Technology Focus Computers/Electronics · Technology Focus: Test & Measurement Nulling Infrared...

Technology Focus

Computers/Electronics

Software

Materials

Mechanics

Machinery/Automation

Manufacturing

Bio-Medical

Physical Sciences

Information Sciences

Books and Reports

06-03 June 2003

https://ntrs.nasa.gov/search.jsp?R=20110024140 2020-05-09T09:44:21+00:00Z

INTRODUCTIONTech Briefs are short announcements of innovations originating from research and develop-

ment activities of the National Aeronautics and Space Administration. They emphasizeinformation considered likely to be transferable across industrial, regional, or disciplinary linesand are issued to encourage commercial application.

Availability of NASA Tech Briefs and TSPsRequests for individual Tech Briefs or for Technical Support Packages (TSPs) announced herein shouldbe addressed to

National Technology Transfer CenterTelephone No. (800) 678-6882 or via World Wide Web at www2.nttc.edu/leads/

Please reference the control numbers appearing at the end of each Tech Brief. Information on NASA’s Commercial Technology Team, its documents, and services is also available at the same facility or on the World Wide Web at www.nctn.hq.nasa.gov.

Commercial Technology Offices and Patent Counsels are located at NASA field centers to providetechnology-transfer access to industrial users. Inquiries can be made by contacting NASA field centersand program offices listed below.

Ames Research CenterCarolina Blake(650) [email protected]

Dryden Flight Research CenterJenny Baer-Riedhart(661) [email protected]

Goddard Space Flight CenterNona Cheeks(301) [email protected]

Jet Propulsion LaboratoryArt Murphy, Jr.(818) [email protected]

Johnson Space CenterCharlene E. Gilbert(281) [email protected]

Kennedy Space CenterJim Aliberti(321) [email protected]

Langley Research CenterSam Morello(757) [email protected]

John H. Glenn Research Center atLewis FieldLarry Viterna(216) [email protected]

Marshall Space Flight CenterVernotto McMillan(256) [email protected]

Stennis Space CenterRobert Bruce(228) [email protected]

Carl RaySmall Business InnovationResearch Program (SBIR) &Small Business TechnologyTransfer Program (STTR)(202) 358-4652 [email protected]

Dr. Robert NorwoodOffice of CommercialTechnology (Code RW)(202) 358-2320 [email protected]

John MankinsOffice of Space Flight (Code MP)(202) 358-4659 [email protected]

Terry HertzOffice of Aero-SpaceTechnology (Code RS)(202) 358-4636 [email protected]

Glen MucklowOffice of Space Sciences(Code SM)(202) 358-2235 [email protected]

Roger CrouchOffice of Microgravity ScienceApplications (Code U)(202) 358-0689 [email protected]

Granville PaulesOffice of Mission to Planet Earth(Code Y) (202) 358-0706 [email protected]

NASA Field Centers and Program Offices

NASA Program Offices

At NASA Headquarters there are seven major program offices that develop and oversee technology projects of potential interest to industry:

NASA Tech Briefs, June 2003 1

5 Technology Focus: Test & Measurement

5 Nulling Infrared Radiometer for MeasuringTemperature

6 The Ames Power Monitoring System

7 Hot Films on Ceramic Substrates for MeasuringSkin Friction

8 Probe Without Moving Parts Measures FlowAngle

9 Detecting Conductive Liquid Leaking FromNonconductive Pipe

11 Computers/Electronics

11 Adaptive Suppression of Noise in VoiceCommunications

12 High-Performance Solid-State W-Band PowerAmplifiers

12 Microbatteries for Combinatorial Studies ofConventional Lithium-Ion Batteries

13 Correcting for Beam Aberrations in a Beam-Waveguide Antenna

14 Advanced Rainbow Solar Photovoltaic Arrays

15 Metal Side Reflectors for Trapping Light in QWIPs

17 Software

17 Software for Collaborative Engineering of LaunchRockets

17 Software Assists in Extensive EnvironmentalAuditing

17 Software Supports Distributed Operations via the Internet

17 Software Estimates Costs of Testing RocketEngines

18 yourSky: Custom Sky-Image Mosaics via theInternet

18 Software for Managing Inventory of FlightHardware

19 Materials

19 Lower-Conductivity Thermal-Barrier Coatings

19 Process for Smoothing an Si Substrate AfterEtching of SiO2

20 Flexible Composite-Material Pressure Vessel

20 Treatment To Destroy Chlorohydrocarbon Liquidsin the Ground

23 Mechanics

23 Noncircular Cross Sections Could Enhance Mixingin Sprays

25 Machinery/Automation

25 Small, Untethered, Mobile Robots for InspectingGas Pipes

27 Manufacturing

27 Paint-Overspray Catcher

27 Preparation of Regular Specimens for Atom Probes

28 Inverse Tomo-Lithography for MakingMicroscopic 3D Parts

31 Physical Sciences

31 Predicting and Preventing Incipient Flameout inCombustors

32 MEMS-Based Piezoelectric/Electrostatic InchwormActuator

32 Metallized Capillaries as Probes for RamanSpectroscopy

35 Information Sciences

35 Adaptation of Mesoscale Weather Models toLocal Forecasting

35 Aerodynamic Design Using Neural Networks

36 Combining Multiple Gyroscope Outputs forIncreased Accuracy

38 Improved Collision-Detection Method for RoboticManipulator

06-03 June 2003

This document was prepared under the sponsorship of the National Aeronautics and Space Administration. Neither the United States Governmentnor any person acting on behalf of the United States Government assumes any liability resulting from the use of the information contained in thisdocument, or warrants that such use will be free from privately owned rights.



Technology Focus: Test & Measurement

Nulling Infrared Radiometer for Measuring TemperatureA microwave-radiometer self-calibration principle would be adapted to measurement of infrared.Stennis Space Center, Mississippi

A nulling, self-calibrating infrared ra-diometer is being developed for use innoncontact measurement of temperaturein any of a variety of industrial and scien-tific applications. This instrument is ex-pected to be especially well-suited to mea-surement of ambient or near-ambienttemperature and, even more specifically,for measuring the surface temperature ofa natural body of water. Although this ra-diometer would utilize the long-wave-length infrared (LWIR) portion of thespectrum (wavelengths of 8 to 12 µm), itsbasic principle of operation could also beapplied to other spectral bands (corre-sponding to other temperature ranges) inwhich the atmosphere is transparent andin which design requirements for sensitiv-ity and temperature-measurement accu-racy could be satisfied.

The underlying principle of nullingand self-calibration is the same as that of atypical microwave radiometer, but be-cause of differences between the charac-teristics of signals in the infrared and mi-crowave spectral regions, the principlemust be implemented in a different way.The instrument (see figure) would include an infrared photodetectorequipped with focusing input optics [e.g.,lens(es) and/or mirrors] and an inputLWIR band-pass filter. An optomechani-cal device would periodically chop theinput to the photodetector between a ref-erence gray body (ideally, a blackbody)and a test body (the temperature of whichone seeks to measure). The referencebody would be mounted on or in a heater,a thermoelectric cooler, or other temper-ature-control device suited to the particu-lar application. If needed, a band-passLWIR filter could be placed in front of thephotodetector.

The AC component of the photodetec-tor output would be fed to a lock-in am-plifier. The reference or synchronizationsignal for the lock-in amplifier would bederived from a device that monitored themotion of the chopper. The output of thelock-in amplifier would be a rectified sig-nal approximately proportional to the dif-ference between the radiances of the testbody and the reference body. This signalwould be used as an error signal in a feed-

back control loop, which would adjust thepower supplied to the temperature-con-trol device and thereby adjust the temper-ature of the reference body in an effort toreduce the error to zero. The temperatureof the reference body would be measuredby any of a variety of commercially avail-able contact temperature sensors, whichcan routinely afford accuracy and long-term stability within 0.1 K. Hence, as longas the error voltage remained at zero andassuming that the emissivities and otherradiant properties of the test and refer-ence bodies were sufficiently similar, itcould be assumed that the measured ref-erence-body temperature was a close ap-

proximation of the test-body temperature.Initial development efforts have been

concentrated on the chopper, which is amajor innovative feature of the instru-ment design. The desired characteristicsof the chopper include pure chopping,low power consumption (a few milli-watts), high reliability (millions of cycles),and a chopping frequency of severalhertz. “Pure chopping” signifies that atany given instant of time, the infrared ra-diation incident on the photodetectorwould be that from the test body and/orthe reference body and not from thechopper itself and that, moreover, duringone designated portion of the chopping

Lock-In Amplifier Controller

Power Amplifier

Chopper

Reference Gray- or Blackbody

Temperature Sensor

ThermoelectricCooler

Photodetectorand Optics

PhotodetectorOutput

Test Body

Synchronizing SignalFrom Chopper

From TestBody

FromReference

Body

From TestBody

LeafSprings

Reflective Prism

FromReference

Body

LensPhotodetector

RADIOMETER

ENLARGED, SIMPLIFIED VIEW OF CHOPPER AND PHOTODETECTORAT THREE POINTS IN THE CHOPPING CYCLE

This Infrared Radiometer would exploit a nulling, self-calibrating principle that, heretofore, has beenthe basis of design and operation of many microwave radiometers. The leaf-spring resonator designof the chopper would be an important element of the overall design of the instrument.

6 NASA Tech Briefs, June 2003

cycle, the photodetector would “see” onlythe test body while during another desig-nated portion of the cycle it would “see”only the reference body.

The chopper design under considera-tion at the time of reporting the informa-tion for this article calls for an electro-mechanical resonator comprising twopermanent magnets and a reflective

prism mounted on leaf springs. Each per-manent magnet would interact with oneof two electromagnet coils. One electro-magnet coil would be driven by amplifierto excite vibrations. The other electro-magnet coil would serve as a pickup coil,providing feedback to the amplifier to setup oscillations at the mechanical reso-nance frequency.

This work was done by Robert Ryan of Lock-heed Martin Corp. for Stennis Space Center.

This invention is owned by NASA, and apatent application has been filed. Inquiriesconcerning nonexclusive or exclusive licensefor its commercial development should be ad-dressed to the Intellectual Property Manager,Stennis Space Center; (228) 688-1929. Referto SSC-00124.

The Ames Power Monitoring SystemPower demand can be managed to reduce cost.Ames Research Center, Moffett Field, California

The Ames Power Monitoring System(APMS) is a centralized system of powermeters, computer hardware, and special-purpose software that collects and storeselectrical power data by various facilitiesat Ames Research Center (ARC). This sys-tem is needed because of the large andvarying nature of the overall ARC powerdemand, which has been observed torange from 20 to 200 MW. Large portionsof peak demand can be attributed to onlythree wind tunnels (60, 180, and 100 MW,respectively). The APMS helps ARC avoidor minimize costly demand charges by en-abling wind-tunnel operators, test engi-neers, and the power manager to monitortotal demand for center in real time.These persons receive the informationthey need to manage and schedule en-ergy-intensive research in advance and toadjust loads in real time to ensure that theoverall maximum allowable demand isnot exceeded.

The APMS (see figure) includes aserver computer running the WindowsNT operating system and can, in princi-ple, include an unlimited number ofpower meters and client computers. Asconfigured at the time of reporting theinformation for this article, the APMSincludes more than 40 power metersmonitoring all the major research facili-ties, plus 15 Windows-based client per-sonal computers that display real-timeand historical data to users via graphicaluser interfaces (GUIs). The power me-ters and client computers communicatewith the server using Transmission Con-trol Protocol/Internet Protocol (TCP/IP) on Ethernet networks, variously,through dedicated fiber-optic cables orthrough the pre-existing ARC local-areanetwork (ARCLAN).

The APMS has enabled ARC toachieve significant savings ($1.2 mil-lion in 2001) in the cost of power and

electric energy by helping personnel tomaintain total demand below monthlyallowable levels, to manage the overallpower factor to avoid low power factorpenalties, and to use historical systemdata to identify opportunities for addi-tional energy savings. The APMS alsoprovides power engineers and electri-cians with the information they need toplan modifications in advance and perform day-to-day maintenance of the ARC electric-power distribution system.

This work was done by Leonid Osetinsky ofJacobs/Sverdrup Technology and DavidWong of Ames Research Center. For fur-ther information, please contact:

Leonid Osetinsky, P.E.Jacobs/Sverdrup, Ames R&D GroupNASA Ames Research CenterTel. No.: (650)604-2396E-mail: [email protected]

The APMS collects, stores, and displays data on the consumption of electric power in major subsystems of the ARC power-distribution system.

ARCLAN

ARC-ENG Domain

BuildingGUI

BuildingGUI

BuildingGUI

ARCLAN Server

APMS Server

Facility GUI

Facility GUI

Facility Power Meters

Facility Power Meter

APMS Domain

BuildingGUI

BuildingPower Meter

BuildingPower Meter

BuildingPower Meter

BuildingPower Meter


Hot-film sensors, consisting of a metal-lic film on an electrically nonconductivesubstrate, have been used to measure skinfriction as far back as 1931. A hot film ismaintained at an elevated temperaturerelative to the local flow by passing anelectrical current through it. The powerrequired to maintain the specified tem-perature depends on the rate at whichheat is transferred to the flow. The heat-

transfer rate correlates to the velocity gra-dient at the surface, and hence, with skinfriction. The hot-film skin friction mea-surement method is most thoroughly de-veloped for steady-state conditions, butadditional issues arise under transientconditions.

