Reviewer Recommendation: Accept with minor revisions

Summary:

The submitted manuscript describes a system used to stabilize the output frequency of a high powered Nd:YAG laser. The laser design and the system are described in detail as well as the testing done on the system to ensure functionality and long term stability.

In my opinion, the submitted manuscript is very well written from a technical perspective. I also believe the manuscript is within the scope of AMT. However, given how the manuscript is written, I personally would have submitted it to a more optics/laser focused journal. The sourcing is lean but I think sufficient as well. I have no major issues with the technical detail from manuscript as written and would suggest it be published subject to some very minor revisions that I think would clarify a few confusing details.

One issue worth mentioning in a general comment is that while I think the manuscript is within the scope of AMT, the reader needs to bring motivation and knowledge of the atmosphere with them. As I read this manuscript, the laser is clearly targeted at a scientific application (Doppler wind measurements) but the description seems to have almost surgically removed the application. For example, while NLCs and winds are mentioned in the intro in a general way, it seems odd to me to have only mentioned wind accuracy in the conclusion.  From an engineering perspective, the content is great. From an atmospheric science perspective, several things remain unclear to me: does this locking method improve your measurements noticeably or are other sources of error (for example shot noise) dominant? What was the accuracy you achieve before this system?  Does this system enable scientific exploration that is otherwise inaccessible? These seem to be the most relevant questions to me as an AMT reader. 

Major Comments:

1. None

Minor Comments:

1. The title refers to ALOMAR but the manuscript mentions the Kühlungsborn RMR lidar as well. I might consider a more generic name if there are 3 such systems not all running at ALOMAR.
2. In my opinion, while it is true that strict stabilization of the laser frequency is required for wind measurements, the manuscript hardly mentions it. This is why I comment that I would have targeted a more optics focused journal. In response, I would suggest:
   1. Link the frequency accuracy to your wind measurement accuracy in the abstract.
   2. Comment on how much of your error budget for winds is occupied by frequency stability. Here I simply mean, if you have improved your wind measurements markedly, show that. If your main error source is shot noise and the stability is contributing less to that budget, it would be good to know.
   3. Describe the level of improvement this locking system provides over your previous system.  
   4. With an accuracy of ~11 cm/s, can you describe what scientific questions can be addressed that were not addressable with a previous stabilization system?
3. Line 28: Is 4 picometers approximately equivalent to 4 GHz? Not 4 MHz?
4. Line 96-97: In the text, you specify 2 hours to reach stability but only show 1 hour. I would tend to show 2 hours of data in Figure 2 to show how stable looks. For example, I wonder if the voltage jumps become less frequent of disappear completely.
5. Line 113: It is not clear to me why a diffuser is helpful here. I would think increasing the angular distribution of light would be generally detrimental to etalon performance. Can you please comment?
6. Line 119: I am a bit confused here. I can take this statement to mean either that you read data from 1 intra-pulse period every second from the seed laser or that you read 100 and average/sum them into 1 data point. Can you clarify?
7. Line 126-127: I presume given that Figure 5 and 7 use fractional pixels that you are fitting a curve to the data. Is that true? If you are fitting something, can you say what that is? If you are just using the raw data, can you specify how FWHM is calculated and how you get to fractional pixels?
8. Line 133: This statement is a bit vague in my opinion. Adding parallelization ad infinitum will slow down your code eventually with communication (not calculation) being the bottleneck. Is this statement just meant to say that you are doing calculations on an FPGA in a parallel way or are you talking about multi-threading in your software?
9. Figure 4 Inset: I am not clear why you would use time as the x-axis. Wavelength makes more sense to me.
10. Line 157: I would modify this statement to “…light enters the LPS continuously”.
11. Line 169: 10% of data seems rather large to leave without a physical explanation for lack of correlation. Are there any physical processes that could explain this data?
12. Figure 5: I would change your colors to accommodate red/green colorblind readers.
13. Line 222: Do you mean resonance-fluorescence lidars? My understanding of fluorescence lidar is that the spectral features are relatively broad (order 1-10 nm at least). It is not clear to me that any amount of precise laser locking should affect that.

