continuous-wave (cw) regime. The need for cw OPOs to operate outside the cosseted environment of the research laboratory precludes the use of doubly-resonant and, to a lesser extent, pump-enhanced geometries due to their susceptibility to mechanical vibration [5], [6]. Externally-pumped singly-resonant devices often exhibit an unacceptably high (in the context of battery operable, potentially man-portable devices) oscillation threshold, particularly when the power requirements of the primary pumping source are considered. These devices only exhibit good down-conversion efficiency from the pump wave to the generated signal and idler waves when pumped sufficiently far (~2.5x) above their threshold condition [7]. Even when long-interaction length nonlinear materials exhibiting high nonlinear coefficients, such as periodically-poled lithium niobate, are used, this threshold is still typically ~2-5W [4]. Therefore, whilst offering very high down-conversion efficiency when generating multi-watt output in the down-converted fields, these devices are very inefficient in the case when only comparatively low powers (~<1W) are required since this requires them to be operated close to their oscillation threshold.

In the intracavity geometry the nonlinear medium of the optical parametric oscillator is located within the cavity of the parent pump laser [2]. Because the intensity of the circulating coherent radiation within the cavity is much greater than that which can be coupled out of the cavity (typically some 10 to 20 times greater), then the parametric generation process can be brought to above threshold with a much smaller pump laser (in terms of both its power and size) compared to the case where the nonlinear medium is placed external to the pump laser. This results in highly compact and efficient devices for the generation of widely-tunable coherent radiation, in particular allowing moderate-power diode-lasers to be used as the primary excitation sources for the pump laser itself [8]. Whilst such devices exhibit high power output and excellent down-conversion efficiency in compact and robust mechanical designs, it is found to be the case that they are highly susceptible to the onset of spontaneous and very long-lived bursts of relaxation oscillations which, due to their impact on the amplitude and frequency stability of the device, essentially render the device useless in all but crude ‘mean power’ applications when operated in the cw mode [9]. One possible solution to this problem is to operate the pump laser in the Q-switched pulsed mode [10], which avoids the issue of relaxation oscillations because of the short pulse duration. An additional advantage of this configuration is that the very high peak powers circulating within the pump cavity desensitize the system from the effects of optical loss, thereby significantly extending the tuning range of the device. For example, we have recently developed compact sources based on this technique in which continuous tuning ranges in the mid-infrared extending from 1.3 to 4.7μm [11], are attained in a single device without the need to change any of the optical components. However, operating the ICOPO in such a pulse mode significantly increases the line-width of the down-converted waves. In general the time-averaged linewidth exceeds the transform limit associated with such short duration pulses, since the high peak pump powers involved allow the parametric process to operate over a very much enlarged range of its phase-matched bandwidth. In order to realize the true potential of the ICOPO technique in the context of high-resolution spectroscopic applications, then, a solution to the problem of relaxation oscillations must be found such that a return can be made to continuous-wave operation.

The pump lasers currently used in the above are neodymium lasers, with primary pumping by diode lasers. Neodymium lasers in general suffer from relaxation oscillations since their upper laser state lifetime is much longer than the typical decay times for coherent radiation within the passive optical cavity of the laser. However, such relaxation oscillations are generally heavily damped and hence can be successfully managed resulting in practical continuous-wave devices. In the case of the intracavity optical parametric oscillator two cavities are now involved, the cavity of the (neodymium) pump laser itself (referred to as the pump-wave cavity), but also the cavity associated with the resonated down-converted wave (singly-resonant, referred to as the signal-wave cavity)), which is coupled through the nonlinear medium to the pump-wave cavity. We have experimentally observed that under such circumstances both the frequency and, importantly, the damping time of the relaxation oscillations are substantially increased [9,12]. Such long damping times lead to spontaneous and long-lived bursts of relaxation oscillations (see Fig. 1 ) and hence preclude the ability to effectively manage the relaxation oscillations, essentially rendering the device useless as a spectroscopic source.

 

Fig. 1 (1.9MB) (Media 1) showing the typical free-running stability of an Nd-based ICOPO. The oscillations occurring in this clip occur spontaneously: no external triggering mechanism was employed.