Fabricating hot-film substrates usinglow-thermal-conductivity ceramics canoffer advantages over traditional quartz or

polyester-film substrates. First, a low con-ductivity substrate increases the fractionof heat convected away by the fluid, thusincreasing sensitivity to changes in flowconditions. Furthermore, the two-part,composite nature of the substrate allowsthe installation of thermocouple junc-tions just below the hot film, which canprovide an estimate of the conductionheat loss.

12.7 mm

15.9 mm

ThermocoupleLeads

Wall HRSI

RCG

Hot Film (ThicknessExaggerated for Visibility)

VelocityProfile

Sleeve ofMachinable

Ceramic

CROSS SECTION OF SENSOR INSTALLED IN WALL PHOTOGRAPH OF SENSOR AND LEADS

Hot Films on Ceramic Substrates for Measuring Skin FrictionLow-thermal-conductivity ceramic substrates, based on Space Shuttle tile technology, serve to increase sensitivity.Dryden Flight Research Center, Edwards, California

Figure 1. The Composite Ceramic Substrate of this hot-film sensor reduces conduction losses and increases sensitivity.

325

350342

333

325

308

308

308

308

316 316

308316333

325

350342

333

325

308

308

308

308

316 316

308316333

QUARTZ SUBSTRATE

308

308

308325

342308

308

308325

342

COMPOSITE CERAMIC SUBSTRATE

Figure 2. Steady-State Temperature Contours, determined from conjugate heat-transfer analyses, illustrate the effect of the lower thermal conductivity ofthe composite ceramic substrate relative to a quartz substrate. Temperatures are indicated in °C.


Figure 1 depicts a hot-film sensor ofthis type. The substrate is primarily com-posed of high-temperature reusable shut-tle insulation (HRSI), a lightweight (den-sity = 352 kg/m3), porous, ceramicmaterial originally developed to protectthe space shuttle from aerodynamic heat-ing. A hard, non-porous coat of reaction-cured glass (RCG) extends over the faceof the cylinder and about one-third of theway down the side providing a surface onwhich the metallic hot film and its leadscan be deposited. Small-diameter [0.005in. (0.127 mm)] thermocouple wires arerouted through the HRSI. Small groovesin the end of the HRSI cylinder, form thelands of the thermocouples and are deepenough such that the wires lie flush withthe HRSI surface prior to being coatedwith the RCG. The three thermocouplejunctions are placed in a line. The sub-strates are placed in a machinable-ce-

ramic sleeve that provides electrical isola-tion for the hot-film leads. Type R ther-mocouples must be used because thehigh firing temperature of the RCG coat-ing precludes the use of the more-sensi-tive thermocouples of type K’s.

The hot film itself is approximately0.004 in. (≈0.102 mm) wide and 1/4 in.(6.35 mm) long. Fabrication of the hotfilm and its leads begins with hand paint-ing the desired pattern using organo-metallic inks. The painted substrate isthen heated in an oven, which removesthe solvents from the ink leaving only agold-alloy film (see Figure 1 photo). Thesensor thermocouples provide feedbackcontrol to the oven. These techniquescould be used for the fabrication ofother temperature and heat-flux gaugeson high-temperature ceramics.

Conjugate heat-transfer analyses wereperformed on different substrate mate-

rials in air at moderate velocity gradients(7,500 s–1). For the composite ceramicsubstrate, the ratio of heat leaving thesensor via convection to total heat pro-duced is about 4 times higher than for aquartz substrate. Figure 2 depicts steady-state temperature contours for quartzand a composite ceramic substrate.Preliminary bench tests comparing hotfilms on composite ceramic and machin-able-ceramic substrates indicate that, atoverheat ratios of 1.2 and in horizontalorientations, the higher conductivitymachinable-ceramic substrates requireover 2.5 times the power.

This work was done by Greg Noffz of Dry-den Flight Research Center, Daniel Leiserof Ames Research Center, Jim Bartlett of Lan-gley Research Center, and Adrienne Lavine ofUCLA. For further information, contact theDryden Commercial Technology Office at(661) 276-3689. DRC-01-48

Probe Without Moving Parts Measures Flow AngleFlow angle is computed from forces measured by use of strain gauges.Dryden Flight Research Center, Edwards, California

The measurement of local flow angle iscritical in many fluid-dynamic applica-tions, including the aerodynamic flighttesting of new aircraft and flight systems.Flight researchers at NASA Dryden FlightResearch Center have recently devel-oped, flight-tested, and patented theforce-based flow-angle probe (FLAP), anovel, force-based instrument for themeasurement of local flow direction.Containing no moving parts, the FLAPmay provide greater simplicity, improvedaccuracy, and increased measurement ac-cess, relative to conventional moving-vane-type flow-angle probes.

Forces in the FLAP can be measuredby various techniques, including thosethat involve conventional strain gauges(based on electrical resistance) andthose that involve more advanced straingauges (based on optical fibers). A cor-relation is used to convert force-mea-surement data to the local flow angle.The use of fiber optics will enable theconstruction of a miniature FLAP, lead-ing to the possibility of flow measure-ment in very small or confined regions.This may also enable the “tufting” of asurface with miniature FLAPs, capable ofquantitative flow-angle measurements,similar to attaching yarn tufts for qualita-tive measurements.

The prototype FLAP was a small,aerodynamically shaped, low-aspect-

ratio fin about 2 in. (≈5 cm) long, 1 in.(≈2.5 cm) wide, and 0.125 in. (≈0.3 cm)thick (see Figure 1). The prototypeFLAP included simple electrical-resis-tance strain gauges for measuringforces. Four strain gauges weremounted on the FLAP; two on theupper surface and two on the lower sur-face. The gauges were connected to

form a full Wheatstone bridge, config-ured as a bending bridge.

In preparation for a flight test, theprototype FLAP was mounted on the air-data boom of a flight-test fixture (FTF)on the NASA Dryden F-15B flight re-search airplane. The FTF is an aerody-namic fixture for flight-research experi-ments that is carried underneath the

Figure 1. The Prototype FLAP was a fin instrumented with simple electrical-resistance strain gauges.


F-15B fuselage (see Figure 2). Measure-ment data were collected as the FLAPwas flown on the F-15B at subsonic andsupersonic speeds up to mach 1.7 and al-titudes up to 45,000 ft (≈13.7 km). FLAPdata were also collected under high-angle-of-attack and high-vertical-acceler-ation flight conditions. The flight dataanalyzed to date have verified the feasi-bility of the FLAP concept.

In a second-generation FLAP nowunder development, the electrical-resistance-strain-gauge force-measure-ment system of the prototype FLAP is re-placed with a fiber-optic-strain-gaugeforce-measurement system. This FLAPwill also be flown on the NASA DrydenF-15B airplane.

This work was done by Stephen Corda andM. Jake Vachon of Dryden Flight ResearchCenter. Inquiries concerning nonexclusiveor exclusive license for its commercial develop-ment should be addressed to Dryden Flight Re-search Center, Commercial Technology Office,(661) 276-3689. Refer to DRC-01-09.

Conventional Flow-Angle Vanes

Airflow

F-15B FTFAir-Data Boom

FLAP

Figure 2. The Air-Data Boom of the F-15B FTF was used to carry the FLAP in a flight test.

A method that can be implementedwith relatively simple electronic cir-cuitry provides a capability for detect-ing leakage of an electrically conduc-tive liquid from an electricallynonconductive underground pipe. Al-ternatively or in addition, the methodcan be applied to locate the pipe,whether or not there is a leak. Althoughthe method is subject to limitations(some of which are described below), itis still attractive as an additional optionfor detecting leaks and locating pipeswithout need for extensive digging.

The method is based on capacitivecoupling of an alternating electricalsignal from the liquid to a portableelectronic unit that resembles a metaldetector. A signal voltage is applied tothe liquid at some convenient pointalong the pipe: for example, the signalcould be coupled into the liquid via anaboveground metal pipe fitting, the in-terior surface of which is in contactwith the liquid. The signal is conductedthrough the liquid in the pipe; in thecase of diffusive leak of liquid into thesurrounding ground, the signal is con-ducted through the leak, into the por-tion of the adjacent ground that has

become soaked with the liquid. (A dripleak cannot be detected by this methodbecause there is no conductive path be-tween the liquid inside and the liquidoutside the pipe.)

The portable unit includes an electri-cally conductive plate connected to theinput terminal of an amplifier. Whenthe plate is brought near the pipe orthe leaked liquid, a small portion of thesignal power is coupled capacitivelyfrom the liquid to the plate. The userscans the plate near the ground surfaceto find the locus of maximum signalstrength. The leak can be identified as arelatively wide area, contiguous withthe location of the pipe, over which thesignal is detectable.

In order for this method to work, theliquid must be sufficiently conductive,and must be significantly more con-ductive than the ground is. Thus, forexample, the method does not workfor pure water, which is nonconductive,and does not work where the groundhas been soaked by a source other thana leak (e.g., heavy rain). It should bepossible to apply this method to, for ex-ample, common polyvinyl chloride(PVC) pipes that contain impure water

(e.g., swimming-pool water) leakinginto fairly dry ground.

The resistance of a typical column ofwater in a PVC pipe is of the order ofmegohms. The combination of thisorder of magnitude of resistance withthe order of magnitude of capacitancein a typical practical case dictates theuse of a signal frequency or frequen-cies no higher than the low kilohertzrange. Using the audibility of signals inthe frequency range to make a virtueout of necessity, one could feed the de-tector-amplifier output to a set of ear-phones so that the user could keep vi-sual attention focused on scanning theplate of the portable unit while listen-ing for the signal.

This work was done by Rober t C.Youngquist of Kennedy Space Center.Further information is contained in a TSP (seepage 1).

Inquiries concerning rights for the com-mercial use of this invention should be ad-dressed to the Technology Programs andCommercialization Office, Kennedy SpaceCenter, (321) 867-8130. Refer to KSC-12255.

Detecting Conductive Liquid Leaking From Nonconductive PipeA capacitive detector is scanned over the ground above the pipe.John F. Kennedy Space Center, Florida


Computers/Electronics

Adaptive Suppression of Noise in Voice CommunicationsA digital signal processor effects SNR-dependent spectral subtraction.John F. Kennedy Space Center, Florida

A subsystem for the adaptivesuppression of noise in a voice-communication system effects ahigh level of reduction of noisethat enters the system through mi-crophones. The subsystem in-cludes a digital signal processor(DSP) plus circuitry that imple-ments voice-recognition and spec-tral-manipulation techniques.

The development of the adap-tive noise-suppression subsystemwas prompted by the followingconsiderations: During process-ing of the space shuttle atKennedy Space Center, voicecommunications among testteam members have been signifi-cantly impaired in several in-stances because some test partici-pants have had to communicatefrom locations with high ambientnoise levels. Ear protection forthe personnel involved is com-mercially available and is used insuch situations. However, com-mercially available noise-cancel-ing microphones do not providesufficient reduction of noise thatenters through microphones andthus becomes transmitted on out-bound communication links.

In operation, noise or noise-cor-rupted speech enters the adaptive noise-suppression subsystem through a micro-phone. The output of the microphoneis sent through a high-gain amplifierand an antialiasing low-pass filter. Theoutput of the filter is sampled by an ana-log-to-digital converter.

At this point, the DSP suppressesnoise by executing a spectral-subtractionalgorithm that incorporates a depen-dence on the signal-to-noise ratio(SNR). The noise level and the SNR foruse in the algorithm are determined bya subalgorithm that includes the follow-ing steps: First, each frame of raw dataput out by the analog-to-digital con-verter is examined to determinewhether it is a voiced or an unvoicedframe. An estimate of the noise is ob-tained during each unvoiced frame. Arunning average of the noise is then

computed and used to approximate theexpected value of the noise.

Using the SNR computed as describedabove, the SNR-dependent algorithmpre-emphasizes the frequency compo-nents of the input signal that contain theconsonant information in humanspeech. The algorithm then determinesthe SNR and adjusts the proportion ofspectral subtraction accordingly. Afterspectral subtraction, de-emphasis filter-ing is performed and low-amplitude sig-nals are squelched. The resulting digitalsignal is processed through a digital-to-analog converter, then through asmoothing/voice-band filter, which is aband-pass filter with low and high 3-dBroll-off frequencies of 300 Hz and 3 kHz,respectively. The resulting analog signalis used to modulate a transmitter in thecommunication system.

In a demonstration of the adaptivenoise-suppression system, the words

“test, one, two, three, four, five” werespoken into the microphone. The fig-ure contains graphs of the original sam-pled noise-corrupted speech signal andthe signal after spectral subtraction.Spectral subtraction increased the SNRby approximately 20 dB. A listening testverified that the original noise was vir-tually eliminated, and that little or nodistortion in the form of musical noisewas introduced.

This work was done by David Kozel of Pur-due University, James A. DeVault ofKennedy Space Center, and Richard B.Birr formerly of I-NET, Inc. Further informa-tion is contained in a TSP (see page 1).

