Remote Spectral Capture designed an analog Fourier Transform detector chip and compact Mach-Zehnder interferometer for the USAF laser warning receiver (LWR) application. This design, illustrated in Figure 1, satisfies the need for reliable near real-time indication of laser illumination by producing the complete spectra of coherent visible and near infrared light as often as every 100 m sec and detects both continous wave (CW) and pulsed lasers. The design discriminates non-laser phenomenon by analyzing the full spectrum wavelength characteristics of the incident light. Analysis results indicate that the design meets all USAF LWR specifications including sensitivity to detect 0.5 mW/cm^2 signals, 20 nSec laser pulses and CW lasers while maintaining an extremely low false alarm rate. The device is also suitable for a Fourier transform readout chip in a hand-held spectrometer for detecting volatile organic compounds in soils at hazardous waste sites.

Platforms that would use laser warning receivers, such as USAF fighter aircraft, place constraints on the size, power consumption, and have severe environmental requirements (temperature and vibration). The major challenge of a high performance laser detection system is to devise a design for a small, wide-field-of-view sensor that is extremely resistant to false alarms. Major sources of false alarms have been characterized, including high power Xenon flash lamps under high atmospheric scintillation conditions. A successful LWR system must specifically rejects these sources while simultaneously detecting laser sources. The LWR system must provide the Air Force with a low risk approach and address the important issues of performance, cost, size, weight and ruggedness. RSC developed LWR specifications in conjuction with a our Wright Labs Project Officer to derive verifiable system specifications. These specifications were flowed down into Subsystem-level allocations that promote a methodical and proven means of designing and producing viable building blocks for integration into a high performance LWR system.

The top level USAF aircraft LWR system requirements summarized in Table 1. A complete draft LWR System Specification was developed jointly with Wright Labs and provided a Phase I deliverable.

RSC's technical approach was to exploit characteristics of an interferometric spectrometer that can simultaneously provide wavelength, coherency, and temporal discrimination while providing superior optical throughput and wide field of view. RSC designed and analyzed a laser warning receiver, as shown in Figure 2, which consists of an Optical Subsystem, a Detector Subsystem and a Processing Subsystem. The Optical Subsystem uses a fiber bundle with 1024 fibers (32X32 array) to collect light from a window in the skin of the aircraft to provide a 90 degree field of view and a direction of arrival capability. The Optical Subsystem also consists of a beam splitter, an input lens, a monolithic interferometer, and a Fourier transform lens to produce a fringe pattern on a proprietary Fourier transform detector chip. The Detector Subsystem's includes RSC's analog Fourier Transform Chip which converts the fringe energy in the spatial interferogram into wavelength bin amplitudes which are then processed by a Processing Subsystem. The Detector Subsystem also includes a 32 x 32 pixel imager chip for DOA inputs to the Processing Subsystem.The Processing Subsystem reduces the full spectral information into declaration of a laser event as appropriate and ensures discrimmination against non-laser phenomena.

Design of the Fourier transform detector chip was driven by a combination of the LWR performance requirements (complete spectral readout at high speeds, CW and pulsed laser detection, laser characteristic capture) and experimental measurements of dynamic range and noise floor of a low power, low noise P-channel cascode amplifier. The general approach to the design of the Fourier transform detector chip used Hspice to simulate circuit performance and a suite of Integraph design tools for mask layout. As illustrated in Figure 3, the circuitry on the chip is highly repetitive. The 680 photo diodes are identical and all of the odd diode amplifiers, including bias and coupling networks, are also identical. The even diode amplifiers are the same as the odd diode amplifiers except that they are reversed in orientation. Thus, we only had to design one column channel in detail. After simulation, the Intergraph layout tools were used to repeat it many times. The capacitor matched filters were laid out by repeating a single capacitor in all columns and rows. This layout was then written to a file in Caltech Intermediate Format (CIF) format, which was then modified to provide the proper capacitor weighting for the corresponding optical design. The modified CIF file was then imported back into the layout. The row amplifiers and output shift register stages were also identical. Finally, the non-repetitive edge circuitry for input/output pads, clock drivers, and test element groups (TEGS) was designed and laid out.

