Spatially Resolved Backscatter Ceilometer Design Team Hiba Fareed, Nicholas Paradiso, Evan Perillo, Michael Tahan Design Advisor Prof. Gregory Kowalski Sponsor, Spectral Sciences Inc. Steve Richstmeier, Neil Goldstein Abstract The project sponsor, Spectral Sciences (SSI) requires a device that can measure the base height of clouds as well as obtain a coarse resolution image of the cloud bottoms. SSI will use this data to improve existing field surveys and provide supporting data for their projects. The presented design allows for imaging the cloud to obtain spatial information. Existing ceilometers only effectively measure the single scattering photon returns which are used to determine cloud base height. To this date, no commercial device has collected information on the diffuse, multiple scattering, photons. The proposed design consists of a wide field of view lens that forms an image onto a digital micromirror-array device (DMD) which is used as a spatial filter for selecting fields of view on the cloud. The entire image of the DMD is condensed onto a single pixel avalanche photodiode (APD). This design rests at the intersection of acceptable signal to noise and commercially available hardware. A rigorous signal to noise and feasibility analysis was performed and an optical and opto-mechanical design was developed. Experimental work is currently being performed and initial lab testing of the final device is predicted to finish by the end of April. 12 11 For more information, please contact g.kowalski@neu.edu. 88
The Need for Project Commercially available ceilometers are underpowered and only detect cloud base height. Spectral Sciences needs to detect thickness and spatial variations in addition to base height. This device would be useful as a research tool to characterize the effects of light scattering in clouds, estimate cloud thickness, and calculate cloud base height. The Sponsor company, Spectral Sciences Incorporated, requested the device because of a need to detect these cloud parameters, commercial devices only allow for detection of base height. As a prototype this device will provide information needed by SSI for many atmospheric detection and remote sensing projects within the company. The Design Project Objectives and Requirements Design, build, and test a ceilometer capable of detecting a cloud s base height, thickness, and spatial variation. Design Objectives The device must be capable of detecting clouds using Lidar (Radar using light) to determine base height while also having the capability to generate low resolution images of the cloud (3-9 resolution units) with a sufficient field of view to determine the thickness of the cloud. Because the detection range of clouds is large (up to 3km) a very sensitive detector must be used, furthermore the contribution of sunlight competes with the signal of interest. For these two reasons maximizing signal to noise is a design objective. The system must have a wider than typical field of view compared to normal ceilometers in order to detect off axis light from the clouds to determine photon scattering time which is inherently related to cloud thickness. The device must be portable, weatherproof, sealed to any stray light and water or dust. The laser transmitter needs to be low enough power to be eye safe (Class 1). The eye safe requirement is put in place to avoid the possibility of damaging pilots eyesight if planes happen to fly through the testing zone. Design Requirements A larger than normal field of view for the optical receiver must be designed in order to meet the objective of detecting cloud thickness. Typically Lidar ceilometers utilize a telescope receiver to maximize collecting area and minimize FOV, a 89
detection range of ±0.1 is common. The wide field design requires both large collecting area and large FOV which is optically more difficult to achieve. The design has a ±2 FOV which appears to be 210m by 210m on a 3km high cloud. From signal to noise analysis using a Monte-Carlo Model for light scattering it was determined that signals on order of 1e- 10 to 1e-7 W must be detected. Solar noise during the day can compete with these signals even after filtering, and signal to noise is anywhere from 13 to 2 worst case. Data acquisition speed is critical because cloud speed can be up to 9m/s. The device must collect all required data points on the cloud faster than 2 seconds to minimize the smearing effect caused by observing different parts of the cloud at different times. Design Concepts considered Two design concepts were considered which met all objectives APD Array System Two designs were considered that would meet the requirements of the system. Telephoto Lens with Avalanche Photodiode Array This design used a long focal length camera lens to form an image onto an array of high sensitivity Avalanche Photodiodes (APD). The resolution would be fixed by the APD array, while the field of view and collection efficiency would be determined by the optics. The lens f-number could be as low as f/1.0 which would provide very good light collecting ability. Limitations to this system lie in the fixed resolution array sensor that is able to only view 4X4 pixels on the cloud. The array sensor was determined to be not feasible considering the project timeline and team expertise. 90