This invention is owned by NASA, and apatent application has been filed. Inquiriesconcerning nonexclusive or exclusive licensefor its commercial development should be ad-dressed to the Technology Programs and Com-mercialization Office, Kennedy Space Center,(321) 867-6373. Refer to KSC-11937.

0–4

–2

0

2

4

1.0 2.0Time, Seconds

NOISE-CORRUPTED SPEECH

NOISE-CORRUPTED SPEECH AFTER SPECTRAL SUBTRACTION

3.0 4.0 5.0

0–4

–2

0

2

4

1.0 2.0Time, Seconds

Am

plitu

de, V

olts

Am

plitu

de, V

olts

3.0 4.0 5.0

The Noise in a Noisy Speech Signal was reduced nearly to zero by the spectral-subtraction process.


Integrated arrays of microscopic solid-state batteries have been demonstratedin a continuing effort to develop micro-scopic sources of power and of voltagereference circuits to be incorporatedinto low-power integrated circuits. Per-haps even more importantly, arrays ofmicroscopic batteries can be fabricatedand tested in combinatorial experiments

directed toward optimization and discov-ery of battery materials.

The value of the combinatorial ap-proach to optimization and discovery hasbeen proven in the optoelectronic, phar-maceutical, and bioengineering indus-tries. Depending on the specific applica-tion, the combinatorial approach caninvolve the investigation of hundreds or

even thousands of different combina-tions; hence, it is time-consuming andexpensive to attempt to implement thecombinatorial approach by building andtesting full-size, discrete cells and batter-ies. The conception of microbattery ar-rays makes it practical to bring the ad-vantages of the combinatorial approachto the development of batteries.

Microbatteries for Combinatorial Studies of ConventionalLithium-Ion BatteriesThousands of combinations of battery materials can be evaluated economically.NASA’s Jet Propulsion Laboratory, Pasadena, California

The figure shows one of four solid-state power amplifiers, each capable ofgenerating an output power ≥240 mWover one of four overlapping frequencybands from 71 to 106 GHz. (The bandsare 71 to 84, 80 to 92, 88 to 99, and 89to 106 GHz.) The amplifiers are de-signed for optimum performance at atemperature of 130 K. These amplifierswere developed specifically for incorpo-ration into frequency-multiplier chainsin local oscillators in a low-noise, far-in-frared receiving instrument to belaunched into outer space to make as-trophysical observations. The designs ofthese amplifiers may also be of interestto designers and manufacturers of ter-restrial W-band communication andradar systems.

Each amplifier includes a set of sixhigh-electron-mobility transistor(HEMT) GaAs monolithic microwaveintegrated-circuit (MMIC) chips, mi-crostrip cavities, and other componentspackaged in a housing made from A-40silicon-aluminum alloy. This alloy waschosen because, for the original in-tended spacecraft application, it offersan acceptable compromise among thepartially competing requirements forhigh thermal conductivity, low mass,and low thermal expansion. Problemsthat were solved in designing the am-plifiers included designing connectorsand packages to fit the available space;designing microstrip signal-power split-ters and combiners; matching of im-pedances across the frequency bands;

matching of the electrical characteris-tics of those chips installed in parallelpower-combining arms; control and lev-elling of output power across the bands;and designing the MMICs, microstrips,and microstrip cavities to suppress ten-dencies toward oscillation in severalmodes, both inside and outside the de-sired frequency bands.

This work was done by Todd Gaier, LoreneSamoska, Mary Wells, Robert Ferber, JohnPearson, April Campbell, and Alejandro Per-alta of Caltech for NASA’s Jet PropulsionLaboratory and Gerald Swift, Paul Yocum,and Yun Chung of TRW, Inc. Further infor-mation is contained in a TSP (see page 1).NPO-30724

This Photograph Shows One of the Amplifiers described in the text. A WR-10 waveguide input port ison the left end. The output port is on the right end, facing away. DC input and sensing conductorsenter the package via a 21-pin connector. (Module dimensions: 20 × 49 × 60 mm.)

High-Performance Solid-State W-Band Power AmplifiersOutputs ≥240 mW are available at frequencies from 71 to 106 GHz.NASA’s Jet Propulsion Laboratory, Pasadena, California


Microbattery arrays (see figure) arefabricated on substrates by use of con-ventional integrated-circuit manufac-turing techniques, including sputter-ing, photolithography, and plasmaetching. The microbatteries incorpo-rate the same cathode materials ofinterest for conventional lithium-ionbatteries such as LiCoO2 and Li-CoxNi1–xO2. If multiple depositionsources are used in the fabrication of agiven array, then the chemical compo-sitions of battery components of inter-est can be varied across the substrate,making it possible to examine whatamounts to almost a continuum ofcompositions. For example, assuming

that a 4-in. (10-cm) substrate is used,the test-device pitch is 50 µm, and theconcentration gradient is 80 percentacross the substrate, then the composi-tions of adjacent test cells can be ex-pected to differ by only about 0.04 per-cent. Because thousands of test cellscan be fabricated in a single batch, itbecomes practical to test thousands ofcombinations of battery materials byuse of microbattery arrays, as con-trasted with only about 10 to 20 combi-nations by use of macroscopic cells.

The following is an example of a pro-cedure for fabricating an array of [LiCoxNi1–xO2 cathode/lithium phos-phorus oxynitride solid electrolyte/nickelanode] cells like those shown in the fig-ure, with a gradient in the cathodecomposition.1. A Ti film is deposited on a 4-in.

(≈10-cm) oxidized Si substrate. A Mofilm is subsequently deposited onthe Ti film.

2. The Mo-Ti bilayer is patterned by use ofphotolithography and wet etching todefine the cathode current collectors.

3. The substrate is patterned with thicknegative photoresist, with vias in thephotoresist opened over selectedareas of the current collectors.

4. A film of LiCoxNi1–xO2 cathode mate-rial is deposited on the substrate withthe desired gradient in x and selec-tively removed by use of a lift-offprocess, such that LiCoxNi1–xO2 ispresent only on the cathode currentcollectors.

5. A film of the solid electrolyte lithiumphosphorous oxynitride is deposited.

6. A film of Ni is deposited.7. The Ni film is patterned to define the

anode current collectors.8. The Ni film is selectively removed by

ion milling.9. A protective coat (for example, vapor-

deposited Parylene) is applied.All depositions described above are

performed by magnetron sputtering. Theprocedure can be readily modified toyield gradients in the cathode withother cations of interest: for example, LiCoxNi1–xO2 could be replaced withLiCoxNiyMn1–x–yO2 with the desired gra-dients in x and y.

The cells can be tested by use of a com-mercial semiconductor-parameter ana-lyzer connected to the test cells via tung-sten probe needles. By applying a currentof the order of 5 nA to a cell from the cath-ode to the anode, the cell can be chargedby oxidizing the cathode and reducing theLi at the anode. The charged cell can bedischarged by reversing the polarity of thecurrent. The cells can be tested in thesame manner as that used to test conven-tional lithium-ion cells to obtain informa-tion on such characteristics as cycle lifeand charge/discharge capacities, all as afunction of compositions of the cathode.

This work was done by William West, JayWhitacre, and Ratnakumar Bugga of Cal-tech for NASA’s Jet Propulsion Labora-tory. Further information is contained in aTSP (see page 1).NPO-21216

This is a Scanning Electron Micrograph of a mi-crobattery array. The lightly shaded contactpads are cathode current collectors. The darkercontact pads are anode current collectors. Thecathodes and solid electrolyte of each cell aresandwiched between the anode and cathodecontact pads and are not visible in this view.

Correcting for Beam Aberrations in a Beam-Waveguide AntennaThe transmitting feed is moved to compensate for movement of the target.NASA’s Jet Propulsion Laboratory, Pasadena, California

A method for correcting the aim of abeam-waveguide microwave antennacompensates for the beam aberration thatoccurs during radio tracking of a targetthat has a component of velocity trans-verse to the line of sight from the trackingstation. The method was devised primarilyfor use in tracking of distant target space-craft by large terrestrial beam-waveguideantennas of NASA’s Deep Space Network(DSN). The method should also be adapt-able to tracking, by other beam-waveguideantennas, of targets that move with largetransverse velocities at large distancesfrom the antennas.

The aberration effect arises when-ever a spacecraft is not moving along

the line of sight as seen from an an-tenna on Earth. In such a case, thespacecraft has a cross-velocity compo-nent, which is normal to the line-of-sight direction. In order to obtain op-timum two-way communication, theuplink and downlink beams must bepointed differently for simultaneousuplink and downlink communica-tions. At any instant of time, the down-link (or receive, Rx) beam must bepointed at a position where the space-craft was 1/2-round-trip light time(RTLT) ago, and the uplink (or trans-mit, Tx) beam must be pointed wherethe spacecraft will be in 1/2 RTLT(see figure). In the case of a high-

gain, narrow-beam antenna such as isused in the DSN, aiming the antennain other than the correct transmittingor receiving direction, or aiming at acompromise direction between thecorrect transmitting and receiving di-rections, can give rise to several dB ofpointing loss.

In the present method, the antenna isaimed directly at the apparent positionof the target, so that no directional cor-rection is necessary for reception of thesignal from the target. Hence, the effortat correction is concentrated entirely onthe transmitted beam. In physical terms,the correction is implemented by mov-ing the transmitting feed along a small


displacement vector chosen so that thedirection of the transmitted beam is al-tered by the small amount needed tomake the beam point to the anticipatedposition of the target at the anticipated

time of arrival of the transmitted signal atthe target.

The angular and linear coordinatesmentioned in the following sentences aredefined in the figure. The angular sepa-

ration between the transmitting and re-ceiving beams is described in terms of theseparation angle α and the clock angle β.The transmitting feed is mounted on anX-Y translation table. The problem is tocompute the polar coordinates ρ and φ ofthe amount by which the transmittingfeed must be displaced in the X-Y planeto move the direction transmitting beam,away from the direction of the receivingbeam, by the amounts of the required an-gular separation. The problem becomesone of computing the ρ and φ needed toobtain the required α and β. (Then therequired X and Y are calculated from ρand φ by simple trigonometry.)

The algorithm used to control the X-Ytable implements a closed-form repre-sentation of the coordinates ρ and φ asfunctions of the coordinates α and β.This representation can be obtained byexperimentation and/or physical-optics-based, computational-simulation studiesof electromagnetic scattering by the per-tinent antenna optics configuration forvarious combinations of ρ and φ. Therepresentation is of the general form

ρ = c1α + c2α2 andφ = φF – β + θEL – θAZ – nπ/2 radians,where c1 and c2 are coefficients deter-

mined by the computational study; fF is re-lated to the feed position on the floor; θELand θAZ are the elevation and azimuth an-gles, respectively; and n is one of the inte-gers between –1 and +2, determinedthrough measurements of beam offsetsobtained at known feed offsets.

This work was done by Manuel Franco,Stephen Slobin, and Watt Veruttipong of Cal-tech for NASA’s Jet Propulsion Labora-tory. Further information is contained in aTSP (see page 1).NPO-30534

YY

X

(x,y)

ReceivingFeed

at (0,0)

ReceivingFeed

at (x,y)

Transmitting Feed on X-Y Table

Receiving Feed

Azimuth Angle

Ele

vatio

n A

ngle

Predicted Position of TargetApparent Position of Target

Antenna Axis forReception andTransmission

BEAM-WAVEGUIDE ANTENNA WITH ABERRATION

ANGULAR-SEPARATIONCOORDINATES

COORDINATES FOR DISPLACEMENTSOF TRANSMITTING FEED

(α,β)(α,β)

ϕ

βα

ρ

ReceivingPosition

TransmittingPosition

t = 0

TransmittedBeam

ReceivedBeam

T ≡ Round Trip Time

t = – T2 t = + T

2

The Transmitting Feed Is Displaced in X and Y by amounts needed to point the transmitting beam awayfrom the receiving beam (which coincides with the antenna axis) by angular-separation amounts α and β.

Advanced Rainbow Solar Photovoltaic ArraysConcentrated sunlight is spectrally dispersed onto adjacent cells with different bandgaps.NASA’s Jet Propulsion Laboratory, Pasadena, California

Photovoltaic arrays of the rainbow type,equipped with light-concentrator andspectral-beam-splitter optics, have beeninvestigated in a continuing effort to de-velop lightweight, high-efficiency solarelectric power sources. This investigationhas contributed to a revival of the conceptof the rainbow photovoltaic array, whichoriginated in the 1950s but proved unre-alistic at that time because the selection ofsolar photovoltaic cells was too limited.Advances in the art of photovoltaic cells

since that time have rendered the con-cept more realistic, thereby promptingthe present development effort.

A rainbow photovoltaic array com-prises side-by-side strings of series-con-nected photovoltaic cells. The cells ineach string have the same bandgap,which differs from the bandgaps of theother strings. Hence, each string oper-ates most efficiently in a unique wave-length band determined by its bandgap.To obtain maximum energy-conversion

efficiency and to minimize the size andweight of the array for a given sunlight-input aperture, the sunlight incident onthe aperture is concentrated, then spec-trally dispersed onto the photovoltaic-array plane, whereon each string of cellsis positioned to intercept the light in itswavelength band of most efficient oper-ation. The number of cells in eachstring is chosen so that the output po-tentials of all the strings are the same;this makes it possible to connect the

strings together in parallel to maximizethe output current of the array.