Summary

In the paper „Frequency control and monitoring of the ALOMAR RMR lidar’s pulsed high-power Nd:YAG lasers“ the authors give a technical overview on the build-up-time resonator stabilization and real time frequency monitoring of their lidar‘s Nd:YAG laser. The frequency stabilization of the injection seeder by iodine spectroscopy is described in another paper (line 54), therefore the title might be considered misleading, because the part of frequency control is described in another paper. The main scope of the paper is performance monitoring of the injection seeder build up time resonator stabilization. Of course, injection seeding can be seen as “frequency control”, but the resonator control mostly controls, which adjacent longitudinal modes to the injection seeder start to oscillate and how fast they start (build up time). Additionally, the Piezo shifts the whole longitudinal mode spectrum. So dependent on how good the resonator length is stabilized to the injection seeder, the bandwidth of the pulsed laser is reduced to nearly single longitudinal mode operation and low frequency offset to the injection seeder.

For the application the bandwidth and the relative frequency control is crucial, but bandwidth is only discussed in figure 7. For a full analysis of the presented system tuning the piezo of the laser over a full spectral range (½ µm) would have been interesting.

Furthermore, clarification about the setup is needed, since not all components mentioned in the publication are depicted in the setup.

Therefore, I advise major revisions.

Comments on the individual sections:

Section 2: System setup

• In figure 1 components are missing (e.g. diffuser/lens/fiber couplers). The components should be described in this section.

Section 3: Pulsed Nd:YAG laser frequency control

• In line 76 the authors write: “When the seed laser radiation is within the bandwidth of a longitudinal mode of the power laser resonator, it is resonantly amplified in the Nd:YAG rod. “
  
  
  
  Is the true? When the Q-switch is operated only one round trip of seed laser radiation will be present in the resonator which is specially broadened due to the Q-switch So the adjacent longitudinal modes will always receive a higher start energy than other longitudinal modes (in classical laser theory they start for 0.5 photons :) ). I’ll add some measurements in the appendix. As the resonator mode is better matched to the seed laser wavelength and mode, other modes get suppressed more until the gain of the laser is used up (works with homogeneous broadened gain media).

• In Figure 2 the authors depict the build-up-time and the piezo control voltage after switching the laser on. The histogram of the laser build-up-time is shown but nowhere discussed. An evaluation is difficult because the build-up time (minimum) is not only dependent on the resonator length control but also on the gain in the laser itself. And the gain might be dependent on the temperature of the pump diodes. For this a synchronous plot of the output energy could explain the 2 ns increase in minimal build up time at 8:20.

Section 4: Laser pulse spectrometer

• The authors should motivate their design considerations and add a literature study on the topic.
  
  E.g. a fast real time wavemeter for injection seeded lasers using a fabry-perot as pulsed wavemeter was presented in 1993 by Hahn et. al. „Fabry–Perot wavemeter for shot-by-shot analysis of pulsed lasers“ DOI:10.1364/AO.32.001095. or ”A simple real-time wavemeter for pulsed lasers”, Ja-Yong Koo and I Akamatsu DOI 10.1088/0957-0233/2/1/009.
• Why did they choose a camera instead of a line camera? The camera produces more data, which might be redundant and is therefore more demanding on the signal processing.
• Why an Etalon and not a Fizeau interferometer was chosen for the Laser pulse spectrometer?
  
  E.g. in „An absolute frequency reference unit for space borne spectroscopy “, by H. Schäfer et. al. DOI: 10.1117/12.2536012 a fiber coupled wavemeter using a collimator and a Fizeau wedge is presented to compare the wavelength of the injection seeder with the injection seeded pulse from an OPO – they even omitted means of chopping out the cw signal of the injection seeder due to different integration times.

Further literature to be considered:

• Fizeau wavemeter for pulsed laser wavelength measurement, Mark B Morris, Thomas J. McIlrath, and James J. Snyder, https://doi.org/10.1364/AO.23.003862

• Low-cost wavemeter with a solid Fizeau interferometer and fiber-optic input, Benedikt Faust and Lennart Klynning, https://doi.org/10.1364/AO.30.005254

• A simple real-time wavemeter for pulsed lasers, Ja-Yong Koo and I Akamatsu, https://doi.org/10.1088/0957-0233/2/1/009

Further minor questions the authors should consider:

• Why a FSR of 1 GHz was chosen for the Etalon (line 107)?
• What was the reflectivity of the mirrors of the Etalon?
• How is the seed laser light coupled out of the fiber (line 111) – is there a collimator used?
• What is the beam size of the laser and the seed laser on the diffuser?
• Where is the ground glass diffuser (line 113) depicted in Figure 1?
• Where is the 500mm lens depicted in Figure 1?
• Is the chopper necessary? (E.g. Schäfer et. al. did not need this)
• In line 113 the authors describe the lens is imaging the interference patterns. Actually, it transforms the angular interference pattern into a spacial interference pattern. An angle of 1mrad is transformed into a displacement of 0.5mm. So, 0.8 mrad would be 0.4mm ~57-66 pixels for a 6-7 µm camera pixel pitch.
• A line camera would not require such “overkill” hardware like a FPGA running RT-Linux and resource hungry LabVIEW… of course a small FPGA could evaluate a line camera in real time with a defined latency. The authors should focus on describing the technical necessities or concepts and then the details of their implementation (as implementations might changes – but keeping in mind the review criteria:
  
  
  
  Scientific significance:
  
  Does the manuscript represent a substantial contribution to scientific progress within the scope of Geoscientific Instrumentation, Methods and Data Systems (substantial new concepts, ideas, methods, or data)?.
  
