Document
AIAA-92-4111-CP
FLIGHT EXPERIENCE WITH LIGHTWEIGHT, LOW-POWER
MINIATURIZED INSTRUMENTATION SYSTEMS
Philip J. Hamory" James E. Murray"" NASA Dryden Flight Research Facility Edwards, California Abstract POPU pushover-pullup RAM random access memory Engineers at the NASA Dryden Flight Research Fa- Symbols cility (NASA-Dryden) have conducted two flight re- search programs with lightweight, low-power miniatur- an normal acceleration, g ized instrumentation systems built around commercial ax axial acceleration, g data loggers. One program quantified the performance of a radio-controlled model airplane. The other pro- vehicle drag coefficient Cn gram was a laminar boundary-layer transition experi- vehicle lift coefficient CL ment on a manned sailplane.
lift-to-drag ratio L/D The purpose of this paper is to report NASA-Dryden ij dynamic pressure, psf personnel's flight experience with the miniaturized in- T total thrust, lb strumentation systems used on these two programs.
The paper will describe the data loggers, the sensors, true airspeed, ft/ sec 'Vtrue and the hardware and software developed to complete angle of attack, deg a the systems. The paper also describes how the sys- angle-of-attack breakpoint, deg no tems were used and covers the challenges encountered to make them work. Examples of raw data and derived lie symmetric elevon deflection, deg results will be shown as well. Finally, future plans for left-elevon deflection, deg liel these systems will be discussed.
right-elevon deflection, deg lier For some flight research applications where minia- l:1p, left-engine static pressure rise, psi turized instrumentation is a requirement, the authors conclude that commercially available data loggers and right-engine static pressure rise, psi f:1Pr sensors are viable alternatives. In fact, the data loggers Introduction and sensors make it possible to gather research-quality data in a timely and cost-effective manner.
Engineers at NASA-Dryden fly radio-controlled model aircraft for preliminary validation of risky design Nomenclature concepts. For programs with limited budgets, the engi- neers also operate a manned, self-launching sailplane as Acronyms a low-speed flight research platform. Though the two A/D analog to digital research efforts are quite different, their instrumenta- CMOS complementary metal oxide semiconductor tion system requirements are similar. Each system re- quires up to 24 channels of data acquisition, onboard CPT control position transducer recording, and operation using batteries, in addition to *Electrical Engineer.
small size, weight, and price.
•• Aerospace Engineer.
Copyright ©1992 by the American Institute of Aeronau- The instrumentation systems currently in use tics and Astronautics, Inc. No copyright is asserted in the at NASA-Dryden are overly complex for these United States under Title 17, U.S. Code. The U.S. Govern- applications. A class of instruments, commonly called ment has a royalty-free license to exercise all rights under data loggers, was investigated as an alternative. Data the copyright claimed herein for Governmental purposes. All other rights are reserved by the copyright owner.
loggers are commercially available data acquisition units with onboard storage. From the Tattletale® analog to digital (A/D) converter that can operate at line of data loggers, the Model 4 was selected for up to approximately 600 samples/sec. It has an RS- the radio-controlled model aircraft. Later, when the 232 serial communication port and can be programmed Model 7 became available, it was selected for the in TTBASIC, which is the BASIC computer language manned sailplane. with extra commands for data logging and host inter- face. Software development can be conducted on an This report describes the flight experience gained IBM PC® or compatible machine.
with miniaturized instrumentation systems built around the Model 4 and Model 7 data loggers. Fol- The Model 4 measures 3.725 by 2.25 by 0.8 in., lowing a brief outline of the data loggers' features, the weighs 2.2 oz, and consumes a maximum of 100 mW. It report will cover the transducers the data loggers were has an onboard 3-V lithium battery to sustain volatile interfaced to and how they were used. Additional dis- memory. The internal bus is brought out to 32 pins.
cussions include the software and hardware developed These pins are arranged so that one can build a sys- to complete the systems, and the challenges faced in tem by stacking memory expansion boards and user- getting the systems to work properly. Examples of raw designed application boards on top of the Model 4.
and derived data will be shown. Finally, future plans Model 7 Data Logger for use of these systems will be outlined.