According to the original rainbow pho-tovoltaic concept, the concentrated sun-light was to be split into multiple beams byuse of an array of dichroic filters designedso that each beam would contain light inone of the desired wavelength bands. Theconcept has since been modified to pro-vide for dispersion of the spectrum by useof adjacent prisms. A proposal for an ad-vanced version calls for a unitary concen-trator/spectral-beam-splitter optic in theform of a parabolic curved Fresnel-like

prism array with panels of photovoltaiccells on two sides (see figure). The sur-face supporting the solar cells can be ad-justed in length or angle to accommo-date the incident spectral pattern.

An unoptimized prototype assemblycontaining ten adjacent prisms andthree photovoltaic cells with differentbandgaps (InGaP2, GaAs, and InGaAs)was constructed to demonstrate feasibil-ity. The actual array will consist of a light-weight thin-film silicon layer of prismscurved into a parabolic shape. In an ini-tial test under illumination of 1 sun at

zero airmass, the energy-conversion effi-ciency of the assembly was found to be20 percent. Further analysis of the datafrom this test led to a projected energy-conversion efficiency as high as 41 per-cent for an array of 6 cells or strings(GaP, AlGaAs, InGaP2, GaAs, and twodifferent InGaAs cells or strings).

This work was done by Nick Mardesichand Virgil Shields of Caltech for NASA’s JetPropulsion Laboratory. Further informa-tion is contained in a TSP (see page 1).NPO-21051


A

A

VIEW A-A MAGNIFIED, SHOWING STRINGS OFPHOTOVOLTAIC CELLS

Series Connections

GaP

AlGaAs

InGaP2

GaAs

InGaAs, Bandgap 0.74 eV

InGaAs, Bandgap 0.62 eV

Sunlight

Panel ofPhotovoltaic

Cells

Fresnel-LikePrism Array

UV

IncidentSpectrum

IR

A Curved Fresnel Prism would concentrate and spectrally disperse sunlight onto adjacent strings of solar photovoltaic cells, each string optimized for thepart of the spectrum incident upon it.

Focal-plane arrays of quantum-well in-frared photodetectors (QWIPs) equippedwith both light-coupling diffraction grat-ings and metal side reflectors have beenproposed, and prototypes are expected tobe fabricated soon. The purpose servedby the metal side reflectors is to increasequantum efficiency by helping to traplight in the photosensitive material ofeach pixel.

The reasons for using diffraction grat-ings were discussed in several prior NASATech Briefs articles. To recapitulate: In anarray of QWIPs, the quantum-well layersare typically oriented parallel to the focalplane and therefore perpendicular ornearly perpendicular to the direction of in-cidence of infrared light. By virtue of theapplicable quantum selection rules, lightpolarized parallel to the focal plane (asnormally incident light is) cannot excite

charge carriers and, hence, cannot be de-tected. Diffraction gratings scatter nor-mally or nearly normally incident light intodirections more nearly parallel to the focalplane, so that a significant portion of thelight attains a component of polarizationnormal to the focal plane and, hence, can

excite charge carriers. Unfortunately, lightscattered in directions parallel or nearlyparallel to the focal plane can escape side-ways from the QWIP of a given pixel, as illustrated in Figure 1. The escaped lighthas made only a single pass through the in-terior photosensitive volume of the QWIP.

Metal Side Reflectors for Trapping Light in QWIPsQuantum efficiency would be increased because light would make multiple passes.NASA’s Jet Propulsion Laboratory, Pasadena, California

Diffraction Grating

Thin GaAs Substrate

QWIP Photosensitive Volume

Incident Infrared Light

One Pixel

Figure 1. Light Is Diffracted almost parallel to the focal plane for maximum quantum efficiency in a QWIP oftypical prior design. However, in the absence of reflectors like those shown in Figure 2, much of the diffractedlight is lost from the QWIP of a given pixel after a single pass through the photosensitive QWIP volume.


The quantum efficiency of theQWIP would be increased by trappinglight so that it makes multiple passesthrough the photosensitive volume. As

shown in Figure 2, the sides of theQWIP of each pixel would be coatedwith gold to reflect escaping light backinto the interior.

This work was done by Sarath Gunapala,Sumith Bandara, John Liu, and David Tingof Caltech for NASA’s Jet Propulsion Lab-oratory. Further information is contained ina TSP (see page 1).

In accordance with Public Law 96-517,the contractor has elected to retain title tothis invention. Inquiries concerningrights for its commercial use should be ad-dressed to

Intellectual Assets OfficeJPLMail Stop 202-2334800 Oak Grove DrivePasadena, CA 91109(818) 354-2240E-mail: [email protected] to NPO-30507, volume and number

of this NASA Tech Briefs issue, and thepage number.

QUIPPhotosensitive

Volume

Top MetalContact

DiffractionGrating

Indium Bump

SiN DielectricLayer

Au ReflectiveLayer

Detector CommonConductor

Thin GaAs Substrate

Incident Infrared Light

One Pixel One Pixel

Figure 2. Reflective Layers of Gold in a QWIP array of the type now under development would makethe light traverse the interior of each QWIP multiple times.


Software for CollaborativeEngineering of LaunchRockets

The Rocket Evaluation and CostIntegration for Propulsion and Engi-neering software enables collaborativecomputing with automated exchange ofinformation in the design and analysis oflaunch rockets and other complex sys-tems. RECIPE can interact with and in-corporate a variety of programs, includ-ing legacy codes, that model aspects of asystem from the perspectives of differenttechnological disciplines (e.g., aerody-namics, structures, propulsion, trajec-tory, aeroheating, controls, and opera-tions) and that are used by differentengineers on different computers run-ning different operating systems.RECIPE consists mainly of (1) ISCRM —a file-transfer subprogram that makes itpossible for legacy codes executed intheir original operating systems on theiroriginal computers to exchange dataand (2) CONES — an easy-to-use file-wrapper subprogram that enables the in-tegration of legacy codes. RECIPE pro-vides a tightly integrated conceptualframework that emphasizes connectivityamong the programs used by the collab-orators, linking these programs in amanner that provides some configura-tion control while facilitating collabora-tive engineering tradeoff studies, includ-ing “design to cost” studies. Incomparison with prior collaborative-en-gineering schemes, one based on theuse of RECIPE enables fewer engineersto do more in less time.

This program was written by Thomas TroyStanley of International Space Systems, Inc.,for Marshall Space Flight Center. Furtherinformation is contained in a TSP (see page 1).MFS-31692

Software Assists in Exten-sive Environmental Auditing

The Base Environmental Manage-ment System (BEMS) is a Web-based ap-plication program for managing andtracking audits by the EnvironmentalOffice of Stennis Space Center in con-formity with standard 14001 of the In-ternational Organization for Standard-ization (ISO 14001). (This standardspecifies requirements for an environ-mental-management system.) BEMS

saves time by partly automating whatwere previously manual processes forcreating audit checklists; recording andtracking audit results; issuing, tracking,and implementing corrective-action re-quests (CARs); tracking continuous im-provements (CIs); and tracking audit re-sults and statistics. BEMS consists of anadministration module and an auditormodule. As its name suggests, the ad-ministration module is used to adminis-ter the audit. It helps administrators toedit the list of audit questions; edit thelist of audit locations; assign mandatoryquestions to locations; track, approve,and edit CARs; and edit completed au-dits. The auditor module is used by au-ditors to perform audits and recordaudit results: it helps the auditors to cre-ate audit checklists, complete audits,view completed audits, create CARs,record and acknowledge CIs, and gener-ate reports from audit results.

This program was written by ChristopherCallac and Charlie Matherne of LockheedMartin Corp. for Stennis Space Center.

Inquiries concerning rights for the commer-cial use of this invention should be addressedto the Intellectual Property Manager, StennisSpace Center, (228) 688-1929. Refer to SSC-00180.

Software Supports Distributed Operations viathe Internet

Multi-mission Encrypted Communica-tion System (MECS) is a computer pro-gram that enables authorized, geograph-ically dispersed users to gain secureaccess to a common set of data files viathe Internet. MECS is compatible withlegacy application programs and a vari-ety of operating systems. The MECS ar-chitecture is centered around maintain-ing consistent replicas of data filescached on remote computers. MECSmonitors these files and, whenever oneis changed, the changed file is commit-ted to a master database as soon as net-work connectivity makes it possible to doso. MECS provides subscriptions for re-mote users to automatically receive newdata as they are generated. Remote userscan be producers as well as consumers ofdata. Whereas a prior program that pro-vides some of the same services treatsdisconnection of a user from the net-work of users as an error from which re-

covery must be effected, MECS treatsdisconnection as a nominal state of thenetwork: This leads to a different designthat is more efficient for serving manyusers, each of whom typically connectsand disconnects frequently and wantsonly a small fraction of the data at anygiven time.

This program was written by Jeffrey Norris,Paul Backes, and Robert Steinke of Caltechfor NASA’s Jet Propulsion Laboratory.Further information is contained in a TSP (seepage 1).

In accordance with Public Law 96-517,the contractor has elected to retain title to thisinvention. Inquiries concerning rights for itscommercial use should be addressed to



Software Estimates Costs ofTesting Rocket Engines

Simulation-Based Cost Model (SiCM),a discrete event simulation developed inExtend™, simulates pertinent aspects ofthe testing of rocket propulsion test arti-cles for the purpose of estimating thecosts of such testing during time inter-vals specified by its users. A user entersinput data for control of simulations; in-formation on the nature of, and activityin, a given testing project; and informa-tion on resources. Simulation objectsare created on the basis of this input.Costs of the engineering-design, con-struction, and testing phases of a givenproject are estimated from numbers andlabor rates of engineers and techniciansemployed in each phase, the duration ofeach phase; costs of materials used ineach phase; and, for the testing phase,the rate of maintenance of the testing fa-cility. The three main outputs of SiCMare (1) a curve, updated at each itera-tion of the simulation, that shows overallexpenditures vs. time during the intervalspecified by the user; (2) a histogram ofthe total costs from all iterations of thesimulation; and (3) table displayingmeans and variances of cumulative costs

Software


for each phase from all iterations. Otheroutputs include spending curves foreach phase.

This program was written by C. L. Smithof Lockheed Martin Corp. for Stennis SpaceCenter.

Inquiries concerning rights for the commercialuse of this invention should be addressed to theIntellectual Property Manager, Stennis SpaceCenter, (228) 688-1929. Refer to SSC-00168.

yourSky: Custom Sky-ImageMosaics via the Internet

yourSky (http://yourSky.jpl.nasa.gov)is a computer program that supplies cus-tom astronomical image mosaics of skyregions specified by requesters usingclient computers connected to the In-ternet. [yourSky is an upgraded versionof the software reported in “Software forGenerating Mosaics of Astronomical Im-ages” (NPO-21121), NASA Tech Briefs,Vol. 25, No. 4 (April 2001), page 16a.] Arequester no longer has to engage inthe tedious process of determining whatsubset of images is needed, nor even toknow how the images are indexed inimage archives. Instead, in response to arequester’s specification of the size andlocation of the sky area, (and optionallyof the desired set and type of data, reso-lution, coordinate system, projection,and image format), yourSky automati-cally retrieves the component image

data from archives totaling tens of ter-abytes stored on computer tape anddisk drives at multiple sites and assem-bles the component images into a mo-saic image by use of a high-performanceparallel code. yourSky runs on theserver computer where the mosaics areassembled. Because yourSky includes aWeb-interface component, no specialclient software is needed: ordinary Web-browser software is sufficient.

This program was written by Joseph Jacobof Caltech for NASA’s Jet Propulsion Lab-oratory. Further information is contained ina TSP (see page 1).

This software is available for commerciallicensing. Please contact Don Hart of the Cal-ifornia Institute of Technology at (818) 393-3425. Refer to NPO-30556.

Software for Managing Inventory of Flight Hardware

The Flight Hardware Support RequestSystem (FHSRS) is a computer programthat relieves engineers at Marshall SpaceFlight Center (MSFC) of most of thenon-engineering administrative burdenof managing an inventory of flight hard-ware. The FHSRS can also be adapted toperform similar functions for other or-ganizations. The FHSRS affords a com-bination of capabilities, including thoseformerly provided by three separate pro-

grams in purchasing, inventorying, andinspecting hardware. The FHSRS pro-vides a Web-based interface with a servercomputer that supports a relationaldatabase of inventory; electronic routingof requests and approvals; and elec-tronic documentation from initial re-quest through implementation of qual-ity criteria, acquisition, receipt,inspection, storage, and final issue offlight materials and components. Thedatabase lists both hardware acquiredfor current projects and residual hard-ware from previous projects. The in-creased visibility of residual flight com-ponents provided by the FHSRS hasdramatically improved the re-utilizationof materials in lieu of new procure-ments, resulting in a cost savings of over$1.7 million. The FHSRS includes sub-programs for manipulating the data inthe database, informing of the status of arequest or an item of hardware, andsearching the database on any physicalor other technical characteristic of acomponent or material. The softwarestructure forces normalization of thedata to facilitate inquiries and searchesfor which users have entered mixed orinconsistent values.