  E.g. in line 126 the authors write: “Then, for each peak of this function the position and amplitude of the maximum as well as the full width at half maximum (FWHM) are determined. “. The accuracy of the frequency estimation is dependent on the FWHM and the SNR of the measurement – the authors should give a Cramer-Rao limit for their estimator and compare their observations with this.

4.1 Calibration

The method is well described.

4.2 Sensitivity

• The authors should state which slope is used for stabilization of the seed-laser. I suppose the increasing slope in the figure 4 (UT 12:16) is used. Here the type of evaluation (fitting/center of gravity measurement etc...) is important, since a sub pixel evaluation is performed and plotted in figure 5.  The chosen criteria in lines 165 and following do not express anything about thermal drift within the Etalon. There is no place where the authors discuss the sensitivity of the laser pulse spectrometer on pressure and temperature. Because it is used for relative measurements, this does not matter too much. 
  
  But for the sensitivity the bandwidth of the laser might be important, the bandwidth of the cw seed laser is small compared to the bandwidth of the 10-12ns pulses q-switched laser. The bandwidth of the q-switched laser with 10-12ns is 50-100 Mhz (dependent on the time bandwidth product of the pulse). With a Finesse of 20 and an FSR of 1 GHz the Airy linewidth of the interference pattern is approximately 50 MHz, therefore the expected peak FWHM of the lase pulse including the instrument function of the Etalon would be 2-3 times higher than the   peak FWHM of the cw-laser. This would decrease the ‘sensitivity’.

5 FCaM Performance

• In line 180 the authors note:”The individual measurements of the power laser frequency stability reproduce the sinusoidal variation of the cavity length nearly perfectly.” and in line 209 the authors write: “The imprinted cavity length modulation for the BuT minimization method results in approx. ±10 MHz frequency modulation around the mean frequency of the power laser, which potentially can be reduced.”
  
  The authors should discuss why they do not consider using the measured frequency offset with the spectrometer for cavity control. E.g.  in DOI: 10.1117/12.2536012 this is the proposed way for a space bourne system. It is a much cleaner signal which is not subject to build-up-time jitter (due to residual inversion etc..).

• In lines 189 ff. the authors begin to speculate about a single event – whether the numerical evaluation worked properly cannot be determined without the raw data of the event.
  
  The sentence: “Destructive interference of adjacent longitudinal modes (mode beating) could explain the observed reduced pulse intensity but should result into a spectral broadening instead of narrowing. In the end, it is unclear which process led to the observed behavior.” could/should be verified by opening the control loop of the laser and scanning the Piezo over a full spectral range (532 nm). With a frequency deviation of 40MHz~1/4 FSR detuning I would expect mode beating between two adjacent modes – furthermore I would expect a lower energy/intensity in the main peak. How stable is the FWHM fit when the intensity is reduced and a second peak appears 160 Mhz away from the main peak? - The peak should be still separated but close to each other due to the finesse and spectral width of the pulse.

Therefore, I would advise the characterization of the measurement setup for all possible detunings of the laser cavity.

Appendix: Unpublished measurements of a laser with the legendary Lightwave Electronics 101 injection seed laser (measured 2005)

Influence of the laser resonator length on the impulse form and spectrum (Image in the supplement.pdf)

The left figure shows the temporal intensity of q-switched pulses of an injection seeded laser with a resonator length of approximately 80 cm. The pulses were measured using a photodetector with ~2GHz bandwidth. The cavity length is controlled using a Piezo translator where the cavity end mirror is mounted on. Additionally, an unseeded laser pulse is generated by blocking the seed laser.

In the middle the Fourier transform power spectra of the measured pulses is displayed. The laser pulse with the lowest build up time shows the least mode beating. As the cavity is more detuned higher order mode beating appears.

In the right figure the measured laser build-up-time (the rate is approximately 0.05 V/ns) by the injection seeder electronics for different cavity detuning is shown. The measured pulses are indicated by horizontal lines.