The Model 7 data logger contains a 68332 mi- System Description of Tattletale Data croprocessor, a low-power version of the Motorola 68020 microprocessor (Motorola, Inc., Phoenix, Ari- Loggers zona) and has 2 Mbytes of onboard memory. For Model 4 Data Logger measurement of analog signals, the Model 7 has a 4-channel, 12-bit A/D converter that can operate at up The Model 4 data logger contains a 6301 micropro- to 100,000 samples/sec. It has an RS-232 serial com- cessor (Hitachi America Ltd., San Jose, California), a munication port and a high-speed parallel interface.
descendant of the Motorola 6800 microprocessor (Mo- The Model 7 is programmed in Think C (Symantec torola, Inc., Phoenix, Arizona). The Model 4 has Corp., Cupertino, California). Think C is an Amer- 32 kbytes of onboard memory. For measurement of ican National Standards Institute (ANSI)-C language analog signals, the Model 4 has an 11-channel, 10-bit Table 1. Comparison of Tattletale data loggers.
Specification Model 4 Model 7 Microprocessor 6301 68332 Onboard memory 32 kbytes 2 Mbytes A/D converter resolution, bits 10 12 Maximum sample rate, 600 100,000 (approximate) samples/sec Host computer PC compatible Macintosh (PC interface planned) Language TTBASIC C RS-232 port Yes Yes Parallel port No Yes Battery-backed RAM Yes No Hard drive No Yes (up to 80 Mbytes) Real- time clock No Yes Size, in. 3. 725 by 2.25 by 0.8 4.0 by 2. 75 by 0.44 Weight, oz 2.2 2.5 Weight including hard drive, oz 8 Maximum power consumption, mW 100 500 Typical power with hard drive, W 3 Approximate cost with startup materials $500 $3300 ® Tattletale is a registered trademark of Onset Computer ® IBM PC is a registered trademark of the IBM Corp., Corp.,.North Falmouth, Massachusetts. Armonk, New York.
., development environment. Software development can acceleration (ax and an, respectively), angle of attack be conducted on a Macintosh® computer. (a), dynamic pressure (ij), and thrust (T).
The Model 7 measures 4.0 by 2.75 by 0.44 in., weighs Accurate measurement of the thrust produced by 2.5 oz, and consumes a maximum of 500 mW. It is also the ducted-fan engines was central to obtaining accu- rate drag measurements. An engine and fan assembly designed to interface to a hard drive, the same kind of was calibrated in a wind-tunnel test program. Then a drive used in many laptop computers. The hard drive model of engine thrust as a linear function of dynamic adds 1 in. to the vertical measurement and 5.5 oz to pressure and static pressure rise across the fan assem- the total weight. As with the Model 4, user-designed application boards can be stacked upon the Model 7 bly was obtained. Measurement of in-flight thrust also and interfaced to a 120-pin internal bus. required measurement of the static pressure rise across the left- and right-fan assembly, (t::.Pt and t::.Pr> respec- Flight Experience With a tively). Measurement of the trim curve also required Radio-Controlled Model Aircraft measurements of the elevon control surface deflections (Dee and DeJ• Test Objective Because of size and weight constraints, no lateral- or The Model 4 has been used on several model aircraft directional-axis variables were measured. All test ma- flight programs where research-quality data were re- neuvers were flown while minimizing cross-axis motion; quired for quantitative analysis. An objective of one of the analysis assumed purely longitudinal motion.
these flight programs was to measure the performance of the test aircraft, which was powered by two ducted- The selection of sensors for the test program was fan engines. Performance parameters of interest were driven primarily by weight and power considerations.
the lift-curve slope, the lift-to-drag ratio, and the trim Table 2 presents sources and specifications for the sen- curve.
sors used in the test program.
Maneuver Design Instrumentation System A flight test maneuver commonly used to quantify The Model 4 was the core of the instrumentation sys- aircraft performance is a quasi-static pushover-pullup tem used on the model aircraft. A 128 kbyte memory (POPU). Properly executed, a single POPU maneu- expansion board was added to increase data storage ver can simultaneously characterize the lift curve, the capability to approximately 150 kbytes. Two applica- lift-to-drag ratio, and the trim curve over a large angle- tion boards were also needed to complete the hardware.
of-attack range. For its simplicity, the POPU maneu- These were designed at NASA-Dryden and consisted ver was initially selected as the primary maneuver for of an analog multiplexer board and a power supply performance measurement.