Download Full Size | PDF

In this publication we identify and report a solution to the problem of relaxation oscillations based on introducing a further nonlinear optical process, namely second harmonic generation (SHG), within the cavity of the pump laser. The introduction of this element severely reduces the damping time of the oscillations to the point that it is less than even the damping time exhibited by the parent pump laser operating alone, see Fig. 2(c) . The suppression system is entirely passive, and so simple to implement, and results in very little reduction in the power of the down-converted waves. In order to identify the parameter space within which suppression of relaxation oscillations occurs, we have developed a theoretical model based on laser rate equations. In the next section (section 2) we describe this model and its application in some detail so as to provide a guide for the experimental studies that are then described in sections 3 and 4. In section 5 we return again to the rate equations and develop a perturbation approach that returns both the frequency and damping time of relaxation oscillations as a function of system parameters through the solution of a cubic equation.

 

Fig. 2 Transient behaviour of the circulating pump field (P_p) as predicted by the theoretical modeling after perturbation for (a) the pump laser alone, (b) the ICOPO with no SHG suppression (i.e. δ = 0) and (c) the ICOPO with SHG suppression enabled (δ = 0.01). Note that, due to their high frequency, individual oscillations in Figs. 2(b) and (c) are not resolvable. Stated times along the horizontal axis are normalized to the upper state lifetime (in this case 100μs)

Download Full Size | PDF

In a recent publication [13] we have demonstrated that by going to a case where the effective lifetime of the population inversion associated with the gain medium of the pump laser, in this case a VECSEL, is of the order of or less than the decay times for coherent radiation within the passive optical cavities (for both the pump wave and the resonant down-converted wave), then relaxation oscillations are avoided, including in the case of the intra-cavity OPO. The current solution based on the inclusion of up-conversion is a valuable complementary approach for suppressing relaxation oscillations where the above conditions cannot be fulfilled; this is particularly so in the case of the widely-used, highly-developed, and well-established Neodymium laser as considered here.

2. Theoretical Modeling of Relaxation-Oscillations in the IC-OPO and their Suppression

In order to explore the parameter space associated with the susceptibility of the ICOPO to spontaneous and long-lived bursts of relaxation oscillations, and to be able to quantify the effect of SHG within the cavity of the pump laser as a means for their suppression, a rate equation model was developed and solved both numerically, and analytically through a perturbation analysis. The model comprises three rate equations; one describing the rate of change of the population of the upper laser state (n_j) of the gain medium (where it is assumed that no significant population builds up in the lower laser state), and the other two describing the rate of change in the power circulating within the pump (P_p) and signal (P_s) cavities. These equations are:

The parameter k is such that (1 + k) is the number of times above the oscillation threshold of the optical parametric oscillator the device is being operated, while σ_j is the number of times above the oscillation threshold of the pump laser the device is being operated. It may readily be shown that under steady-state operating conditions, optimum down-conversion occurs when (1 + k) ² = σ_j. Following our inclusion of the up-conversion term into the equations, as displayed above, we have proceeded to obtain numerical solutions to these equations as a function of time by using computational algebra programmes such as Mathematica.

Figure 2 shows numerical solutions based on the above equations under operating conditions appropriate to the experimental results discussed in sections 4 and 5; where we have: σ_j = 10, k = 4, τ_u = 100μs, τ_p = 10ns and τ_s = 60ns. Figure 2(a) is for the case where intracavity parametric down-conversion is suppressed and shows the relaxation oscillation behaviour typical of a Nd-laser. We see that in this case the oscillations have a frequency of ~330kHz and damp away in approximately 50μs (half the spontaneous-emission lifetime of the upper laser state). In general the relaxation oscillation frequency is determined by the inverse of the geometrical mean of the upper-state lifetime multiplied by the radiative decay time of the passive pump-wave cavity, while the damping time is of the order of the upper-state lifetime. Figure 2(b) is where intracavity parametric down-conversion is allowed to take place and where the extreme relaxation oscillation behaviour seen experimentally is clearly predicted by the model. Here we see that the frequency of the relaxation oscillations is now around 10MHz, and the damping time has increased substantially to ~1ms. In general for the case where the two cavities involved have similar radiative decay times, then the oscillation frequency is determined by the inverse of this common radiative decay time, while the damping time is of the order of the upper-state lifetime but now multiplied by the ratio of the upper state lifetime divided by the radiative decay time of the cavities. This latter explains the exceptionally long decay time of the relaxation oscillations. Figure 2(c) is for the case where up-conversion of the intracavity pump field is now introduced (δ = 0.01) and clearly illustrates how this process is highly effective in the suppression of the previously observed relaxation oscillations, which now decay in only ~10μs; substantially quicker even than the case of the parent pump laser operating in the absence of parametric down conversion.