This program was designed and written byJohn Salisbury, Scott Savage, and ShirmanThomas of Cortez III for Marshall SpaceFlight Center. Further information is con-tained in a TSP (see page 1).MFS-31661


Materials

Lower-Conductivity Thermal-Barrier CoatingsAdditional stabilizers are incorporated into yttria-stabilized zirconia.John H. Glenn Research Center, Cleveland, Ohio

Thermal-barrier coatings (TBCs)that have both initial and post-exposurethermal conductivities lower than thoseof yttria-stabilized zirconia TBCs havebeen developed. TBCs are thin ceramiclayers, generally applied by plasmaspraying or physical vapor deposition,that are used to insulate air-cooledmetallic components from hot gases ingas turbine and other heat engines.Heretofore, yttria-stabilized zirconia(nominally comprising 95.4 atomic per-cent ZrO2 + 4.6 atomic percent Y2O3)has been the TBC material of choice.The lower-thermal-conductivity TBCsare modified versions of yttria-stabilizedzirconia, the modifications consistingprimarily in the addition of other ox-ides that impart microstructural anddefect properties that favor lower ther-mal conductivity.

TBCs are characterized by porosity,typically between 5 and 20 percent.Porosity reduces the thermal conductiv-ity of a TBC below the intrinsic conduc-tivity of a fully dense (that is, non-porous) layer of the TBC material. Thethermal conductivity of a TBC increasesas its porosity is reduced by the sinteringthat occurs during use at high tempera-ture. For future engines that will operateat higher gas temperatures, TBCs withgreater degrees of both initial insulating

capability and retention of insulating ca-pability will be needed.

The present lower-thermal-conductiv-ity TBCs are made of ZrO2 and Y2O3doped with additional oxides that arechosen to perform three functions:• Create thermodynamically stable,

highly defective lattice structures withtailored ranges of defect-cluster sizesto exploit the effectiveness of suchstructures as means of attenuating andscattering phonons, thus reducingthermal conductivity;

• Produce of highly distorted latticestructures with essentially immobile de-fect clusters and/or nanoscale orderedphases, which effectively reduces con-centrations of mobile defects andmovements of atoms, thus increasingsintering-creep resistance; and

• Exploit the formation of complexnanoscale clusters of defects to in-crease the measures of such desiredmechanical properties such as fracturetoughness.The additional oxides in a TBC accord-

ing to this concept are typically selected asa pair — one from each of two groups ofoxides denoted for this purpose as groupsA and B. Group A includes scandia(Sc2O3) and ytterbia (Yb2O3). These ox-ides are highly stable, and the radii oftheir trivalent cations are smaller than

those of the primary dopant yttria. GroupB includes neodymia (Nd2O3), samaria(Sm2O3), and gadolinia (Gd2O3) whichare also highly stable, and their trivalentcations are larger than those of yttria.

Like yttria, the A and B oxides are re-garded as stabilizers. Preferably, the totalstabilizer content (yttria + A oxide + Boxide) should lie between 4 and 50atomic percent. The concentration of yt-tria should exceed that of each of otherstabilizers, and the concentrations of theA and B oxides should be approximatelyequal. Formulations other than the fore-going preferred one are also possible:Variations include the use of alternativegroup-A oxides (e.g., MgO2, NiO,Cr2O3), the use of two or more group-Aand/or group-B oxides, substitution ofhafnia for zirconia, and substitution ofother primary stabilizers (e.g., dysprosiaor erbia) for yttria.

This work was done by Robert A. Millerof Glenn Research Center and Dong-ming Zhu of Ohio Aerospace Institute.Further information is contained in a TSP (seepage 1).

Inquiries concerning rights for the commer-cial use of this invention should be addressedto NASA Glenn Research Center, CommercialTechnology Office, Attn: Steve Fedor, MailStop 4–8, 21000 Brookpark Road, Cleve-land, Ohio 44135. Refer to LEW-17039.

Process for Smoothing an Si Substrate After Etching of SiO2Reactive-ion etching can be tailored to minimize undesired side effects.NASA’s Jet Propulsion Laboratory, Pasadena, California

A reactive-ion etching (RIE) process forsmoothing a silicon substrate has been de-vised. The process is especially useful forsmoothing those silicon areas that havebeen exposed by etching a pattern ofholes in a layer of silicon dioxide that cov-ers the substrate. Applications in whichone could utilize smooth silicon surfaceslike those produced by this process in-clude fabrication of optical waveguides,epitaxial deposition of silicon on selectedareas of silicon substrates, and prepara-

tion of silicon substrates for deposition ofadherent metal layers.

During etching away of a layer of SiO2that covers an Si substrate, a polymer be-comes deposited on the substrate, andthe substrate surface becomes rough(roughness height ≈ 50 nm) as a result ofover-etching or of deposition of the poly-mer. While it is possible to smooth a sili-con substrate by wet chemical etching,the undesired consequences of wetchemical etching can include compro-

mising the integrity of the SiO2 sidewallsand undercutting of the adjacent areasof the silicon dioxide that are meant tobe left intact.

The present RIE process results inanisotropic etching that removes thepolymer and reduces height of rough-ness of the silicon substrate to <10 nmwhile leaving the SiO2 sidewalls intactand vertical. Control over substrate ver-sus sidewall etching (in particular, pref-erential etching of the substrate) is


A proposed lightweight pressure ves-sel would be made of a composite ofhigh-tenacity continuous fibers and aflexible matrix material. The flexibilityof this pressure vessel would render it(1) compactly stowable for transportand (2) more able to withstand impacts,relative to lightweight pressure vesselsmade of rigid composite materials. Thevessel would be designed as a structuralshell wherein the fibers would be pre-dominantly bias-oriented, the orienta-

tions being optimized to make the fibersbear the tensile loads in the structure.Such efficient use of tension-bearingfibers would minimize or eliminate theneed for stitching and fill (weft) fibersfor strength. The vessel could be fabri-cated by techniques adapted from fila-ment winding of prior composite-mater-ial vessels, perhaps in conjunction withthe use of dry film adhesives. In additionto the high-bias main-body substructuredescribed above, the vessel would in-

clude a low-bias end substructure tocomplete coverage and react peak loads.Axial elements would be overlaid to con-tain damage and to control fiber orien-tation around side openings. Fiber ringstructures would be used as interfacesfor connection to ancillary hardware.

This work was done by Glen Brown, RoyHaggard, and Paul A. Harris of Vertigo, Inc.,for Johnson Space Center. Further informa-tion is contained in a TSP (see page 1).MSC-23020

Flexible Composite-Material Pressure VesselLyndon B. Johnson Space Center, Houston,Texas

achieved through selection of processparameters, including gas flow, power,and pressure. Such control is not uni-formly and repeatably achievable in wetchemical etching. The recipe for thepresent RIE process is the following:

Etch 1 — A mixture of CF4 and O2gases flowing at rates of 25 to 75 and 75to 125 standard cubic centimeters perminute (stdcm3/min), respectively;power between 44 and 55 W; and pres-sure between 45 and 55 mtorr (between

6.0 and 7.3 Pa). The etch rate lies be-tween ≈3 and ≈6 nm/minute.

Etch 2 — O2 gas flowing at 75 to 125stdcm3/min, power between 44 and 55W, and pressure between 50 and 100mtorr (between 6.7 and 13.3 Pa).

This work was done by Tasha Turner andChi Wu of Caltech for NASA’s Jet Propul-sion Laboratory. Further information iscontained in a TSP (see page 1).

In accordance with Public Law 96-517,the contractor has elected to retain title to this

invention. Inquiries concerning rights for itscommercial use should be addressed to

Intellectual Property groupJPLMail Stop 202-2334800 Oak Grove DrivePasadena, CA 91109(818) 354-2240Refer to NPO-20777, volume and number


Treatment To Destroy Chlorohydrocarbon Liquids in the GroundEmulsified iron is injected into the ground and left there.John F. Kennedy Space Center, Florida

A relatively simple chemical treatmentthat involves the use of emulsified ironhas been found to be effective in remedi-ating groundwater contaminated withtrichloroethylene and other densechlorohydrocarbon liquids. These liquidsare members of the class of dense, non-aqueous phase liquids (DNAPLs), whichare commonly recognized to be particu-larly troublesome as environmental con-taminants. The treatment converts theseliquids into less-harmful products.

As a means of remediation of contami-nated groundwater, this treatment takesless time and costs less than do traditionalpump-and-treat processes. At some sites,long-term leakage and/or dissolution ofchlorohydrocarbon liquids from poolsand/or sorbed concentrations in rock andsoil gives rise to a need to continue pump-and-treat processes for times as long asdecades in order to maintain protection ofhuman health and the environment. Incontrast, the effects of the emulsified-iron

treatment are more lasting, decreasing theneed for long-term treatment and moni-toring of contaminated areas.

The material used in this treatment con-sists of iron particles with sizes of the orderof nanometers to micrometers containedwithin the micelles of a surfactant-stabi-lized, biodegradable, oil-in-water emul-sion. The emulsion is simple to prepareand consists of relatively inexpensive andenvironmentally acceptable ingredients:One typical formulation consists of 1.3weight percent of a food-grade surfactant,17.5 weight percent of iron particles, 23.2weight percent of vegetable oil, and 58.0weight percent of water.

The emulsion is injected into theground via a push well. Free-phasechlorohydrocarbon molecules diffusethrough the oil membranes of the emul-sion particles to the surfaces of the ironparticles, where dehalogenation takesplace. The dehalogenation reactions gen-erate hydrocarbon byproducts (primarily

ethylene in the case of trichloroethyl-ene), which diffuse out of the emulsionmicelles and are benign in nature.

Experiments have demonstrated sev-eral aspects of the effectiveness of thistreatment by use of emulsified iron:• This treatment is more effective in de-

grading free-phase trichloroethylenethan is a similar treatment that usesonly pure iron particles.

• Emulsions containing iron can be in-jected into soil matrices, where theybecome immobilized and remain im-mobile, even in the presence of flow-ing water.

• Iron emulsions can exert an effectequivalent to pulling globules oftrichloroethylene into their micelles.

• No chlorinated byproducts from thedegradation of trichloroethylene passout of the micelles. The only degrada-tion products that have been ob-served to leave the iron emulsions areethylene as mentioned previously,


plus trace amounts of other innocu-ous hydrocarbons.Laboratory studies have shown that the

amount of emulsion needed to degrade agiven amount of trichloroethylene is ap-proximately eight times the mass of thetrichloroethylene. Because there are nocontinuing operating costs after the emul-sion has been injected through push

wells, the iron-emulsion treatment offers asubstantial economic advantage over thelong-term pump-and-treat method.

This work was done by Jacqueline Quinnof Kennedy Space Center and ChristianA. Clausen III, Cherie L. Geiger, Debra Rein-hart, and Kathleen Brooks of the Universityof Central Florida. Further information is con-tained in a TSP (see page 1).

This invention is owned by NASA, and apatent application has been filed. Inquiriesconcerning nonexclusive or exclusive licensefor its commercial development should be ad-dressed to the Technology CommercializationOffice, Kennedy Space Center, (321) 867-8130. Refer to KSC-12246.


Mechanics

A computational study has shown thatby injecting drops in jets of gas havingsquare, elliptical, triangular, or othernoncircular injection cross sections, itshould be possible to increase (relativeto comparable situations having circularcross section) the entrainment and dis-persion of liquid drops. This finding haspractical significance for a variety of ap-plications in which it is desirable to in-crease dispersion of drops. For example,

in chemical-process sprays, increaseddispersion leads to increases in chemi-cal-reaction rates; in diesel engines, in-creasing the dispersion of drops ofsprayed fuel reduces the production ofsoot; and in household and paint sprays,increasing the dispersion of dropsmakes it possible to cover larger sur-faces.

It has been known for some years thatsingle-phase fluid jets that enter flow

fields through noncircular inlets entrainmore fluid than do comparable jets en-tering through circular inlets. The com-putational study reported here was di-rected in part toward determiningwhether and how this superior mixingcharacteristic of noncircular single-phase jets translates to a similar benefitin cases of two-phase jets (that is, sprays).

The study involved direct numericalsimulations of single- and two-phase freejets with circular, elliptical, rectangular,square, and triangular inlet cross sec-tions. The two-phase jets consisted of gasladen with liquid drops randomly in-jected at the inlets. To address the moreinteresting case of evaporating drops,the carrier gas in the jets was specified tobe initially unvitiated by the vapor of theliquid chemical species and the initialtemperature of the drops was chosen tobe smaller than that of the gas. Themathematical model used in the studywas constructed from the conservationequations for the two-phase flow and in-cluded complete couplings of mass, mo-mentum, and energy based on thermo-dynamically self-consistent specificationof the enthalpy, internal energy, and la-tent heat of vaporization of the vapor.