Table 2. Sensor specifications.
Variable Model number Range Resolution Accuracy a, deg NASA-Dryden noseboom -0.5 to 40 0.04 0.25 ij, psf SenSym 142SC01D* 0 to 55 0.058 0.144 be deg; Der, deg NASA-Dryden CPT -40 to 20 0.06 0.20 SenSym 142SC01D* 0.0 to 6 0.0006 0.001 !::,_Pi' psi; !::,_Pr' psi IC Sensors 3110-002** -1 to 1 0.002 0.015 ax, g an,g IC Sensors 3110-005** -0.5 to 2.5 0.003 0.02 *SenSym Inc., Sunnyvale, California.
**IC Sensors, Milpitas, California.
Measurement Requirements and Sensor board. The circuitry for each channel on the mul- Selection tiplexer board consisted of a differential amplifier, a third-order Butterworth low-pass filter, and a comple- Minimum measurement requirements for the test mentary metal oxide semiconductor (CMOS) switch.
program were determined from the test requirements The schematic is shown in Fig. 1. The power sup- and maneuver selection. Measurement of the aerody- ply board provided voltage regulation and excitation namic forces (lift and drag) and their nondimensional to the sensors, and contained a voltage reference for coefficients required measurement of axial and normal the A/D converter. Figure 2 shows the complete data acquisition system including the Model 4, the mem- ory expansion board, application boards, battery pack, ®Macintosh is a registered trademark of Apple Computer, and all sensors. These components weighed 1. 7 lb Inc., Cupertino, California.
and consumed 500 mW of power. Figure 3 shows the frequencies, and antenna orientations were tried with- Model 4 package installed in the nose section of the out success. Boosting uplink power from 0.5 to 5W model aircraft. provided some improvement; however, upon elimina- tion of an intermittent ground loop in the instrumen- The instrumentation system is incomplete without tation the preflight checks succeeded consistently.
software. Three software modules were required to operate and obtain data from the system. The first The high-vibration environment in the model air- module controlled the Model 4 in flight. The sec- craft provided another challenge. At high power set- ond module downloaded flight data from the Model 4. tings, the structural vibration from the reciprocating The third module converted the raw flight data into engines was in excess of 5 g. This overwhelmed the an engineering format compatible with software used lower level acceleration signals from the lift and drag on several computers at NASA-Dryden. The first forces during the test maneuver. Different mounting and third modules were written at NASA-Dryden in techniques and mounting locations for the engines and TTBASIC and FORTRAN 77, respectively. The sec- accelerometers were attempted but failed to alleviate ond module, Procomm (Data.storm Technologies, Inc., the problem. Ultimately, a "pullup" maneuver was de- Columbia, Missouri), was obtained commercially. veloped that could be flown with the engines at an idle power setting. Under these conditions the accelerome- Only the first module ran on the Model 4. It was up- ter data were suitable for analysis.
loaded from the PC compatible before flight and was left in standby mode. At the runway, a ground oper- Results of Model Aircraft Flights ator threw a switch to start data logging and released Figure 4 shows a time history of a representative the aircraft for takeoff. The Model 4 recorded eight pullup maneuver. Included are the raw measured vari- channels of data continuously at 25 samples/sec. Data ables, o:, ij, /jet, /jer, LlPt' LlPr' an, and ax, as well as the logging stopped automatically in approximately 6 min derived variables true airspeed (½rue), lift coefficient when random access memory (RAM) was full.
(CL), drag coefficient (CD), T, and lift-to-drag ratio The other two modules ran on the PC compatible. (L/ D). Light turbulence is evident in o:, an, and ax.