3. Experimental Arrangement

In order to evaluate the effect of the SHG upon the transient dynamic behavior of the ICOPO, a system was set up as in Fig. 3 and is similar to our previously reported device [11]. The system is primary-pumped by a thermo-electrically temperature stabilised c-package 3W CW laser diode LD with a 200 × 1μm emission area. The wavelength output of the diode was temperature tuned in order to closely match into the peak absorption feature of the Nd:YVO₄, corresponding to a wavelength of 808.5μm. The coupling optics, which were antireflection coated at the diode pump wavelength, image the diode emission facet into the rear face of the Nd:YVO₄ crystal. The external face of the Nd:YVO₄ crystal was coated to be highly reflecting for the 1.064μm circulating pump field, and therefore defined one end of the laser cavity. This coating was also anti-reflecting for the diode pump radiation, thereby allowing transmission of the incident pumping radiation into the crystal. The c-cut, 3 × 3 × 1mm Nd:YVO₄ crystal was doped 1% at. Nd. Its intracavity facet was antireflection coated for the 1.064μm circulating field. When pumped at 3W, the mirror coated upon the Nd:YVO₄ crystal had a ~75-100mm thermally-induced radius of curvature.

 

Fig. 3 Experimental configuration of the ICOPO with SHG relaxation oscillation suppression.

Download Full Size | PDF

In order to form an appropriate pump field focus in the centre of the parametric down-conversion nonlinear crystal, an intracavity lens (antireflection coated at 1.064μm, 38mm focal length) is included in the cavity. The lens also served as ‘optical ballast’ within the pump cavity, desensitizing it to the effects of thermally-induced lensing in the Nd:YVO₄ and, more significantly, the nonlinear crystal used for down-conversion. An MgO-doped, periodically-poled LiNbO₃ (PPLN) crystal which had dimensions 1 × 6 × 30mm³ and which was triple-band anti-reflection coated for the pump, signal and idler wavelengths, was used for down-conversion. The inclusion of the MgO doping meant that it was unnecessary to heat the PPLN crystal to elevated temperatures in order to obviate the deleterious effects of photo-refractive damage. However, the crystal temperature was held at at 25°C to help stabilise the wavelengths of the down-converted signal and idler waves. The crystal had a single grating period of 30.7μm, resulting in signal and idler wavelengths of 1.56 and 3.35μm respectively. The high-finesse pump cavity is completed by the CaF₂ pump and signal mirror (30mm radius of curvature), through which the useful mid-infrared idler field is extracted. This mirror was coated to be highly reflecting at both the pump and signal wavelengths and anti reflecting at the idler wavelength.

In order to discriminate the signal cavity from that of the pump, a dichroic beamsplitter BS is placed between the intracavity lens and the nonlinear crystal. The left-hand face of this is antireflection coated for the transmitted pump field, while the right hand face is antireflection coated for the pump and broadband highly reflective coated for the incident signal field. The signal cavity is thus defined by the common pump & signal mirror, the dichroic beamsplitter BS and the signal mirror (radius of curvature 75mm), all of which are coated to be broad-band highly reflecting over the signal wavelength tuning range. The separation of the two signal mirrors was chosen such that a near concentric cavity was formed for the signal field in order to produce the appropriate signal waist of ~120μm diameter in the centre of the nonlinear crystal coincident with that of the pump waist.

An antireflection-coated (for the pump field and its second-harmonic), 2 × 2 × 20mm³ lithium triborate (LBO) second-harmonic generation crystal was placed in the cavity between the Nd:YVO₄ crystal and the intracavity lens. This component served to suppress the relaxation oscillations inherent in the device once parametric oscillation occurs. The LBO crystal was placed within a temperature stabilized oven in order to maintain it at the correct temperature for phase matching. The pump mode was weakly focused in this part of the cavity with a beam diameter of ~200μm at the centre of the LBO crystal. We chose LBO as the doubling crystal (as opposed to, say, KTP) as it exhibited type I phase-matching and therefore did not introduce birefringence into the pumping cavity. Such birefringence can perturb the polarization of the circulating pump field and cause the device performance to degrade. The efficiency of the SHG, and its impact on the transient stability of the system, could be varied simply by control of the crystal temperature without changing the round-trip linear loss of the pump cavity. Finally, the large aperture (2 × 2mm²) made it easy to accommodate the near-collimated pump mode without the risk of mode aperturing.