The detector chip performs the functions of wavelength determination, laser pulse extraction, CW laser detection and identification, and incoherent source rejection. Each photo diode in the linear array drives a low-noise CMOS amplifier through a high-pass filter consisting of a bias resistor and a coupling capacitor. This filter rejects low frequency signals so that pulsed laser signals can pass and be processed by the rest of the chip even in the presence of large amounts of background light. The bias resistor is used to supply any DC photo current generated by direct sunlight, CW lasers or similar sources.

A MOS switch has been added in series with the bias resistor so that the presence of CW laser energy can be detected by momentarily pulsing this switch "off". This interrogation pulse diverts the CW signal through the high pass filter to the amplifiers. The CW detection duty cycle (duty cycle = 1 KHz rate / 10 KHz pulsed source frame rate =.1), is kept extremely low to reduce the chances of pulsed lasers going undetected during the CW sampling period. While in the CW detection mode the background suppression feature, which results from the capacitive coupling, is no longer active. This means that if the sun is within the field of view at the same time as a CW laser, there is a direct competition between the sunlight and the laser light which is not present when detecting pulsed lasers.

The output of the chip is a multiplexed stream of matched filter outputs which represent the energy in the wavelength bins. These outputs will be digitized and serially transferred to off-chip laser detection and discrimintion processing. Since all of the multiplications and additions required to perform the matched filtering have already occurred, round-off error is not as great a problem in the proposed system as for systems where the matched filtering is performed digitally. There are fewer wavelength bins than photo diodes and the reduced dynamic range required for each amplifier output is due to the processing gain inherent in the summation over the photo diode channels. RSC's approach reduces the required data rate (because of the parallel nature of the on-chip processing), improves data robustness, and reduces the post-processing computation load considerably. The detector chip can be fabricated with high yield (see Section 1a.8 -- chip producibility) using widely available standard CMOS processing.

Each diode in the array is biased with a resistor capacitively coupled to its amplifier. This RC bias coupling time constant is set to pass 20 - 200 nSec pulses. The low noise, high speed, wide dynamic range amplifiers, shown in Figure 4 utilize a cascode configuration operated at low current -- just above the subthreshold region, for maximum gain. Depletion mode MOSFETs are used to obtain high resistance values in minimal chip area. This amplifier has been designed, modeled, and fabricated; it was repeated on 20 micron centers to accommodate a 10 micron photo diode pitch.

The cascode integrating stage measured open loop gain is 100, the gain-bandwidth product is 40 MHz, and the power consumption is 75 microwatts. Addition of peak detection and the inverting stage raised the power consumption to 200 uW per column amplifier. The use of 680 amplifiers, one for each photodiode, gives 2,720 GHz of on-chip gain-bandwidth at a power consumption of 136 mW.

Hspice simulation of the MOS amplifier with illumination from a 20 nS, 0.5 mW per square cm, laser pulse is shown in Figure 5. This simulation predicts an output from the column amplifiers of 0.27 v. This pulse is then peak detected and applied to an inverting stage to provide both positive and negative drive to the capacitor matrix as illustrated in Figure 3 above.

Central to our LWR system design, is the analog Fourier transform. This time parallel transform requires a capacitor matrix that implements the Fourier coefficients. Each wavelength bin has a matched filter bank of capacitors to weight the spatial interferograms fringe energy contributions across the diode array (each capacitor value equals the magnitude of the weight of its particular component for each bin and the sign is determined by the + or - amplifier output connection). The two capacitor matrices on either side of the diode array contain 680 (diodes) X 70 (wavelength bins) X 2 (Cosine and Quadrature-phase signals) = 96,200 separate capacitors. The values of these capacitors were determined by calculating the fringe locations. For the LWR application, the current specification is to cover the wavelength region from 420 to 1060 nm with 10 nm resolution spacing between each bin. This results in 70 bins with equal wavenumber separation, and implies that there will be 140 rows on each capacitor matrix. Particular laser lines of interest wavelengths can also be added, if desired. Since equal spacing in wavenumbers results in equal length matched filters, it makes more efficient use of chip area. Using an average bin spacing as a measure of our resolution and designing for approximately 12 nm resolution at the red end and about 5 nm resolution at the blue end the same number of bins achieve an average resolution well below 10 nm. The automated FT capacitor weighting approach saved hundreds of layout technician labor hours that would have been needed to size each of the 96,200 capacitors in the layout (See Section 1b.4 for a detailed discussion of the automated capacitor matrix design)