By using an 8086 class PC compatible, a flight's data Figures 5, 6, and 7 show results derived from the set (150 kbytes) could be downloaded in approximately time history data shown in Fig. 4. The turbulence- 6 min and converted to engineering format in approxi- induced and other noises were removed from a subset mately 30 min. Though the total time is comparatively of the raw data using a digital low-pass filter, and these long, it was adequate for the project needs. The total data were used in the following results.
time can be shortened by using a more powerful PC compatible. Figure 5 shows the flight-determined lift curve for the aircraft. The small symbols represent the CL and Implementation Challenges angle-of-attack measurements at each sample point, In anticipation of potential interference between and the solid line is a piecewise linear fit to the data the avionics and instrumentation systems, the power points. There is a change in slope at o: = o: where the switch for the instrumentation system was attached to aerodynamics of the aircraft change character. Both a servoactuator of the avionics system. This gave the above and below o: = a the slope is consistent.
pilot the ability to turn the instrumentation system Figure 6 shows the flight-detemined lift-to-drag ra- on or off in flight. Likewise, the model aircraft re- tio for the aircraft. The small symbols represent the ceiver was programmed to turn off the instrumentation L/ D measurements at each sample point, and the solid system automatically upon loss of the transmitter sig- line is a hand-faired estimate through the data points.
nal. The logger program and any data gathered would The dotted lines above and below the solid lines are not be lost upon shutdown because of the 3-V backup estimates of the error bounds for the data based on in- battery.
strumentation error estimates, data repeatability, ma- There was never any interference between the sys- neuver quality, and signal-to-noise ratio.
tems in flight. However, getting the systems to work Figure 7 shows the flight-determined trim curve for together on the ground was a major challenge initially.
the aircraft. The small symbols represent the 8e and From successful noninstrumented flights it was clear angle-of-attack measurements at each sample point, that the avionics system worked. From practice and the solid line is a piecewise linear fit to the data flights using a different model airplane it was also points. As with the lift curve shown in Fig. 5, there clear that the instrumentation worked. However, for is definite change in the trim curve at approximately an unknown reason, the preflight checks sometimes o: = o:o, where there is a measurable change in the failed with the instrumentation system in place and aircraft aerodynamics.
turned on. Different transmitters, receivers, uplink Flight Experience With a Manned cause it existed on the anemometers. A patch panel be- tween the anemometers and the multiplexer board pro- Sailplane vided easy selection of hot films between flights. The Test Objective power supply board provided voltage regulation and a voltage reference for the system as well as a voltage For programs with limited budgets, a self-launching offset for the differential amplifiers on the multiplexer sailplane (shown in Fig. 8) serves as a low-speed flight board.
research platform. The objective of one flight program was to determine the maximum size of excrescence tol- Figure 11 shows the Model 7, the analog multiplexer erable on a laminar-flow wing before boundary-layer board, and power supply board. Also shown is one transition occurs. Excrescences are rough spots on a of the anemometers and the 72-Whr nickel-cadmium surface.
battery pack that powered the entire system. The size of the battery pack was driven by three factors: (1) Experiment Design the power requirements of the anemometers, (2) the A set of calibration flights determined the natural flight time, and (3) the need to provide reserve capacity boundary-layer transition location on the wing. In sub- to power instrumentation added in the future. The sequent flights, excrescenses of varying thickness were anemometer chassis consumed up to 23 W when fully applied to the wing, and changes in the boundary-layer stocked and flights generally lasted 1 hr. The Model 7 transition location were observed. The excresences package alone consumed only 0.9 W. Figure 12 shows were simulated by using vinyl sheets 0.032 in. or less in the battery pack, Model 7 chassis, and anemometer thickness. The sheets were wrapped around the leading chassis installed in the cockpit.
edge of the wing leaving an aft-facing step as shown in Three software modules were required to operate and Fig. 9. Maneuvers were flown 5000 to 10,000 ft above obtain data from the system. The first module con- sea level, and straight and level flight was used to ob- trolled the Model 7 in flight. The second module oper- tain stabilized data.
ated the parallel interface and downloaded data from Measurement Requirements the Model 7. The third module converted the raw flight data into a format readable by plotting software avail- To pinpoint transition location, an array of hot films able for the Macintosh computer. The three modules (shown in Fig. 10) was laid on the wing such that part were written at NASA-Dryden in C.
of the array would see laminar flow and part would see turbulent flow. Hot-film anemometers developed Only the first module ran on the Model 7. It was at NASA-Dryden provided signal conditioning for the uploaded from the Macintosh computer before flight hot films. Figure 10 shows the hot-film array on the and it remained in standby mode. When triggered by wing. It was a requirement to record eight of the hot- the pilot, it turned on a lamp in the cockpit, gathered film signals at 1500 samples/sec.
eight channels of hot-film data at a rate of 1500 sam- ples/sec for 4 sec, turned off the lamp, and waited for Airdata parameters such as outside air temperature, the next trigger. Data for up to 10 maneuvers could airspeed, and altitude were also measured. For expedi- be captured in this manner during 1 flight.
ency, these measurements were not recorded electron- ically. Rather, the pilot read the measurements from The other two modules ran on the Macintosh com- the cockpit instrument panel and noted them on the puter. A flight's data set (1 Mbyte) could be down- flight card. The pilot also noted the time for correla- loaded in approximately 3 min, and conversion of the tion to the electronic data.
data from binary format only required a few minutes.