In order to evaluate the efficacy of the relaxation oscillation suppression through the use of intracavity SHG, in was necessary to predictably induce relaxation oscillations into the system. This was achieved by periodically perturbing the steady-state of the system either by rapid rising/falling edge modulation of the drive current to the diode-laser used for pumping or via perturbation of the pump-cavity photon lifetime by means of an acousto-optic modulator (AOM) included within the pump-only section of the system. In the latter case, the loss of the pump cavity was momentarily modulated to the extent that the OPO was reduced from operating at ~2.5 to ~1.5 times its oscillation threshold. The frequency of the perturbation was 1kHz with a duty cycle of 90%. Inducing perturbations into the system through the use of an AOM is preferable over chopping the external pumping radiation as it leaves the thermally-induced lens within the Nd:YVO₄ crystal unchanged. Variations in this latter parameter is undesirable as it can destabilize or at least modify the stability of the pump cavity, thereby changing the threshold conditions of the pump laser and OPO. In order to minimize this effect in the case of external modulation of the diode pump power, the drive current was modulated on a 95% duty cycle; the mean power delivered to the Nd:YVO₄ crystal thereby changing very little.

With the modulation schemes unused, laser and OPO threshold were reached for a diode pump power of 350mW and 650mW respectively. At OPO threshold, a circulating pump field of 4.7W was measured within the cavity. At a diode-laser pump power of 3W, 90mW of useful idler radiation was extracted from the device, indicating a total down-converted power from the circulating 1.064μm pump-wave to the resonant signal and non-resonant idler waves of 550mW.

4. Results

4.1 SHG impact on amplitude stability

The transient behavior of the device was evaluated under different operating regimes (a) Nd laser only operating, (b) ICOPO and (c) ICOPO and SHG, as shown in Fig. 4 . In this sequence, the system was driven from its steady state into relaxation oscillations by modulation of the diode drive current as outlined above. By de-tuning the LBO crystal temperature away from its phase matching condition and blocking the outboard signal mirror it is possible to suppress the operation of both the parametric down-conversion and frequency up-conversion whilst leaving the operation of the parent Nd pumping unchanged. The response of the system in this case is indicated in Fig. 4(a), where we see classic Nd-laser relaxation oscillation behavior with a frequency of ~350kHz and a rapid decay of the induced oscillations in less than an upper-state lifetime. In the absence of any further external perturbation, the laser displayed no tendency to spontaneous oscillatory behavior and remained in its steady-state indefinitely.

 

Fig. 4 Experimental measurements of the transient behavior of the ICOPO with (a) down-conversion suppressed (i.e. laser operation only), (b) down-conversion permitted (with no SHG) and (c) both down-conversion and SHG enabled.

Download Full Size | PDF

Parametric down-conversion was then restored by removing the shutter placed in between the pump/signal beamsplitter and the outboard signal mirror; the resulting dynamic behavior is shown in Fig. 4(b). The relaxation oscillation frequency is now increased to 3-4MHz; almost an order of magnitude greater than that of the Nd laser operating alone, and in line with being determined largely by the radiative decay times of the pump and signal cavities as discussed previously (see section 3). The damping time also increased considerably to well beyond the upper state lifetime associated with the Nd laser. A precise figure of the oscillation frequency and, particularly, the damping time of the system when the ICOPO was operating is not possible due to the highly erratic nature of the oscillations. Note, however, that an approximate damping time of 200-300μs (as shown in Fig. 4(b)) is significantly shorter than the ~1ms predicted by our numerical model (see section 3). We speculate that this may be due to the parasitic yet significant pump-wave SHG exhibited by the PPLN crystal itself. A damping time of this order would indicate that the PPLN SHG effect upon the dynamics of the system is equivalent to a value for δ of 0.0001 (this point is further discussed in section 5).