The results of the numerical simula-tions yielded information on (1) the dif-ferent spreading behaviors occurring fordifferent inlet cross sections and (2) thedifferences between flow fields in thepresence and absence of liquid drops.The most important consequence of in-teraction of drops with the flows wasfound to be the production of enhancedstreamwise vorticity that alters entrain-ment and the mixing of species accord-ing to the inlet geometry. At the time sta-tion corresponding to steady-stateentrainment, the potential cores of two-phase jets were found to be shorter thantheir single-phase counterparts by anorder of magnitude (see figure).Whereas the two-phase circular jets werefound to exhibit symmetric entrainmentpatterns at a location well past thestreamwise locations of the potentialcores, the noncircular jets were found,at the same location, to depart stronglyfrom symmetry. The phenomenon of

Noncircular Cross Sections Could Enhance Mixing in SpraysPreliminary results suggest that elliptical cross sections may be best.NASA’s Jet Propulsion Laboratory, Pasadena, California

Circle

Square

Ellipse

Rectangle

Triangle

Circle

Square

Ellipse

Rectangle

Triangle

0

0–1.0

–0.5

0

0.5

1.0

–1.0

–0.5

0

0.5

1.0

0.1 0.2 0.3 0.4 0.5

1 2 3 4

SINGLE PHASE

TWO PHASE

x1/DJ

x 3/D

Jx 3

/DJ

x1/DJ

5 6 7 8

The Potential Core of a jet is defined as the region beyond which the velocity is no longer equal tothat at the inlet. These plots are outlines of computationally simulated potential cores of single- andtwo-phase jets issuing from inlets with the noted cross sections. DJ denotes the equivalent jet diame-ter, which, for a noncircular inlet, is defined as the diameter of a circular inlet of equal cross-sectionalarea. The symbols x1 and x3 denote Cartesian coordinates parallel and perpendicular, respectively, tothe initial jet axis.


upstream-vs.-downstream exchange ofmajor and minor axes of elliptical crosssections (“axis switching” for short) ofsingle-phase jets was not observed in thetwo-phase jets.

Considerations of the distributions ofthe number density of drops, liquidmass, and evaporated species distribu-

tions lead to recommending ellipticalcross sections as optimal ones in thatthey result in optimal combinations ofdispersion and mixing. All of the com-putations were performed for pre-transi-tional jets (that is, jets on the laminarside of the transition between laminarand turbulent flow). Further investiga-

tions would be necessary to elucidate theeffects of turbulence.

This work was done by Josette Bellan andHesham Abdel-Hameed of Caltech forNASA’s Jet Propulsion Laboratory. Fur-ther information is contained in a TSP (seepage 1).NPO-30400


Machinery/Automation

Small, Untethered, Mobile Robots for Inspecting Gas PipesThese robots would be powered by gas flows.NASA’s Jet Propulsion Laboratory, Pasadena, California

Small, untethered mobile robots de-noted gas-pipe explorers (GPEXs) havebeen proposed for inspecting the interi-ors of pipes used in the local distributionnatural gas. The United States has net-work of gas-distribution pipes with atotal length of approximately 109 m.These pipes are often made of iron andsteel and some are more than 100 yearsold. As this network ages, there is a needto locate weaknesses that necessitate re-pair and/or preventive maintenance.The most common weaknesses are leaksand reductions in thickness, which arecaused mostly by chemical reactions be-tween the iron in the pipes and varioussubstances in soil and groundwater.

At present, mobile robots called pigsare used to inspect and clean the interi-ors of gas-transmission pipelines. Somecarry magnetic-flux-leakage (MFL) sen-sors for measuring average wall thick-nesses, some capture images, and somemeasure sizes and physical conditions.The operating ranges of pigs are limitedto fairly straight sections of wide trans-mission-type (as distinguished from dis-tribution-type) pipes: pigs are too largeto negotiate such obstacles as bends withradii comparable to or smaller than pipediameters, intrusions of other pipes atbranch connections, and reductions indiameter at valves and meters. TheGPEXs would be smaller and would beable to negotiate sharp bends and otherobstacles that typically occur in gas-dis-tribution pipes.

Unlike a pig, a GPEX would be able tooperate or even fail inside a pipe withoutstopping the gas flow. It would be capable

of operating for long times and travelinglong distances without human interven-tion, so that expensive and disruptiveurban excavation could be kept to a min-imum; to make this possible, the GPEXwould generate its own power from theflow of gas. It would communicate infor-mation, including low-rate data like thosefrom an MFL sensor and high-rate imagedata that show corrosion, leaks, and buck-les. Two-way radio communication, forboth retrieval of inspection data and con-trol of the GPEX, would take place at gig-ahertz carrier frequencies, with pipes serv-ing as waveguides. The GPEX musttransform its shape as needed to copewith changes in pipe dimensions, dents,pipe bends, and pipe intrusions.

A prototype GPEX that would be capa-ble of operating inside either a 4-in. (10-cm) or a 6-in. (15-cm) pipe has beendeveloped. At present, it is viewed as in-feasible to package all the needed func-tions of the GPEX inside a single bodythat could negotiate right-angle turns in a10-cm pipe. Therefore, the prototypeGPEX is composed of a train of multipleunits, much like a railroad train. Becausethe source of power would be the flow ofnatural gas in the pipe to be inspected,the first unit in the train would be a com-bination of a power generator and loco-motive. The flow of gas [typically methaneat a flow speed of 10 m/s, a pressure of 60 psi (≈0.4 MPa), and a temperature of15 °C] could supply sufficient power toenable the robot to perform all its tasks,without a major loss of gas pressure.

The initial development effort is ex-pected to focus on the power-genera-

tor/locomotive unit. The power genera-tor would be a permanent-magnetmotor back-driven by a miniature tur-bine driven by the flow of gas. In normaloperation, power of at least 2 W wouldbe required, and the pressure dropwould have to be limited to no morethan 0.3 psi (2 kPa). A further require-ment is that in the event of worst-casefailure of the generator or any otherpart of the GPEX, the pressure drop notexceed 0.6 psi (4 kPa).

The power-generator/locomotive unitis a pair of bodies connected by an ex-tension actuator. Multiple bladders canbe alternately inflated and deflated toexert forces on the pipe wall, and the ex-tension actuator can increase or de-crease the separation between the bodiesto move the unit in “inchworm” fashion.

This work was done by Brian Wilcox ofCaltech for NASA’s Jet Propulsion Labor-atory. Further information is contained in aTSP (see page 1).





Manufacturing

Preparation of Regular Specimens for Atom ProbesSingle- or multiple-tip specimens can readily be prepared.NASA’s Jet Propulsion Laboratory, Pasadena, California

A method of preparation of speci-mens of non-electropolishable materialsfor analysis by atom probes is being de-veloped as a superior alternative to aprior method. In comparison with the

prior method, the present method in-volves less processing time. Also,whereas the prior method yields irregu-larly shaped and sized specimens, thepresent developmental method offers

the potential to prepare specimens ofregular shape and size.

The prior method is called themethod of sharp shards because it in-volves crushing the material of interest

Paint-Overspray CatcherTurning airflow and entrained droplets would be drawn away.Langley Research Center, Hampton, Virginia

An apparatus to catch paint oversprayhas been proposed. Overspray is an un-avoidable parasitic component of spraythat occurs because the flow of air or other

gas in the spray must turn at the sprayedsurface. Very small droplets are carriedaway in this turning flow, and some land onadjacent surfaces not meant to be painted.

The basic principle of the paint-spraycatcher is to divert the overspray into asuction system at the boundary of thearea to be painted. The paint-spraycatcher (see figure) would include atoroidal plenum connected through nar-row throat to a nozzle that would face to-ward the center of the torus, whichwould be positioned over the center ofthe area to be spray-painted. Theplenum would be supported by fourtubes that would also serve as suction ex-haust ducts. The downstream ends of thetubes (not shown in the figure) would beconnected to a filter on a suction pump.The pump would be rated to provide asuction mass flow somewhat greater thanthat of the directed spray gas stream, sothat the nozzle would take in a small ex-cess of surrounding gas and catch nearlyall of the overspray. A small raised lip atthe bottom edge of the nozzle wouldcatch paint that landed inside the nozzle.Even if the paint is directly pistonpumped, the droplets entrain an air flowby time they approach the wall, so thereis always a gas stream to carry the excessdroplets to the side. For long-durationspraying operations, it could be desirableto include a suction-drain apparatus toprevent overflowing and dripping ofpaint from inside the lip. A version with-out an external contraction and with thethroat angled downward would be amore compact version of catcher, al-though it might be slightly less efficient.

This work was done by Leonard M. Wein-stein of Langley Research Center. For moreinformation, contact the Langley CommercialTechnology Office at (757) 864-6005.LAR-15613

Area To Be Spray-Painted

TOP VIEW

SIDE VIEW

Plenum

Plenum

Nozzle

Sprayer

Supporting Arms

Surface To Be Spray-PaintedLipThroat

Suction Exhaust Tube and Support(One of Four)

Nozzle

To Suction Pump With Filter

Suction ExhaustTube and Support

To Suction Pump With Filter

SpraySpray

The Paint-Overspray Catcher would suck the turning flow of gas and entrained paint droplets, pre-venting the droplets from landing on non-target surfaces. The planform of the catcher plenum andnozzle need not be round as shown here: It could have any other convenient shape, depending onthe boundary of the area to be painted.

and selecting microscopic sharp shardsof the material for use as specimens.Each selected shard is oriented with itssharp tip facing away from the tip of astainless-steel pin and is glued to the tipof the pin by use of silver epoxy. Thenthe shard is milled by use of a focusedion beam (FIB) to make the shard verythin (relative to its length) and to makeits tip sharp enough for atom-probeanalysis. The method of sharp shards isextremely time-consuming because theselection of shards must be performedwith the help of a microscope, theshards must be positioned on the pinsby use of micromanipulators, and the ir-regularity of size and shape necessitatesmany hours of FIB milling to sharpeneach shard.

In the present method, a flat slab ofthe material of interest (e.g., a polishedsample of rock or a coated semiconduc-tor wafer) is mounted in the sampleholder of a dicing saw of the type con-ventionally used to cut individual inte-

grated circuits out of the wafers onwhich they are fabricated in batches. Asaw blade appropriate to the material ofinterest is selected. The depth of cutand the distance between successiveparallel cuts is made such that what isleft after the cuts is a series of thin, par-allel ridges on a solid base. Then theworkpiece is rotated 90° and the pat-tern of cuts is repeated, leaving behinda square array of square posts on thesolid base.

The posts can be made regular, long,and thin, as required for samples foratom-probe analysis. Because of theirsmall volume and regularity, the amountof FIB-milling time can be much lessthan that of the method of sharp shards.Individual posts can be broken off formounting in a manner similar to that ofthe method of sharp shards. Alterna-tively, the posts can be left intact on thebase and the base can be cut to a smallsquare (e.g., 3 by 3 mm) suitable formounting in an atom probe of a type ca-

pable of accepting multiple-tip speci-mens. The advantage of multiple-tipspecimens is the possibility of analyzingmany tips without the time-consuminginterchange of specimens.

This work was done by Kim Kuhlman andJames Wishard of Caltech for NASA’s JetPropulsion Laboratory. Further informa-tion is contained in a TSP (see page 1).


Intellectual Assets OfficeJPLMail Stop 202-2334800 Oak Grove DrivePasadena, CA 91109-8099(818) 354-2240E-mail: [email protected] to NPO-30667, volume and number



According to a proposal, basic x-raylithography would be extended to incor-porate a technique, called “inverse to-mography,” that would enable the fabri-cation of microscopic three-dimensional(3D) objects. The proposed inversetomo-lithographic process would makeit possible to produce complex shaped,submillimeter-sized parts that would bedifficult or impossible to make in anyother way. Examples of such shapes orparts include tapered helices, parabo-loids with axes of different lengths,and even Archimedean screws thatcould serve as rotors in microturbines.

The proposed inverse tomo-litho-graphic process would be based partlyon a prior microfabrication processknown by the German acronym “LIGA”(“lithographie, galvanoformung, abfor-mung,” which means “lithography, elec-troforming, molding”). In LIGA, onegenerates a precise, high-aspect ratiopattern by exposing a thick, x-ray-sensi-tive resist material to an x-ray beamthrough a mask that contains the pat-tern. One can electrodeposit metal intothe developed resist pattern to form aprecise metal part, then dissolve the re-sist to free the metal. Aspect ratios of

100:1 and patterns into resist thicknessesof several millimeters are possible.

Typically, high-molecular-weight poly(methyl methacrylate) (PMMA) is used asthe resist material. PMMA is an excellent

resist material in most respects, its majorshortcoming being insensitivity. Conven-tional x-ray sources are not practical forLIGA work, and it is necessary to use asynchrotron as the source. Because syn-

Inverse Tomo-Lithography for Making Microscopic 3D PartsInverse tomography would be used to generate complex three-dimensional patterns.NASA’s Jet Propulsion Laboratory, Pasadena, California

Mask ContainingSinusoidal

Absorber Pattern

X-Rays

RotatingPMMA Rod

Motion ofMask

DEVELOPMENT

EXPOSURE TO X-RAYS

HELIX REMAININGAFTER DEVELOPMENT

A Rotating PMMA Rod would be exposed to collimated x-rays through a mask bearing a sinusoidalabsorber pattern while the mask moved along the rod in synchronism with the rotation. Upon devel-opment of the PMMA (used here as an x-ray photoresist material), a helix would remain.


chrotron radiation is highly collimatedand its wavelength of synchrotron radia-tion is typically <5 Å, there is very littlediffraction and the pattern of a high-con-trast mask is projected deep into a resistwith nearly perfect vertical sidewalls. Ofcourse, the only three-dimensional shapethat can be formed in this way is thelocus of points generated by moving themask pattern along the direction of inci-dence of the radiation.