Hence, flight data were available quickly after landing.
Instrumentation System Implementation Challenges The Model 7 was the core of this instrumentation system. The Model 7 was selected for its high sampling The major challenge involved triggering the system.
rate capability and greater data storage capacity.
Initially, the triggering circuitry consisted of a digital input pin on the Model 7, a 100 kD pullup resistor, Two application boards were stacked on the Model 7 and a 6-ft wire to a switch on the control stick. This to complete the system. One was an eight-channel ana- circuitry was susceptible to false triggering upon engine log multiplexer board, and the other was a power sup- startup.
ply board. Both were designed at NASA-Dryden. The circuitry for each channel on the multiplexer board con- To solve the triggering problem, a 1-μF capacitor sisted of a differential amplifier and a CMOS switch.
was attached to the digital input pin, and the pullup Overvoltage protection for the A/D converter inputs, a resistor was reduced to 1 kD. However, for added ro- lamp driver, and triggering circuitry were included on bustness, the software was changed to look for a trigger this board as well. No anti-aliasing was included be- pulse lasting at least 0.25 sec. Previously it had only The high-speed parallel interface between the looked for a momentary trigger. Furthermore, a lamp Model 7 and the Macintosh computer will be repack- in the cockpit, previously illuminated only while data aged as well. The interface currently resides on a were being gathered, was programmed to blink once tiny circuit board connected perpendicularly to the every 10 sec after the memory was filled. This pro- Model 7. It will be redesigned in-house to attach in vided some visible indication of the system's state.
parallel and will have the extra 1/16 in. on each side Results of Sailplane Flights as mentioned previously. The header strip will be replaced by a D-subminiature connector. These en- Figure 13 shows a time history of eight hot films for one maneuver conducted at 70 knots during a smooth- hancements will make the interface more practical to wing flight; that is, a flight without excrescences. The work with.
signals are offset by chord location measured in per- System Applications cent chord. The lower amplitude signals, such as those from 22.6 to 59.2 percent chord, indicate laminar flow. Eventually this may develop into a "shoebox" instru- The large amplitude signal at 62.8 percent indicates mentation system; that is, a self-contained system, ap- turbulent flow. Thus the transition location for this proximately the size of a shoebox. This system would be capable of measuring the basic parameters needed maneuver was between 59.2 and 62.8 percent chord.
to determine an aircraft's performance: accelerations, Figure 14 shows results derived from flights with aft- rates, positions, pressures, and temperatures. In ad- facing steps located at 42.6-percent chord. The tran- dition, this would be a system that could be rapidly sition location resulting from step heights of 0.012, installed on an aircraft to address last-minute needs or 0.015, and 0.020 in. are compared with the transi- to support projects with very short lead times.
tion location of the smooth wing. Notice that with all step heights the transition location moves for- Concluding Remarks ward. Whereas the location resulting from step heights 0.012 in. and 0.020 in. appear to be independent The flight programs described acquired research- of Reynolds number, the location resulting from the quality data in a timely and cost-effective manner us- 0.015-in. step moves forward at Reynolds numbers ing instrumentation systems built around commercially greater than 1 million. available data loggers and sensors. Data loggers are well suited for applications requiring miniaturized in- Future Plans strumentation and onboard recording.