The chip floor plan follows the architecture shown in Figure 3. The photo diode array consists of 680 diodes on 10 micron centers. The height of the photo diode array would be 2 mm resulting in a central photosensitive area 2 mm high and 6.8 mm wide. The proposed floor plan is shown in Figure 6. Test layouts of the column amplifiers indicate that they would require 600 microns above and below the photo diodes. The capacitive matched filters are 2 mm high, and the row amplifiers and output scanners would require about 1 mm each in width. Pads, test element groups (Tegs), on-chip drivers, etc. would be fit into available space as indicated. The active area 7 mm by 9 mm. Overall chip size is somewhat larger, slightly larger than 10 mm square. A 4 inch wafer would contain approximately 60 chips.

Diodes are easy to fabricate with high yield so they should not contribute significantly to yield loss. The major part of the active area, 30 %, is occupied by the matched filters. These capacitors consist of polysilicon lower electrodes, 800 angstroms of thermal silicon oxide, and polysilicon upper electrodes. The total capacitor area is ~ 0.1 sq cm. At a typical oxide defect density of 1 per sq cm, the yield loss due to the matched filters is expected to be about 10 %. The total yield will certainly be somewhat lower than the expected 90% of the matched filters. The probe yield of CID image sensors of comparable size, 0.42 sq cm, was about 25 %.

A fiber bundle containing 1024 individual fibers illuminates a beamsplitter dividing light between an area imager used for DOA determination and an interferometer for coherent light and wavelength detection. Each fiber in the bundle has a Grin Rod lens attached to its input which matches a 3 degree acceptance angle to the acceptance angle of the fiber. The fiber ends are fanned out so that the Grin lens acceptance angles combine to cover the full 90 degree field of view. The fiber bundle outputs are imaged onto a 32X32 pixel array whose output is time correlated with the coherent light detection to allow calculation of the centroid of the brightest points in the field of view. The optics design has been checked for field of view, with the result that all fibers within a 3 mm diameter circle will illuminate the entire detector array. Thus, a 32 X 32 array of 70 micron fibers will provide full wavelength resolution for light form every fiber.

The Optical Subsystem configuration, shown in Figure 7, employs a bundle of fibers which fan out to cover the desired field of view. The plane where all of these fibers terminate is, in effect, an image plane of the field of view, and this image can be processed by the interferometer and the DOA imager. Figure 7 depicts two rays entering the Mach-Zehnder interferometer from 45 degrees and zero degrees from the optic axis. Note that both of these rays enter the solid glass Mach-Zehnder interferometer essentially parallel to the optic axis. This feature maximizes the number of usable fringes regardless of the type of interferometer. Both of the sheared rays are shown for the ray entering at zero degrees, whereas only one ray is shown for the other ray.

The interferometer must produce two mutually coherent images which are both laterally and axially displaced from each other. The axial displacement prevents fringes from being formed except by light whose coherence length exceeds the amount of displacement while the lateral displacement determines the fringe spacing for a given light wavelength. To prevent the fringes from washing out, all points in the field of view must develop the identical fringe pattern. This is a severe requirement, and it dictates stringent specifications on flatness, alignment, and mechanical stability. These requirements are addressed by using solid glass prisms that are bonded together with optical adhesive.

The purpose of the Fourier transform lens is to focus parallel rays emerging from the two mutually coherent images of the field of view to a specific point on the detector chip. Rays spanning a range of angles produce a fringe pattern on the Fourier Transform detector chip. Although the rays are collimated by an input lens which is focussed on the input aperture, a fan of rays is re-created in one dimension by a cylindrical lens immediately behind the collimating lens.

As described above, the LWR Optical Subsystem consists of an appropriate number of optical fibers that sense the incoming radiation from all directions, a cylindrical lens, a solid glass Mach Zehnder interferometer, a Fourier Transform lens, an optional field corrector lens and a detector chip. The purpose of this section is to detail the interrelationships between these components and to indicate how the recommended system parameters were chosen. The next sections cover the components listed above, and the last section covers the lens designs and the overall layout.

When the two pieces of glass that comprise the interferometer are of different thicknesses, lateral shear is introduced. When the replicated images are viewed from a particular direction, and the corresponding rays in those directions are focused to a point, that point will be in a light or dark fringe depending on the optical path difference in that direction.