Digital Micromirror-Array with High Throughput Optics DMD System A Digital micromirror-array device (DMD) is used to independently select what parts of the image plane will be detected by the sensor. The sensor is only a single pixel so the image on the DMD is integrated and read as a single data point, in this way an image with reasonable resolution is generated in time by selecting different parts of the image at different points in time. This is the basic idea of a single pixel camera that has been applied to cloud Lidar through the use of a high efficiency optical system and a low noise APD sensor. Recommended Design Concept The design uses a DMD to selectively filter fields of view of the cloud onto a single pixel detector and provides spatial resolution without expensive array detectors. Receiver Assembly Exploded View Showing Optical Elements (1) Design Description The design uses a DMD to selectively pass portions of the total optical field of view through an optical system. An image of the object being viewed can be generated using only a single pixel sensor by scanning through a range of field of view. At each instant in time only 1 data point is being taken, however over time a full scale image is generated. The critical design element is the DMD which is an array of 1024 X 768 micron sized mirrors which can independently tilt in one of two angles. One direction of tilt allows the light to pass further into the optical path this is referred to as the ON position. The second tilt angle causes the incident light to reflect into a wall and not continue further in the optical path, this is referred to as the OFF position. Challenging engineering tasks in this project involved the packaging of the optical design into real mechanical mountings. Mechanical clearances and tolerance stack up was a primary concern when designing the receiver assembly. All critical optical elements were pinned to a single plate to reduce error buildup. Careful thought was put into the alignment process, as 91
the mountings provide compensating motions which allow for removal of errors in all 5 required degrees of freedom of the system. Views to the left show the final receiver assembly design in an exploded view with optical elements exposed and a section view through the optical plane. Receiver Cross-section showing Light path Monte Carlo Model: Diffuse Photon propagation through a 3km cloud, 200ns after laser pulse.(160x160m 2 ) (2) Analytical Investigations The system level and feasibility design used a Monte Carlo based cloud modeling program written at Spectral Sciences. The model uses brute force statistics to predict how light will propagate through clouds. Several test cases were run using system parameters that would be found from commercially available components. The critical design driver was signal to noise during worst case daytime conditions. To the left is a typical image of the photon number exiting a cloud bottom after a laser has illuminated only a small central point of the cloud. The optical design for the receiver system was developed in the commercial raytracting program ZEMAX. The optical design needed to meet two goals; high transmission at design wavelength (905nm), and high light throughput. Further consideration was put into making a design that would be simple to integrate mechanically. (3) Experimental Investigations The laser transmitter has undergone experimental testing in lab. The laser has been coupled to a commercial telescope to expand its beam diameter. The goal of beam expansion is to lower its divergence so that it illuminates a small area of the cloud. A small illumination area helps to achieve higher signal to noise when reading base height. The laser expander assembly was bread-boarded and tested for collimation. A CMOS camera capable of detecting the 905nm (IR) wavelength was used to characterize output of the expander. Zemax Model of Receiver Optical System The laser wavelength output was characterized with a 92
spectrometer supplied by SSI. The spectrometer observed the spectral output of the laser with sub-nanometer resolution. The spectral signature of the laser was used to verify the purchase of the optimal bandpass filter. (4) Key Advantages of Recommended Concept The DMD system allows for flexibility, expansion for future applications, and noise reduction. The flexibility lies in the DMD s high resolution (1024 by 768) which is effectively unlimited resolution compared to the resolvable signal resolution. The unlimited resolution allows for on the fly change of resolution by changing the pixel size of the image reconstruction. The device has possible applications outside of Lidar as a standalone ultra-high sensitivity single pixel camera. The noise reduction benefits over the APD array design occur through better sunlight reduction and stray light minimization. Financial Issues The materials cost of the prototype was $20,000 The prototype design material cost was $20,000. All materials were purchased by Spectral Sciences and the device will be owned entirely by SSI after the capstone term. Following successful validation of the concept there is opportunity for future funding, however it is not expected to ever make it to a commercial setting. Recommended Improvements A narrower bandpass filter would improve signal to noise ratio, while a commercial near- IR lens could produce a finer resolution image. A better bandpass filter could be used that would effectively cut the solar noise in half. The optimal filter could not be purchased during this term due to prohibitively long lead time. A higher resolution objective lens could be purchased that 93
would improve the quality of the image formed at the DMD. The lens was not purchased during this term because its cost was prohibitively high. If the eye safe laser requirement were not restricting the design then a higher power pulsed laser could be used to improve signal to noise. Signal to noise scales linearly with the peak power of the laser. In terms of system level improvements any optical complications to detection such as polarization sensitivity, fluorescence, or coherent detection schemes could be added to the existing optical system which would expand detection capabilities.. 94