System Architecture The data loggers provide large bandwidth and on- board storage capabilities in a small-sized and low- For applications where a data logger is appropriate, power package. In the case of the sailplane program, plans call for use of the Model 7. Designs for the analog eight channels of data were acquired at 1500 sam- multiplexer and power supply boards will be finalized ples/sec for 40 sec with 12-bit (0.024-percent) resolu- and printed circuit boards will be manufactured. The tion. The package measured approximately 4 by 2. 75 boards flown on the manned sailplane were prototypes by 2.5 in., weighed approximately 8 oz, and consumed wired at NASA-Dryden upon blank prototyping boards approximately 900 mW of power.
available from the manufacturer. The printed circuit boards will have slightly larger dimensions than the References Model 7 in order to improve the system's structural integrity. Iliff, Kenneth W., "Maximum Likelihood Estima- tion of Lift and Drag from Dynamic Aircraft Ma- Currently, a stack of boards is held together by four neuvers," J. of Aircraft, vol. 14, no. 12, Dec. 1977, screws. These four screws are not located in the cor- pp. 1175-1181.
ners, so the ends of the boards can bend when experi- encing vibration. Application boards designed in-house Chiles, Harry R., The Design and Use of a will have an extra 1/16-in. on each side so that the Temperature-Compensated Hot-Film Anemometer Sys- boards can slide into grooves within a box. The grooves tem for Boundary-Layer Transition Detection on Su- will keep the ends of the boards rigid. With this mod- personic Aircraft, NASA TM-100421, 1988.
ification, a stack should be able to withstand greater vibration.
Third-order Differential Butterworth CMOS amplifier low-pass filter switch ToA/D Data return Data in Digital on/off control 920436 Fig. 1 One channel of analog multiplexer board used with the Model 4.
Dynamic pressure transducer Multiplexer board Fig. 2 Model 4 onboard data acquistion components.
EC 92 05214-4 Fig. 3 Model 4 package installed in model aircraft.
_J d~~ L I ~ ~· L_J___j..__l,____l,_______!__L__L____L._____L---1,_J.....L_____._____,
Is
~~- q, ~P' r psf psi w I-' -.J
I1
an, Oe,1,~ g
degL
Io.2s
o.,.~
ax' g
degL
~
12 12 0 2 4 6 8 10 0 2 4 6 8 10 Time, sec Time, sec 920438 Fig. 4 Time history of pullup maneuver for model aircraft.
Vtrue, ----- T,
fl/sec~ I I I ~ lb~' I I I I I I I I
r ~ r
w I-' (X)
CLt=: ,~,~
( 2 4 6 8 10 12 Time, sec CDt=I I I I I I I I !: 0 2 4 10 6 8 12 Time, sec 920439 Fig. 4 Concluded.
-- -- - -- ------ ----- - --- --- - ------ - - - ----------
LID x Flight data X Flight data -Curve fit - Piecewise linear fit - - - Estimated error bounds a,deg 920441
a=ao
a,deg 920440 Fig. 5 Flight-determined lift curve for model aircraft. Fig. 6 Flight-determined lift-to-drag ratio for model aircraft.
X Flight data - Piecewise linear fit 8e, deg a, deg 920442 Fig. 7 Flight-determined trim curve for model aircraft.
EC 91 504-1 Fig. 8 PIK 20E self-launching sailplane.
Flow direction Flow
+
direction
•
Hot-film Step height sensors ( c.__Jiii!;;; I ; 920516 Fig. 9 Cross-sectional view of the test excrescence.
Fig. 10 Hot-film sensor array on wing.
, I Power supply board Fig. 11 Signal conditioning, data logging, and power components of Model 7-based data acquisition system.
Fig. 12 Instrumentation installation in sailplane cockpit.
62.8 59.2 Amplitude _ 51 8 offset by percent chord 44.4 1""'1~-.,,.,,..,.,,.,,..,¥-~-""-,..,..,,'W'J'-""""'r"ll"w"""'""""'r,t,/w"""'"ll"ll""'\J,~""""'--v,,......J,1"'4,V.,.,.,..,.,,...l,\..J\,,..,.,,,,r,,1....,..,.w,v,l,.r'~lr"'v location 2.05 2.10 2.15 2.20 2.25 2.00 Time, sec 920446 Fig. 13 Time history of hot films on smooth wing at 70 knots.
1.0 ---0- Smooth ---ts- 0.012 in.
---o- 0.015 in.
r Hot-film locations .8
---o- 0.020 in.
+ Hot-film sensors
t 0-0:-0-0-0-0--0
Percent chord
:: _J __ ::!: _____________ _
transition location Aft step \
location _j
.2 1 2 3 Reynolds number x 10 920447 Fig. 14 Transition location for aft step configuration at 42.6--percent chord.