The desired lateral shear is determined by the number of fringes needed to achieve the designed resolution, and in the case at hand, this may result in an axial shear that exceeds the coherence length of some lasers. A several solutions to this design issue exist (see Section 5.2 -- Future Research). It is now appropriate to outline the consideration affecting the interfeometer geometry: namely, the lateral shear requirements.

The geometry shown in Figure 9 defines the optical path difference (OPD). Note equal increments in OPD (which define the fringe spacing) will occur with equal increments in the sine of the acute angle of the similar triangles shown above, whereas equal increments on the chip (which corresponds to diode spacing) occurs with equal increments in the tangent of that same angle. This discrepancy can be handled either optically (by introducing an appropriate distortion) or electronically (by increasing the spatial frequency of each of the matched filters towards the edges of the chip.)

We have selected the second option using an exact expression for the OPD to determine fringe location from this interferometer. Thus, although the following discussion is useful for design purposes, the actual OPD and index of refraction (which determines the capacitor values) are determined by an exact ray trace of the optical system.

Using a and s to represent the axial and lateral shears, and after simplification, the expression for the optical path difference can be shown to be:

where L is the focal length of th F.T.lens and n is the index of refraction of the glass. The substitution a = s/n has been made,which assumes that the light enters and leaves at 45 degrees with respect to the beam splitter and that the optical path difference is the free space value multiplied by n. The variable x is not precisely the position on the chip because the direction of view inside the glass is not the same as the direction of the rays outside the chip (due to refraction at the prism boundary). However, this difference is tantamount to a different diode spacing and will merely change the wavelength scaling.

If the above expression is expanded as a power series, one gets:

Thus, one sees a constant offset (which discriminates against non-laser light), a linear term which is proportional to the index of refraction, a curvature that does not depend on index, and a cubic distortion that does. Although the distortion terms are of the order of a percent or less, they are serious corrections when several hunderd fringes are present, and they must be accounted for.

To derive design parameters from this geometry, one must choose the half-width of the chip, the focal length of the lens, and the number of fringes that one wants on the chip (i.e. the OPD corresponding to the edge of the chip).

Since the smallest number of fringes occurs at the longest wavelength, and since the resolution is the wavelength divided by the number of fringes, the longest desired wavelength determines the worst case resolution. The minimum wavelength can be determined by multiplying the number of available wavelength bins by the desired resolution and subtracting this total from the maximum wavelength.

For the case at hand, the longest wavelength of interest is 1 micron, and we want a minimum of 10 nm resolution across the responsitivity range of silicon. Selecting 400 nanometers as the shortest wavelength implies that we need 60 bins.

At 1 micron we need 100 fringes to achieve 10 nm. resolution, and this means that we will have 100x(1000/400) = 250 fringes at 400 nanometers. To achieve an OPD of 50 microns at each edge of the chip, we need a lateral shear equal to

To avoid aliasing when there are 250 fringes, we need a minimum of 500 diodes plus whatever is needed for apodization since the apodization diodes do not fully contribute to the resolution. The results of our simulations indicate that a linear weighting function applied to 90 diodes on either end of this array provides adequate signal quality. We have therefore selected 680 diodes as the basis for our design resulting in a detector array with a half width of the 3.4 mm. Selecting a system F/3.5 due to availability of high quality achromatic lenses at this size means that x/L at the edge of the chip will be 1/7. Given the halfwidth of the diode array is 3.5 mm, the focal length of the FT lens comes out to be 25 mm. To allow for the fact that the input aperture is a 3 mm. circle rather than a point, it is necessary to provide for a slight expansion of the optical system. Thus, while the input lens is 1/2 inch in diameter, the prisms accomodate 5/8 inch, and the F.T. lens is 3/4 inch in diameter. This will allow the extremal points of the input aperture to fully illuminate the detector.