In a recently developed variant ofLIGA, a rotating PMMA rod is exposed tox-rays through a stationary mask; this tech-nique can be used to make axisymmetricstructures; e.g., objects shaped like wineglasses or baseball bats. The proposedtechnique would also involve stenciling anx-ray image into a rotating PMMA rod,but would differ from prior techniques in

that the mask would be moved in syn-chronism with the rod to generate athree-dimensional pattern. The synchro-nized motions of the mask and rod wouldbe generated by translation and rotationstages actuated by stepping motors undercontrol by a computer.

Describing the x-ray exposure tech-nique in different words, a changingtwo-dimensional pattern would be pro-jected into a three-dimensional one. Intomography, one decodes a three-di-mensional pattern from the changingtwo-dimensional pattern obtained by il-luminating it from a changing direction.In the proposed technique, one wouldessentially reverse this decoding process;that is, one would encode or construct athree-dimensional pattern by illuminat-ing the region of interest in a changing

two-dimensional pattern: That is why theproposed x-ray exposure technique iscalled “inverse tomography.”

The figure depicts an example of theuse of this technique to generate a sim-ple helix. The two-dimensional projec-tion (shadow) of a helix is a sinusoid. Toform the helical pattern in a PMMA rod,one would project x-rays perpendicu-larly toward the rod through a mask witha sinusoidal pattern while rotating therod and translating the mask along therod at a speed of one wavelength of thesinusoid per rotation period.

This work was done by Victor White andDean Wiberg of Caltech for NASA’s JetPropulsion Laboratory. Further informa-tion is contained in a TSP (see page 1).NPO-20593


A method of predicting and pre-venting incipient flameout in a com-bustor has been proposed. Themethod should be applicable to a va-riety of liquid- and gas-fueled com-bustors in furnaces and turbine en-gines. Until now, there have beenmethods of detecting flameouts afterthey have occurred, but there hasbeen no way of predicting incipientflameouts and, hence, no way of act-ing in time to prevent them. Preven-tion of flameout could not only pre-vent damage to equipment but, inthe case of aircraft turbine engines,could also save lives.

For any combustor, excessive depar-ture from optimum operating condi-tions can lead to instability thatquickly ends in flameout. Of particu-lar interest is that for a given tempera-ture and pressure of incoming air,flameout can occur if the fuel/airratio is too high or too low. In manycases, combustors are operated nearertheir lean-mixture (low fuel/air ratio)stability boundaries.

Studies have shown that as a com-bustor approaches instability, pressurefluctuations increase sharply. Onemeasure of such fluctuations that hasbeen found to be especially useful isthe ratio between the magnitude of pres-sure fluctuations (essentially, the acousticpressure) and the time-averaged pressure.Alternatively, one could detect instability-related pressure fluctuations by detectingthe sudden appearance of one or moreacoustic spectral peak(s) at frequenciesknown to be associated with instabilities inthe particular combustor (see figure).

The proposed method is based largelyon the foregoing observations. It callsfor continuous monitoring and analysisof the acoustic pressure and severalother key physical parameters that areindicative of the state of the combustionprocess. The instrumentation needed toimplement the method in a typical in-stallation would include the following:• Meters to measure the flows of fuel

and air into the combustor;• Gauges to measure the pressures of

the entering fuel and air;

• Thermocouples to measure the tem-peratures of the entering fuel and air;

• One or more thermocouple(s) to mea-sure the temperature(s) at a key loca-tion or locations in the combustor;

• A microphone or other acoustic-pres-sure transducer;

• Analog-to-digital converters for sam-pling the outputs of the aforemen-tioned sensors;

• A computer running special softwarefor analyzing the digitized sensor out-puts and responding as needed;

• Digital-to-analog converters to generateactuator-control signals for automatedrapid responses; and

• Output connections to displays thatwould be read by human operators.

Through continuous monitoring ofthe temperatures, pressures, and flowrates, the instrumentation systemwould provide information that would

enable a pilot, power-plant operator, orother responsible person to set the flowrates of fuel and air for safe operation.Upon detecting a sudden large in-crease in acoustic pressure, the systemwould act, much faster than the humanoperator could, to make a temporaryadjustment in the fuel/air ratio to pre-vent flameout. For example, if thecombustor was operating near the leanstability boundary, the system could re-spond by actuating a solenoid valve orfuel injector to increase the flow offuel. At the same time, the systemwould generate a display advising thehuman operator of this action and sug-gesting an adjustment to restore steadysafe operation.

This work was done by Richard Lee Pusterof Langley Research Center. Further in-formation is contained in a TSP (see page 1).LAR-15487

Physical Sciences

Predicting and Preventing Incipient Flameout in CombustorsIncreases in acoustic signals could trigger rapid adjustments to prevent flameouts.Langley Research Center, Hampton, Virginia

Spectrum in Normal Operation

Spe

ctra

l Pow

er D

ensi

tyS

pect

ral P

ower

Den

sity

Spectrum Near Flameout

Frequency

Frequency

Strong Peaks associated with instability appear in the acoustic spectrum of a combustor as it approachesflameout. The frequencies and amplitudes of the peaks depend on the combustor geometry and on temper-ature.


MEMS-Based Piezoelectric/Electrostatic Inchworm ActuatorNanometer steps could be concatenated into overall travel of hundreds of microns.NASA’s Jet Propulsion Laboratory, Pasadena, California

A proposed inchworm actuator, to bedesigned and fabricated according tothe principles of microelectromechani-cal systems (MEMS), would effect linearmotion characterized by steps as smallas nanometers and an overall range oftravel of hundreds of microns. Potentialapplications for actuators like this oneinclude precise positioning of opticalcomponents and active suppression ofnoise and vibration in scientific instru-ments, conveyance of wafers in thesemiconductor industry, precise posi-tioning for machine tools, and posi-tioning and actuation of microsurgicalinstruments.

The inchworm motion would be gen-erated by a combination of piezoelectricdriving and electrostatic clamping. Theactuator (see figure), would include apair of holders (used for electrostaticclamping), a slider (the part that wouldengage in the desired linear motion), adriver, a piezoelectric stack under the dri-ver, and a pair of polymer beams centrallyclamped to the flexure beam via a T bar.The holders would be held stationary.One end of the piezoelectric stack wouldbe held stationary; the other end wouldbe connected to the bottom of the driver,which would be free to move up anddown. All of these components except

the piezoelectric stack andthe polymer beams wouldbe micromachined from a500-µm-thick silicon waferby deep reactive-ion etch-ing. The inchworm motionwould be perpendicular tothe broad faces of thewafer (perpendicular tothe plane of the figure).

The combination ofthe polymer beams andthe centrally clamped flex-ure beam would spring-bias the slider into a posi-tion such that, in theabsence of electrostaticclamping, the gap be-tween the slider on theone hand and both the

driver and the holder on the other handwould be no more than a few microns.This arrangement would make it possi-ble to electrostatically pull the slider intocontact with either the holders or thedriver at a clamping force of the order of1 N by applying a reasonably small volt-age (of the order of 100 V).

The actuation sequence would be thefollowing:1. The slider would be electrostatically

clamped to the driver.2. The piezoelectric stack would push

the driver upward or downward (outof or into the page, respectively).

3. The slider would be electrostaticallyclamped to the holders.

4. The slider and the driver would be re-leased from each other.

5. The driver would be moved down-ward or upward by the piezoelectricactuator while the slider remainedclamped to the holders. This wouldcomplete the sequence for one in-crement of motion.

6. The cycle comprising steps 1 through5 would be repeated as many times asneeded to obtain the desired overallupward or downward travel. The rep-etition rate could be as high as about1 kHz.This work was done by Eui-Hyeok Yang of

Caltech for NASA’s Jet Propulsion Labo-ratory. Further information is contained in aTSP (see page 1).NPO-30672

1 mm

HolderHolderHolder DriverDriverDriver HolderHolderHolder

Flexure BeamT-Bar Polymer Beam

Piezoelectric Stack, Hidden Underneath

SliderSliderSlider

This Piezoelectric/Electrostatic Actuator would produce motion ofthe slider into or out of the page in small increments.

Metallized Capillaries as Probes for Raman SpectroscopyThese would offer several advantages over fiber-optic probes.NASA’s Jet Propulsion Laboratory, Pasadena, California

A class of miniature probes has beenproposed to supplant the fiber-opticprobes used heretofore in some Ramanand fluorescence spectroscopic systems. Aprobe according to the proposal would in-clude a capillary tube coated with metalon its inside to make it reflective. A mi-crolens would be hermetically sealed ontoone end of the tube. A spectroscopicprobe head would contain a single suchprobe, which would both deliver laserlight to a sample and collect Raman or flu-orescent light emitted by the sample.

The capabilities of most prior Ramanand fluorescence fiber-optic probes arelimited because of spurious emission ofRaman and fluorescent light by the coresof the optical fibers themselves. To pre-vent the spurious emissions from over-whelming the desired Raman and/or flu-orescence signals, it is necessary toincorporate spectral filters. In a given ap-plication, such a filter undesirably limitsthe probe to a single laser wavelength. Fil-tration is usually performed by means offree-space optics in the probe head.

These optics are vulnerable to misalign-ment. Alternatively, thin-film filters canbe deposited directly on optical fibers orin proximity to the fibers, but the perfor-mances of such filters are inferior tothose of free-space optical filters.

A typical probe according to the pro-posal would have an outside diameter of<1 mm. Relative to a typical prior Ramanprobe, this probe would be muchsmaller and lighter and could be manu-factured at lower cost. The interior ofthe probe could be evacuated or filled


with a gas (e.g., argon) that does notemit any Raman or fluorescent light inthe spectral region of interest. Hence,there would be no need for filters, ancil-lary filter optics, and associated mount-ing hardware vulnerable to misalign-ment, and so the probe would be betterable to withstand such adverse environ-mental conditions as higher pressures,temperatures, and mechanical shocks.The elimination of filters would make itpossible to operate the probe at morethan one wavelength. A probe accordingto the proposal could perform well inthe deep ultraviolet spectral region,which is important for spectral detectionof organisms. Unlike fiber-optic probescontaining filters, these probes could beused to measure Raman wave-numbershifts smaller than 50 cm–1.

In the development of these probes, itshould be possible to take advantage ofthe knowledge base already accumu-lated in the use of metallized capillary

tubes to deliver laser beams for surgeryand cutting metal. Provided that the re-flectivity of the interior metal coat of acapillary is sufficiently high, the numeri-cal aperture of the capillary sufficientlylow, and the diameter of the capillarysufficiently large, it should be possible toachieve efficient transmission of lightalong the capillary.

The small size, low cost, and multi-wavelength capability of these probesmake them attractive for many Ramanand fluorescence applications. Matricesof samples generated by combinatorialchemistry could be analyzed by an arrayof these probes interfaced to a singlefiber-optic probe head or directly to asingle Raman spectrograph. Access portsto chemical reactors could be keptsmall, reducing the potential hazards ofa seal failure. Probes inserted into largechemical reactors could be muchsmaller in diameter due to the greatly re-laxed stiffness requirements compared

to traditional Raman probes. Also, be-cause of their smallness, they would becapable of withstanding high pressuresand would be well suited for use in un-dersea exploration.

This work was done by Michael Pelletier ofCaltech for NASA’s Jet Propulsion Labo-ratory. Further information is contained in aTSP (see page 1).


Intellectual Assets OfficeJPLMail Stop 202-2334800 Oak Grove DrivePasadena, CA 91109-8099(818) 354-2240E-mail: [email protected] to NPO-30711, volume and number



Information Sciences

Aerodynamic Design Using Neural NetworksThe amount of computation needed to optimize a design is reduced.Ames Research Center, Moffett Field, California

The design of aerodynamic compo-nents of aircraft, such as wings or en-gines, involves a process of obtaining themost optimal component shape that candeliver the desired level of componentperformance, subject to various con-straints, e.g., total weight or cost, that thecomponent must satisfy. Aerodynamic de-sign can thus be formulated as an opti-mization problem that involves the mini-mization of an objective function subjectto constraints.

A new aerodynamic design optimiza-tion procedure based on neural networks

and response surface methodology (RSM)incorporates the advantages of both tradi-tional RSM and neural networks. The pro-cedure uses a strategy, denoted parameter-based partitioning of the design space, toconstruct a sequence of response surfacesbased on both neural networks and poly-nomial fits to traverse the design space insearch of the optimal solution.

Some desirable characteristics of thenew design optimization procedure in-clude the ability to handle a variety of de-sign objectives, easily impose constraints,and incorporate design guidelines and

rules of thumb. It provides an infrastruc-ture for variable fidelity analysis and re-duces the cost of computation by usingless-expensive, lower fidelity simulationsin the early stages of the design evolu-tion. The initial or starting design can befar from optimal. The procedure is easyand economical to use in large-dimen-sional design space and can be used toperform design tradeoff studies rapidly.Designs involving multiple disciplinescan also be optimized.