With the diode-laser modulation switched off and hence the intracavity pump field subject to no external perturbations, the system was characterized by periods of stable operation broken by the spontaneous and variable bursts of prolonged oscillations characteristic of Nd-based ICOPOs [9,14]

Finally, the temperature of the LBO SHG crystal was adjusted for optimal phase-matching. The (expected) dramatic effect this has upon the damping time of the perturbed system is clearly shown in Fig. 4(c). In line with expectations the damping time of the system is very much reduced to the point that the oscillations decay away in less than half of the time of the Nd laser operating without parametric down-conversion (as seen in Fig. 4(a)). Note however that the oscillation frequency in this case is still determined by the photon lifetime of the pump and signal cavities and as such remains largely unchanged from the case of Fig. 4(b), again in line with expectations. When the inducing perturbation was removed the system was characterized by robust and essentially indefinite operation in the steady state. The power performance of the device both with and without SHG oscillation suppression will be addressed in section 4.2 below.

The impact of up-conversion on the stability of the system can be readily seen in real time in the video clip which appears as Fig. 5 . Here, the time base of the oscilloscope is slowed to 1ms/division in order to show the quasi-chaotic nature of the oscillations. The upper trace shows the triggering signal from which the AOM RF amplitude is derived. The lower trace, zeroed at the bottom of the scope screen, shows the signal field leaking from the outboard mirror. It is clear from the left hand side of the oscilloscope trace that even when the loss introduced by the AOM is present, the OPO is still above threshold. Mild Q-switching behavior is evident upon the rising edge of the triggering signal (i.e. when the AOM-induced loss is switched off). This is due to the population inversion within the Nd:YVO₄ crystal saturating at a slightly higher level when the modulator is in its high-loss state. Upon removal of the extra loss the circulating pump-field then instantaneously experiences a gain that exceeds the (now reduced) loss leading to the formation of a weak Q-switched pulse.

 

Fig. 5 (1.9MB) (Media 2) showing the impact of SHG on the stability of the ICOPO

Download Full Size | PDF

At the beginning of the movie, the LBO crystal is held at a temperature where SHG phase matching does not occur, and so up-conversion does not contribute to the dynamics of the system. The relaxation oscillations triggered by modulation of cavity loss by the AOM are clearly visible in the lower trace. It is also evident that oscillations spontaneously occur even after previously induced bursts have decayed. The amplitude stability of the system even in the absence of external perturbation is very similar to that seen in the lower trace. After 15 seconds of the movie, the temperature set-point of the LBO oven is adjusted for optimum phase matching. Due to the thermal mass of the oven insert in which the LBO crystal is housed, it then takes ~30 seconds for the oven temperature to change and stabilize to the correct level. As the temperature is changing, the impact of the varying levels of up-conversion on the stability of the system can be seen. Finally, as the LBO oven temperature settles and SHG is optimized (50 seconds into the movie), the relaxation oscillations are now absent from the system.

An RF spectrum analyzer was employed in order to provide a highly sensitive and quantified measurement of the stability of the system in the presence of up-conversion. The resulting traces are shown Figs. 6(a) and (b) , where the device was allowed to free-run without inducing perturbations either through the diode pump current or AOM pump cavity loss modulation. In Fig. 6(a) the spectrum of the amplitude fluctuations of the circulating signal field is shown without the LBO temperature set for SHG phase matching, and is measured during a relaxation-oscillation event. As the oscillation bursts are unpredictable in as much as they are of random in length and their fundamental frequency varies significantly throughout the burst, the trace shown in Fig. 6(a) is representative of a local oscillation event rather than the global behavior of the system. The non-sinusoidal yet periodic nature gives rise to the comb-like trace, the first harmonic component of which is at the fundamental oscillation frequency (in this case, ~4MHz) followed by its higher harmonic components. The stability behavior depicted in Fig. 6(a) is in sharp contrast to that of Fig. 6(b), where the LBO crystal temperature has now been optimized for SHG. This trace exhibits none of the frequency components associated with oscillatory behavior. The noise floor for the detection system used in this measurement is at −70dB. The presence of low frequency amplitude fluctuations within the system brought about by mechanical and acoustic perturbation manifests itself by a rise of the curve at low (kHz) frequencies. This could be substantially improved by utilizing superior mechanical design techniques such as more robust mounting of optical components and tubing off the circulating pump and signal fields within the cavity.