Figure 10 Basic Diode Amplifier Circuit

Figure 11 Noise Equivalent Circuit

Rd = 1000m /7m (100 W /sq) = 14.3 K

The noise resistance, Rn = Rd/3 = 4.8 K

Ed = 0.129 sqrt (4.8 K) = 9 nV/sqrt Hz

Rl =200 K

Er = 58 nV/sqrt Hz

Root Sum Square of Ed and Er = 58.7 nV/sqrt Hz

The limiting bandwidth for noise at the diode node is set by Rl and (Cd + Cc).

fn = p /2 1/(2p RC) = 1/4RC = 1.12 MHz

Input Johnson noise voltage = 62 m V

Cc = 0.4 pF

Cf = 0.74 pF

Gain from the diode node to the amplifier output is Cc/Cf = 0.284

The amplifier noise output from this component is 18 m V.

Ca = 0.27 pF

Ea = 0.129 sqrt (1/gm) = 15 nV/sqrt Hz

Noise Gain for amplifier noise (Ea) = (Cc + Cf)/Cf = (0.4 + 0.27 + 0.74)/0.74 = 1.91

Noise Bandwidth = p /2 fc = 630 KHz.

Amplifier Johnson noise at output =1.91 ( 17.5) sqrt (630K) nV = 27 m V

The KTC noise upon reset at the amplifier output is Vktc = 401 sqrt (C (pF)) Qe/Cf

= 401 sqrt(0.72 + 0.27 + 0.74) 1.6e-19/0.74e-12 = 114 m V

Typical dark current levels at 25° C are 1 nA/cm2 + 1 pA/cm of edge. This leads to a total of 3e-13 Amps which is very small compared to the solar illumination current.

Solar illumination is 1000 W/sq meter on an input aperture of 0.07 sq cm. Solar current will be 0.65 uA per diode (optical thruput = 0.2, responsivity = 0.315 A/W, 680 diodes illuminated). The shot noise is 0.36 nA. The photo diode bias resistor (200K) converts this to 72 uV rms. Noise gain from the diode node to the output is 0.284 as noted above.

The root-sum-square of the noise levels identified is 121 m V. While it may be possible to use correlated double sampling to reduce the KTC noise, this would double the Johnson noise power and lead to an output noise level of 55 m V. This small improvement would not justify the added complication of correlated double sampling.

The noise voltage of all 680 channels adds in quadrature; if we assume a weighting of one (the worst case), the maximum noise on any filter (frequency bin) output is 3.2 mV. The minimum signal at this point is 30 mV (0.5 mW/sq cm at 20 nS pulse width). The false alarm rate is very sensitive to the detection threshold. A detection threshold of 50% of the minimum signal gives a false alarm rate of 5.2 per hour. If the detection threshold is set to 75% of the minimum signal, the false alarm rate drops to 3.8 x 10-6 per hour. Actual signal to noise ratio and resulting false alarm rates will be a function of the collection optics configuration at the input of our interferometer.

The innovative FTS device, with its inherent stability, wide field of view, compactness and low power dissipation would be of interest to instrument vendors in a variety of remote, portable spectral estimating applications. Private industrial companies that use or produce chemicals could employ such a device to monitor processes or instrument storage tanks to detect leakage.

Mr. Michon was RSC's Principal Investigator for RSC's SBIR Phase I effort. Mr. Michon is the co-inventor (together with Dr. Tiemann) of the monolithic analog Fourier transform chip. He completed the LWR application and chemical sensing application imager/Fourier transform chip design and foundry layout in December 1995. His recent work includes integrated circuit design and circuit performance simulation under the Strategic Environmental Research and Development Program (SERDP). He will be testing the SERDP developed LWR and chemical spectrometer chips in February 1996. Mr. Michon has spent the last 20 years developing advanced solid-state image sensors. Mr. Michon is a recognized world expert in Charge Injection Device (CID) imager technology and has researched and designed ultra low power VLSI imager circuits with performance demands equivalent to those we will be addressing in our research including television sensor, electronic still photography sensors, star trackers, and medical imaging detectors. In addition to our brassboard LWR requiring this type of detector technology experience, follow-on effort in LWR Direction of Arrival determination imager will also require Mr. Michon's imager experience. Mr. Michon has worked in the field of Prior to the CID image sensing work, he was involved in high-speed, random access active memory development, and remote computer data handling equipment. Mr. Michon holds 25 patents, 13 in solid-state image sensing techniques. Mr. Michon has contributed to more than 50 publications including:

Charge Injection Modeling in Solid State Image Sensors

J.M. Pimbley and G.J. Michon

IEEE Trans. on Electron Devices, Vol. ED---34, No.2, Feb. 1987

Conference Paper: "CID Image Sensor and Improved Sensitivity"