Some practical applications of the de-sign procedure that have demonstrated

Methodologies have been developedfor (1) configuring mesoscale numericalweather-prediction models for execu-tion on high-performance computerworkstations to make short-rangeweather forecasts for the vicinity of theKennedy Space Center (KSC) and theCape Canaveral Air Force Station(CCAFS) and (2) evaluating the perfor-mances of the models as configured.These methodologies have been imple-mented as part of a continuing effort toimprove weather forecasting in supportof operations of the U.S. space program.The models, methodologies, and resultsof the evaluations also have potentialvalue for commercial users who couldbenefit from tailoring their operationsand/or marketing strategies based onaccurate predictions of local weather.

More specifically, the purpose of de-veloping the methodologies for config-uring the models to run on computers atKSC and CCAFS is to provide accurateforecasts of winds, temperature, andsuch specific thunderstorm-related phe-nomena as lightning and precipitation.The purpose of developing the evalua-tion methodologies is to maximize theutility of the models by providing users

with assessments of the capabilities andlimitations of the models.

The models used in this effort thus farinclude the Mesoscale AtmosphericSimulation System (MASS), the Re-gional Atmospheric Modeling System(RAMS), and the National Centers forEnvironmental Prediction Eta Model(“Eta” for short). The configuration ofthe MASS and RAMS is designed to runthe models at very high spatial resolu-tion and incorporate local data to re-solve fine-scale weather features. Modelpreprocessors were modified to incorpo-rate surface, ship, buoy, and rawinsondedata as well as data from local wind tow-ers, wind profilers, and conventional orDoppler radars.

The overall evaluation of the MASS,Eta, and RAMS was designed to assess theutility of these mesoscale models for sat-isfying the weather-forecasting needs ofthe U.S. space program. The evaluationmethodology includes objective and sub-jective verification methodologies. Ob-jective (e.g., statistical) verification ofpoint forecasts is a stringent measure ofmodel performance, but when usedalone, it is not usually sufficient for quan-tifying the value of the overall contribu-

tion of the model to the weather-fore-casting process. This is especially true formesoscale models with enhanced spatialand temporal resolution that may be ca-pable of predicting meteorologicallyconsistent, though not necessarily accu-rate, fine-scale weather phenomena.Therefore, subjective (phenomenologi-cal) evaluation, focusing on selected casestudies and specific weather features,such as sea breezes and precipitation, hasbeen performed to help quantify theadded value that cannot be inferredsolely from objective evaluation.

This work was done by John T. Manobianco,Gregory E. Taylor, Jonathan L. Case, Allan V.Dianic, and Mark W. Wheeler of ENSCO, Inc.,John W. Zack of MESO, Inc., and Paul A.Nutter formerly of ENSCO, Inc., for KennedySpace Center. For further information, con-tact John Manobianco at (321) 853-8202 orvia e-mail at [email protected] orrefer to “Evaluation of the 29-km Eta Model.Part I: Objective Verification at Three SelectedStations” and “Evaluation of the 29-km EtaModel. Part II: Subjective Verification overFlorida” in Weather and Forecasting, Volume14 (February 1999), published by the Ameri-can Meteorological Society.KSC-12241

Adaptation of Mesoscale Weather Models to Local ForecastingBoth objective and subjective evaluation methodologies are needed.John F. Kennedy Space Center, Florida


some of its capabilities include the inversedesign of an optimal turbine airfoil start-ing from a generic shape and the redesignof transonic turbines to improve their un-steady aerodynamic characteristics.

In one practical application, the pro-cedure was used to reconstruct theshape of a turbine airfoil given a desiredpressure distribution and some relevantflow and geometry parameters. Theshape of the airfoil was not known be-forehand. Instead, it was evolved from asimple curved section of nearly uniformthickness. The evolved optimal airfoilclosely matched the shape of the origi-nal airfoil that was used to obtain thepressure distribution. The progressionof the design is depicted in the figure.The airfoil shape evolution is shown onthe left, while the corresponding pres-sure distributions and the target pres-sure distribution are shown on the right.The surface pressures approach the tar-get distribution as the design progressesuntil the optimal airfoil shown at thebottom has a pressure distribution thatmatches closely the target.

The technology developed and imple-mented in the neural-network-based de-sign optimization procedure offers aunique capability that can be used inother aerospace applications such as ex-ternal aerodynamics and multidiscipli-nary optimization, and has potential ap-plications beyond aerospace design.

This work was done by Man Mohan Raiand Nateri K. Madavan of Ames ResearchCenter. Further information is contained in aTSP (see page 1).

This invention is owned by NASA, and apatent application has been filed. Inquiriesconcerning nonexclusive or exclusive licensefor its commercial development should be ad-dressed to the Patent Counsel, Ames ResearchCenter, (650) 604-5104. Refer to ARC-14281.

The Design Procedure can evolve a generic shape into an optimized airfoil that matches a target pres-sure distribution. Position coordinates (x and y) are normalized to the chord length (c); local pressure(p) is normalized to the turbine inlet total pressure (pt, ∞).

Target Design

0.0 0.2 0.4 0.6 0.8 1.00.00.0

0.2

0.4

0.6

0.8

1.0

1.0

1.0

1.0

1.0

1.0

1.0

1.0

Initial AirfoilDesign

MidwayThrough EulerOptimization

Process

Beginning ofNavier-StokesOptimization

Process

FinalOptimized

Design

0.2 0.4

x/c

y/c

p/p t

,∞

x/c

AIRFOIL SHAPES PRESSURE DISTRIBUTIONS

0.6 0.8 1.0

Evolved Design

A proposed method of processingthe outputs of multiple gyroscopes toincrease the accuracy of rate (that is,angular-velocity) readings has been de-veloped theoretically and demon-strated by computer simulation. Al-though the method is applicable, inprinciple, to any gyroscopes, it is in-tended especially for application to gy-

roscopes that are parts of microelectro-mechanical systems (MEMS). Themethod is based on the concept thatthe collective performance of multiple,relatively inexpensive, nominally iden-tical devices can be better than that ofone of the devices considered by itself.The method would make it possible tosynthesize the readings of a single,

more accurate gyroscope (a “virtual gy-roscope”) from the outputs of a largenumber of microscopic gyroscopes fab-ricated together on a single MEMSchip. The big advantage would be thatthe combination of the MEMS gyro-scope array and the processing cir-cuitry needed to implement themethod would be smaller, lighter in

Combining Multiple Gyroscope Outputs for Increased AccuracyA lightweight, low-power, compact MEMS gyroscope array could perform comparably to alarger more-conventional gyroscope.NASA’s Jet Propulsion Laboratory, Pasadena, California


Gyroscope 1

Rate (Angular Velocity)Random-Walk Noise Sources

Angle Random-WalkNoise Sources

Noise-Mixing Matrix

MultipleGyroscope

Outputs

Minimum-VarianceEstimate of Rate

True Rate

Truth Model

OptimalFilter

Gyroscope 2

Gyroscope 3

Rate Error–

Noise-MixingMatrix

Readings From Multiple Gyroscopes would be combined by use of an optimal (minimum-variance) filter, thereby synthesizing the readings of whatamounts to a more accurate virtual gyroscope.

weight, and less power-hungry, relativeto a conventional gyroscope of equalaccuracy.

The method (see figure) is one ofcombining and filtering the digitizedoutputs of multiple gyroscopes to ob-tain minimum-variance estimates ofrate. In the combining-and-filtering op-erations, measurement data from thegyroscopes would be weighted andsmoothed with respect to each other ac-cording to the gain matrix of a mini-mum-variance filter. According toKalman-filter theory, the gain matrix ofthe minimum-variance filter is uniquelyspecified by the filter covariance, whichpropagates according to a matrix Ric-cati equation. The present method in-corporates an exact analytical solutionof this equation.

The analytical solution reveals a wealthof theoretical properties and enables theconsideration of several practical imple-mentations. Among the most notable the-oretical properties are the following:

• Even though the terms of the Riccatiequation grow in an unboundedfashion, the Kalman gain can beshown to approach a steady-statematrix. This result is fortuitous be-cause it simplifies implementationand can serve as a basis of practicalschemes for realizing the optimal fil-ter by use of a constant- gain matrix.

• The analytical solution enables thedevelopment of a complete statisti-cal theory that characterizes thedrift of the virtual gyroscope andprovides theoretical limits of im-provement obtainable by use of anensemble of correlated sensors as asingle virtual sensor.

• The minimum-variance gain matrixhas been analyzed in detail. Thestructure of the gain matrix indicatesthe presence of a marginally unstablepole, which would make implementa-tion impossible if it were not properlyunderstood and compensated. In ad-dition, a simple algebraic method forcomputing the optimal gain matrixhas been developed, making it possi-ble to avoid the Riccati equation com-pletely.

• The notion of statistical common-mode rejection (CMR) has beencharacterized mathematically. For sta-tistically uncorrelated gyroscopes, it isshown that the component drift vari-ances add like parallel resistors (e.g.,in units of rad2/sec3). For identicaldevices, this means that the combineddrift is reduced by a factor of compared to the individual gyroscopedrift, where N is the number of gyro-scopes being combined.

• Potential improvement in drift ismuch more impressive when the de-

vices are correlated. For correlatedgyroscopes, an exact mathematicalexpression is developed for the com-bined drift. The expression indi-cates that the internal noise sourcescount to the extent that they infectmultiple devices coherently (e.g.,with common sign), and can be re-moved to the extent that they infectmultiple devices incoherently (e.g.,with randomized signs). Noise elimi-nated by correlations between de-vices can potentially reduce the vir-tual gyroscope drift far beyond the

factor attainable using uncor-related devices.

This work was done by David S. Bayard ofCaltech for NASA’s Jet Propulsion Labo-ratory. Further information is contained in aTSP (see page 1).




1 N

1 N


Improved Collision-Detection Method for Robotic ManipulatorCollisions are detected in a computationally efficient manner.NASA’s Jet Propulsion Laboratory, Pasadena, California

An improved method has been de-vised for the computational prediction ofa collision between (1) a robotic manip-ulator and (2) another part of the robotor an external object in the vicinity of therobot. The method is intended to beused to test commanded manipulatortrajectories in advance so that executionof the commands can be stopped beforedamage is done. The method involvesutilization of both (1) mathematicalmodels of the robot and its environmentconstructed manually prior to operationand (2) similar models constructed auto-matically from sensory data acquiredduring operation. The representation ofobjects in this method is simpler andmore efficient (with respect to both com-putation time and computer memory),relative to the representations used inmost prior methods.

The present method was developed es-pecially for use on a robotic land vehicle(rover) equipped with a manipulator armand a vision system that includes stereo-scopic electronic cameras. In this method,objects are represented and collisions de-tected by use of a previously developedtechnique known in the art as the methodof oriented bounding boxes (OBBs). Asthe name of this technique indicates, an ob-ject is represented approximately, for com-putational purposes, by a box that enclosesits outer boundary. Because many parts of arobotic manipulator are cylindrical, theOBB method has been extended in this

method to enable the approximate repre-sentation of cylindrical parts by use of oc-tagonal or other multiple-OBB assembliesdenoted oriented bounding prisms(OBPs), as in the example of Figure 1. Un-like prior methods, the OBB/OBP methoddoes not require any divisions or transcen-dental functions; this feature leads togreater robustness and numerical accuracy.The OBB/OBP method was selected for in-corporation into the present method be-cause it offers the best compromise be-tween accuracy on the one hand andcomputational efficiency (and thus compu-tational speed) on the other hand.

OBBs are also used to represent the ter-rain and any objects on the terrain sensedby the stereoscopic vision system. A con-ceptual multiresolution map pyramid ofthe manipulator work space is computedfrom the stereoscopic sensory data and isthen used in a coarse-to-fine sequence todetect collisions between the manipulatorand terrain. As described next, tests forcollisions are performed in a hierarchicalsequence to minimize the amount of com-putation needed to detect collisions.

Starting with the second-highest level ofthe pyramid, each level is characterized by

twice the horizontal spatial resolution ofthe level above it (see Figure 2). For ex-ample, at the highest level of the pyramid(coarsest resolution) there is a single ter-rain OBB that encloses all of the senseddata points. The model for each manipu-lator link is one low-resolution OBP. If nocollisions between any of the OBPs andthe coarsest-resolution terrain OBB aredetected, then there is no need for fur-ther computation to detect collisions withterrain. On the other hand, if collisionsare detected at the coarsest resolution,then tests for collisions are performed oneach of the terrain OBBs at the secondcoarsest resolution. This process contin-ues to successively finer levels of resolu-tion until the finest resolution is reachedor no more collisions are detected. Simi-lar tests for collisions are performed witha similarly hierarchical model of the non-manipulator parts of the robot (body,cameras, sensors, suspension, andwheels).

This work was done by Chris Leger of Cal-tech for NASA’s Jet Propulsion Labora-tory. Further information is contained in aTSP (see page 1).NPO-30356

Octagonal Boundary of OBP

Boundaries of Constituent OBBs

Figure 1. A Rodlike Robot Arm of Circular CrossSection, viewed here along its axis, can be rep-resented by an octagonal OBP assembled fromfour smaller OBBs.

Figure 2. A Hierarchy of OBBs of successively finer resolution is used to represent terrain elevation asa function of horizontal coordinates.

National Aeronautoics andSpace Adninistration

Technology Focus Computers/Electronics · Technology Focus: Test & Measurement Nulling Infrared...

Documents

Transcript of Technology Focus Computers/Electronics · Technology Focus: Test & Measurement Nulling Infrared...