 

Fig. 6 Amplitude spectrum of ICOPO in (a) the absence and (b) the presence of second-harmonic generation.

Download Full Size | PDF

4.2 SHG impact on down-converted power

In order to assess the impact of SHG on the power operation of the OPOs, the extracted idler output power and, hence, the total down-converted power of the ICOPO was measured as a function of diode laser primary pumping power both in the presence and absence of SHG; the results of which are shown in Fig. 7 . In the case of the ICOPO operating in the absence of SHG, the output power exhibited significant (10%) fluctuation on the ~1-10s timescale, coincident with bursts of relaxation oscillations. The measured power indicated in Fig. 7 is for the case where averaging over >10s has been applied. When SHG oscillation suppression was enabled, the fluctuation in output power was significantly reduced to ~3%, indicating the enhanced stability of the device operating in this regime.

 

Fig. 7 SHG impact upon down-converted power.

Download Full Size | PDF

It is evident from Fig. 7 that operating an ICOPO with intracavity SHG results in very little loss in the down-conversion efficiency. In our case, implementing the SHG resulted in a ~3% drop in down-converted power, in reasonable agreement with the loss predicted by the numerical model as discussed in section 5 below.

Once above threshold, the OPO ‘clamps’ the intensity of the circulating pump-field at the value corresponding to this threshold, and so the production of second-harmonic light does not significantly increase as the primary pumping power is increased. The combination of the pump-field clamping at this relatively low level and the weak focusing within and the low nonlinearity of the LBO crystal produces relatively low levels of green light production on the order of 10’s mW. As is evident from the results presented in section 4.1, however, this low-level production is sufficient to eliminate the spontaneous onset of relaxation oscillations. A further increase in the power of the generated second-harmonic radiation would improve the stability of the system further at the expense of greater loss in down-converted power. However, our model predicts that even when very high pump-wave damping times are brought about by increasing the value of δ only a modest price is paid in terms of the down-conversion efficiency of the device.

5. Perturbation Solutions to Rate Equations & their Application

Further useful insights may be gained by carrying out a (small) perturbation analysis of the three coupled equations. In such an analysis we assume that the pump and idler intracavity field powers (P_p and P_s) and the population inversion (n_j) are perturbed by amounts that are small compared to their steady-state values, and then retain only terms linear in these perturbations. This procedure allows the three equations to be decoupled, and hence allows solutions to be sought for each of the three perturbed quantities independently. Trial solutions are then sought and successfully obtained in the form of damped simple-harmonic-oscillations, namely:

It is also important to know what impact a particular level of up-conversion has on the down-conversion. It may readily be shown that in the perturbation approximation the steady-state intracavity signal field, which is normalized to unity in the absence of up-conversion, reduces to:

Solving the above cubic equation for parameters appropriate to the conditions of the experimental investigation, and which were used in the previous numerical modeling (σ_j = 10, k = 4, τ_u = 100μs, τ_p = 10ns, τ_s = 60ns, and hence γ = 6, and η = 10⁴) but with no up-conversion present (δ = 0), we obtain an oscillation frequency of 6.5MHz, with a 1/e damping time of 8.4ms. It should be noted that while these values are reasonably consistent with our previous modeling (see section 2), exact agreement is not to be expected since the current approach is based on small perturbations from the steady state, whereas the previous modeling (section 2) was carried out for conditions encountered in the experimental studies where perturbations of the order of +/−60% of steady state values are involved . If we now introduce an up-conversion parameter δ equal to 0.01, then the frequency is predicted by the above cubic equation to be largely unchanged, but the damping time is now reduced to ~2μs. Such an up-conversion would result in a predicted loss in the down-conversion (signal) power of only 1%. Note also that with an up-conversion parameter equal to as little as 0.0001, the current model also predicts that the damping time becomes ~200μs, in line with previous experimental observations.

5. Conclusions

We have shown that the nonlinear loss introduced into the cavity of the circulating pump field by means of a second-harmonic generation crystal represents a simple and highly effective means of eliminating the problem of relaxation oscillations associated with continuous-wave, intracavity, optical parametric oscillators based on neodymium pump lasers. Such a technique should be applicable in general to other pump lasers where severe relaxation oscillations are experienced due to the lifetime of the upper state of the laser transition being long compared to the radiative decay times associated with the (passive) cavities of the pump field and the resonant down-converted field.