G.J. Michon and T.L. Vogelsong

SPSE 26th Fall Symposium on Electronic Imaging, Oct. 13-17, 1986

Simple Measurement of the Properties of a Distributed Resistor-Capacitor Line

J.M. Pimbley and G.J. Michon

IEEE Transactions on Electron Devices, Vol. E-D-5, No. 8, Aug. 1988

M. P. Dierking, J. F. McCalmont, M. A. Karim, F. L. Baxley, R. A. Mann, J. J. Tiemann, G. J. Michon, D. J. O'Donnell, H. W. Tomlinson

SERDP '95 Conference paper entitled, "Field Portable FTS Fiber Optic VOC Sensor";

R. Mann, G. Michon, D. O'Donnell, J.Tiemann, H. Tomlinson

DoD Photonics '96 Conference paper, entitled "Interferometric Laser Warning Receiver with Dual-use Environmental Sensing Application";

H. Tomlinson, G. Michon, J.Tiemann, H. Burke, R. Mann, D. O'Donnell,

Pittcon '96 Conference paper, entitled, "Fourier Transform Spectrometer Detector with Integrated Interferogram Processor"

Jerome Tiemann, Ph.D., Coolidge Fellow, Chief Scientist, RSC Inc. and GE CR&D Scientist

Education: ScB. Physics, Mass. Inst. of Tech.

Ph.D. Physics, Stanford University

Honors: Coolidge Fellowship Award 1975

Elected Fellow of IEEE 1976

Elected to Nat. Acad. of Eng. 1984

Dr. Tiemann is coinventor of the analog Fourier transform chip (together with Mr. Michon). Under the USAF SBIR Phase I, Dr. Tiemann architected RSC's LWR system and designed and breadboarded the Mach-Zehnder interferometer based optical subsystem. He devised and modeled the direction of arrival determination scheme. Dr. Tiemann devised and coded the automated capacitor weighting method for the analog FT chip. Dr. Tiemann is currently performing research and development on electronic systems and system architectures with particular emphasis on signal processing functions and with applicability to integrated circuit fabrication resources. His work has involved optics and optical techniques since 1989. He was principle system architect for Comband (R) bandwidth compression system, and the PASS ultra-sonic imager product of Medical Systems Division of GE. He has had extensive research and development on discrete-time analog signal processing systems usign charge coupled devices (CCDs) and charge injection devices (CIDs). Dr. Tiemann's inventions are covered by 85 patents. Dr. Tiemann has contributed to over 50 technical papers, including the following relevant publications and pending publications:

M. P. Dierking, J. F. McCalmont, M. A. Karim, F. L. Baxley, R. A. Mann, J. J. Tiemann, G. J. Michon, D. J. O'Donnell, H. W. Tomlinson

SERDP '95 Conference paper entitled, "Field Portable FTS Fiber Optic VOC Sensor";

R. Mann, G. Michon, D. O'Donnell, J.Tiemann, H. Tomlinson

DoD Photonics '96 Conference paper, entitled "Interferometric Laser Warning Receiver with Dual-use Environmental Sensing Application";

H. Tomlinson, G. Michon, J.Tiemann, H. Burke, R. Mann, D. O'Donnell,

Pittcon '96 Conference paper, entitled, "Fourier Transform Spectrometer Detector with Integrated Interferogram Processor"

Daniel O'Donnell, President RSC, Lockheed Martin Principal Systems Engineer

Education: BSEE Northwestern University

MSEE University of Southern California

Mr. O'Donnell co-founded RSC in 1992. Mr. O'Donnell has been President of RSC for over three years. In addition to his business duties, Mr. O'Donnell performed the MATLAB modeling of the LWR system performance under the Phase I SBIR effort. While at TRW, Hughes and Lockheed Martion, Mr. O'Donnell has over twelve years of system engineering, circuit design, system integration and test experience for small and large scale aerospace system. His work has included Fourier transform spectrometer system performance analysis for deployment harsh enviroments. Having performed considerable research on space-based instrument applications, Mr. O'Donnell understand the implications of vibration, air currents, and mechanical distortions which are expected in air frame mounted LWR environments. He is currently lead engineer for a team working on an optical signal processing system integration effort. Mr. O'Donnell has contributed to the following relevant publications and pending publications:

