The CPS is a sensor that is clearly needed for makinng profiles in clouds with upsonds, dropsonds or even teathered balloons. The original description by Fujiwara et al. was a good introduction but left out four very important details that also limit the usefullness of the present paper. 1) What is the intensity profile of the lasr beam across the sample areas of the two detection systems, 2) What is the shape of the elaser beam, 3) What are the actual dimensions of the two parallelagrams and 4) For the collection geometry what do the Mie scattering cross sections look like for water droplets.

Lacking this information in the present paper, all of the corrections that are related to the pulse width and the sizing are irrelevant since they all assume a beam shape and laser intensity that is uniform, assumptions that are likely not the case.

I am puzzled that readily available software like Zmax was not employed to model the actual optical system. The figures in 3a and b are speculative, in the words of the author. Speculation has no place in a technical article.

The size calibration is based on water equivalent sizes of crown glass and PSL particles, but these water equivalent sizes have to come from theoretical considerations. Nowhere is this descriibed.

The cloud physics community that uses optical spectrometers are now using precise droplet generators to map the sample areas of spectrometers similar to the CPS. This needs to be done for the CPS if this technology is to be accepted and inversions have to be applied as there is no qualifier in the system to constrain the particles through the sample area. I recommend that the authors read the papers related to the IAGOS Backscatter Cloud Probe.

Finally, trying to model the flow through the CPS with no measurement validation is unconvincing. The much siimpler and more convincing approach is to do the measurements in a low speed wind tunnel that are em[ployed around the world to calibrate anemometers.

From a presentation perspective of the material, once the study is repeated more vigorously, this is an AMT paper so most of the introduction is irrelevant except for the last paragraph that describes the objectives. The title is misleading and needs to be more explicit and the photos, while pretty, are also irrelevant to the topic.

Finally, comparisons in the field are irrelevant until the corrections are justified properly and in additions the OPCs against which they are compared have their own uncertainties that have to be explain to put the comparisons into context.

The authors present a new instrument, a Cloud Particle Sensor (CPS), that can be used for vertical probing of cloud particle number concentration, phase and size from radiosondes or similar platforms. The technical details of the instrument were introduced in a paper by Fujiwara et al. (2016) and in the current paper, the authors present field data taken from 40 CPS sonde launches in the Arctic region. While development of reliable and inexpensive in-situ measurement techniques for tethered balloons and radiosondes is important to complement aircraft and remote sensing measurements, which are restricted in time and space, the proposed method to correct for CPS raw data to get scientifically meaningful quantitative values for total concentration and liquid water content (LWC) is not convincing. The sensitive area of CPS is not characterised either using theoretical calculations nor using droplet injector mapping. The assumption to use cross-sectional area of the CPS inlet as sensitive area is wrong and the applied correction factor to account for flow dynamics is not convincing. Without proper characterisation of the sensing area, the presented results cannot be recommended for publication.

Major comments

1. In order to derive particle concentrations, the information of the sensitive area of the instrument is crucial. The sensitive area is defined as the overlap between the laser beam profile and the detector filed-of-views (FOV). However, no information is given for their geometry.  Instead, the authors claim that the “detection domain” is 1x1x0.5 cm based on the dimensions of the slits and in the derivation of total concentration the counts are divided by the cross-sectional area of the CPS inlet (1 cm2), which is not the same as the sensitive area.

2. Since the sensitive area is not known, estimation of total concentration is not possible.

3. The analysis is based on the assumption that the particles have a constant PSW. In order this to be true, the laser profile should be uniform in the detection area and the flow of the particles should be constant. Why no technical efforts were made to fulfil these conditions  (e.g. beam shaping, focusing air flow, etc.)?

4. The size calibration is based on standard particles. Although the differences in refractive index is taken into account, the authors could have repeat the calibration using water droplets distributed with piezo-injector, which is the common practise for cloud instruments. Same method could have been used to map the sensing area.

5. According to flow calculations, the flow speed at the inlet is reduced by 17.2% due to the instrument housing. At the same time air pressure increases. After reaching the instrument, the flow speed accelerates to a value close to the flow speed before the instrument. The authors interpret these calculations so that “the chance that the air mass can enter the CPS inlet in a unit of time is reduced to 17.2%” and calculate a correction factor for the total counts to be 5.8 (= 1/0.172). However, I consider this reasoning to be incorrect. 

Minor comments

p.2, lines 49-52: Cloud phase can be determined in-situ using number of different methods but the authors only mention Cloud Particle Imager. I would suggest referring to paper by Baumgardner et al., 2017.

p.4, line 93: The terms “particle signal width” (PSW) and particle transit time are used. Why not to use the term particle time-of-flight (TOF) that is frequently used in the community?

p.5, Section 2.5: No explanation is given for the chosen DOP threshold. Additionally, the separation is based on particle sphericity rather than actual ice/water phase. This should be mentioned.

Fig. 6: I don’t see that accumulated relative PSW frequency is a good way to illustrate the PSW distribution. Why not to show normalised PSW frequency? Why is the data limited to water cases (DOP>0.5)?

Fig. 10: What is the upper detection limit of the OPC? 

Fig. 10b: Why is the OPC counting particles with concentration >100 L-1 below the cloud base?

References

Baumgardner, D., Abel, S. J., Axisa, D., Cotton, R., Crosier, J., Field, P., … Um, J. (2017). Cloud Ice Properties: In Situ Measurement Challenges. Meteorological Monographs, 58, 9.1-9.23. https://doi.org/10.1175/amsmonographs-d-16-0011.1

This manuscript describes the application of cloud particle sensor (CPS) sondes on free and tethered balloon systems. The CPS sondes definitely promise potential for investigation of the vertical structure of clouds in remote locations as a light-weight and low-cost instrument and the manuscript fits the scope of AMT. However, there are several points regarding the measurement technique and data analysis which need to be resolved before this paper can be recommended for publication.

How does the laser beam profile look like? In optical particle detection, the laser beam profile in the sample volume plays an important role. This is a major point that needs to be added to this manuscript. Best practice is to use optical modeling software and also measure the intensity profile at different places within the sample volume with a photodiode on a translation stage or a camera. 

Another point that deserves more investigation is the correction factor for the total particle counts. The authors already mentioned inhomogeneities in the sample volume, so one of the basic assumptions is actually violated. In my opinion, this issue can only be resolved by intercomparison with state-of-the-art optical cloud particle spectrometers in a laboratory setup. 

Although the flow conditions have been investigated via CFD modeling, there are no results from real measurements shown. In particular, boundary layer effects and "slow flow zones" are much clearer to see in an experimental flow characterization in a wind tunnel. Ideally, a particle generator is part of the laboratory setup to also investigate how the flow conditions influence detectability of cloud hydrometeors. In addition, a particle generator producing water droplets should be used to calibrate the CPS sondes. Further experiments in an icing wind tunnel would be helpful to investigate the ability of the sensor to distinguish ice from supercooled liquid water under realistic conditions.