M. P. Dierking, J. F. McCalmont, M. A. Karim, F. L. Baxley, R. A. Mann, J. J. Tiemann, G. J. Michon, D. J. O'Donnell, H. W. Tomlinson

SERDP '95 Conference paper entitled, "Field Portable FTS Fiber Optic VOC Sensor";

R. Mann, G. Michon, D. O'Donnell, J.Tiemann, H. Tomlinson

DoD Photonics '96 Conference paper, entitled "Interferometric Laser Warning Receiver with Dual-use Environmental Sensing Application";

H. Tomlinson, G. Michon, J.Tiemann, H. Burke, R. Mann, D. O'Donnell,

Pittcon '96 Conference paper, entitled, "Fourier Transform Spectrometer Detector with Integrated Interferogram Processor"

Roger A. Mann Vice president RSC, Lockheed Martin (formerly GE Aerospace) Project Manager

Education: BSEE Cornell University

MSSE Virginia Tech

Mr. Mann has over fourteen years of system engineering and integration and management experience integrating remote sensing, signal processing and communications systems at General Electric Co., Martin Marietta and Lockheed Martin. Under the USAF Phase I SBIR, Mr. Mann developed the draft LWR specification. While at Lockheed Martin, Mr. Mann has managed the integration of several systems valued over 100 million dollars. While at RSC, he has managed the start-up operations involving transition from technical feasibility to product development. At Lockheed Martin, Mr. Mann currently manages a team of twenty engineers on communications and optical signal processing system integration project. Mr. Mann is a member of IEEE and AFCEA. Mr. Mann has contributed to the following relevant publications and pending publications:

M. P. Dierking, J. F. McCalmont, M. A. Karim, F. L. Baxley, R. A. Mann, J. J. Tiemann, G. J. Michon, D. J. O'Donnell, H. W. Tomlinson

SERDP '95 Conference paper entitled, "Field Portable FTS Fiber Optic VOC Sensor";

R. Mann, G. Michon, D. O'Donnell, J.Tiemann, H. Tomlinson

DoD Photonics '96 Conference paper, entitled "Interferometric Laser Warning Receiver with Dual-use Environmental Sensing Application";

H. Tomlinson, G. Michon, J.Tiemann, H. Burke, R. Mann, D. O'Donnell,

Pittcon '96 Conference paper, entitled, "Fourier Transform Spectrometer Detector with Integrated Interferogram Processor"

Harold Tomlinson, Vice president and Technical Director RSC, Optical Processing Program Manager, General Electric Co., Corporate Research and Development;

Education: B.S., Physics St. Lawrence University

MS, Biomedical Engineering Rensselaer Polytechnic Inst.

Mr. Tomlison co-founded RSC. He currently oversees the technical direction of RSC products and services. Under the SERDP effort, Mr. Tomlinson worked with Mr. Michon to develop the testing circuitry for RSC analog FT chip. Mr. Tomlinson has worked on development of imager systems for light detection using focal plane charge processing for over 7 years. Previously, Mr. Tomlinson worked in GE's Robots and Vision Systems Department as a design engineer, developing image processing algorithms for real-time machine vision systems. He has also developed and implemented high speed image segmentation algorithms for a 3D vision inspection system. Mr. Tomlinson holds 8 patents in the areas of CID imagers and vision processing architectures. He is a member of IEEE and SPIE. Mr. Tomlinson has contributed to the following relevant publications and pending publications:

M. P. Dierking, J. F. McCalmont, M. A. Karim, F. L. Baxley, R. A. Mann, J. J. Tiemann, G. J. Michon, D. J. O'Donnell, H. W. Tomlinson

SERDP '95 Conference paper entitled, "Field Portable FTS Fiber Optic VOC Sensor";

R. Mann, G. Michon, D. O'Donnell, J.Tiemann, H. Tomlinson

DoD Photonics '96 Conference paper, entitled "Interferometric Laser Warning Receiver with Dual-use Environmental Sensing Application";

H. Tomlinson, G. Michon, J.Tiemann, H. Burke, R. Mann, D. O'Donnell,

Pittcon '96 Conference paper, entitled, "Fourier Transform Spectrometer Detector with Integrated Interferogram Processor"