1

Drag and Atmospheric Neutral Density Explorer

(DANDE)

Colorado Space Grant Consortium and

CU Aerospace Engineering Sciences

COSGC Symposium

April 20, 2013Boulder, Colorado

Outline

• Over-View of the DANDE Spacecraft and science instruments.

• Discussion of time offset measurement algorithm.

• Discussion of ACC data processing algorithms.

• Emphasis will be placed on these two specifics aspects of the DANDE Mission

2

DANDE Mission

Mission StatementExplore the spatial and temporal variability of the

neutral thermosphere at altitudes of 350 - 200 km, and investigate how wind and density variability translate to

drag forces on satellites.

DRAG and ATMOSPHERIC

NEUTRAL DENSITY

EXPLORER

3

DANDE Mission

Mission StatementExplore the spatial and temporal variability of the

neutral thermosphere at altitudes of 350 – 200 325 - 1500 km, and investigate how wind and density variability translate to drag forces on satellites.

DRAG and ATMOSPHERIC

NEUTRAL DENSITY

EXPLORER

4

410 km

390 km

370 km

350 km

330 km

1999 2000 2001

reboost

decay

X5 flare7/14/2000

ISS drops 10kmin several days

E. Semones et.al. WRMISS 10 http://www.oma.be/WRMISS/workshops/tenth/pdf/ex02_semones.pdf

Operational Importance of DragThe density of the atmosphere in this region varies greatly (300% to 800%*) due to space weather and not yet understood coupled processes.

•Forbes et. Al. “Thermosphere density response to the 20-21 November 2003 solar and geomagnetic storm from CHAMP and GRACE accelerometer data”, Journal of Geophysical Research, Vol. 111, June 2006•European Space Agency Debris Tracking

5

Operational Importance of DragThe density of the atmosphere in this region varies greatly (300% to 800%*) due to space weather and not yet understood coupled processes.

•Forbes et. Al. “Thermosphere density response to the 20-21 November 2003 solar and geomagnetic storm from CHAMP and GRACE accelerometer data”, Journal of Geophysical Research, Vol. 111, June 2006•European Space Agency Debris Tracking

6

Satellite DragSatellite drag measurements suffer from errors caused by•Unknown acceleration contribution from in-track winds

•coefficient of drag accuracy

DANDE is designed to address these issues and provide acceleration, composition, and wind measurements simultaneously along with a well determined drag coefficient at ~350 km

CHAllenging Minisatellite Payload

Starshine I

7

8

How Measurements are Made

• Identifying all components of the constituents of the drag equation.• With a near-spherical shape, an a-priori physical drag coefficient may be calculated and a physical density can

be obtained from the measurements

atmosphere

ρ - densityV

A

FD

CD

VW

aMVVACF WDD 2

21

accelerometersWTS sensor

a priori knowledge

tracking

a priori knowledge

a priori knowledge comparison

solutionmeasured

a priori

solution

solved

8

Dual Instrument Approach

9

2 2.5 3 3.5 4 4.5 5 5.5 60

0.2

0.4

0.6

0.8

1

1.2 PHI =0.39138

SDEA Voltage (V)

Out

put C

urre

nt (n

A)

Vi = 15.6V

Vmax =3.7922

dV =0.27567

raw datalow-pass filterGaussian fitpeak centerhalf-peak width

Accelerometer Subsystem (ACC)As the spacecraft rotates, it modulates the drag signal to a known frequency. The unique accelerometer subsystem takes advantage of this by filtering the noise from unwanted frequencies and processing the data onboard from all accelerometers. This provides the sub-μg precision required for drag measurements.

Wind & Temperature Spectrometer (WTS)This instrument was invented by Dr. Fred Herrero at NASA Goddard and implemented for the DANDE platform by students at CU with NASA support. It is capable of measuring the wind direction and magnitude, atmospheric temperature, as well as the O:N2 ratio.

1/6 Hz drag signal

Frequency [Hz]

Measured Energy Spectrum

10

Accelerometer Instrument

Acceleration Measurements• Accelerometer Sensor Heads

– QA-2000 X 6 Units– Phased by 60°

• Analogue Filtering– 4th Order Bandpass

• 0.05-0.5Hz• 60dB Gain in the bandpass

– High Pass removes centripetal acceleration

– Low Pass removes intrinsic noise– Each ACC head filtered independently

• ADC acts as 6 independent accelerometers in one package– Simultaneous sampling of each ACC head

11

Acceleration Measurements

12

ANALOG FILTERING

A/D CONVERSION

DATA FITTING

70 ng

1x100 1x1021x10-31x10-5

1.6x10-10

4.0x10-15

Frequency [Hz]

PSD

[g2 /Hz]

1.0x10-12

1x100 1x1021x10-31x10-5

1.6x10-10

4.0x10-15

Frequency [Hz]

PSD

[g2 /Hz]

1.0x10-12spin rate

Low frequency bias

Acceleration Measurements

FD

ACC-4

ACC-6

ACC-3 ACC-

2

ACC-

5

ACC-1

ω

13

14

Wind & Temperature Spectrometer

Wind & Temperature Spectrometer

15

Acceleration Measurements

16

• Atmosphere Sample Enters from the left

• Collimator rejects natural ions, leaving only neutral particles

• Ionizing filament ionizes the neutral particle stream

• SDEA deflects the now charged atmosphere constituents a variable amount (depending on ramp voltage)

NMS: Angular Distribution Schematic

17

channel 8

channel 9

channel 10

.....

channel 11

channel 12

channel 7

channel 2

channel 1Ionizer

CollimatorSmall Deflection Energy Analyzer

NMS Data Product

18

• Energy of each particle is used to give the velocity at which that particle entered the instrument

• Spread in each peak gives a measure of the temperature of the constituent

• Displacement from the center anode gives the wind angle (as illustrated previously)

1919

Time Mapping

Time Mapping

• WHAT is TOM?– Program written in order to accurately calculate and place a time on downlinked

data

• WHY do we need TOM?– DANDE does not have GPS– DANDE system reset DANDE system time set back to 1/1/1970 at 12:00 a.m. – Need to ‘map’ DANDE system time to local time on ground in order to have

accurate and usable science data

20

Time Mapping

• How are Time Offset Measurements Taken?– Mission operators send “MeasureTimeOffset” command from

InControl client to DANDE satellite– Command asks for current time on DANDE satellite– Sends this information to ground server– Time elapsed since sent command is called “command time”– Assume DANDE time measurement occurred at half of the

“command time”– Process repeated 5 times to attain necessary accuracy

21

Time Mapping

• HOW can we use TOM?

– Fine Time Mapping: to the units of micro-seconds, needs two time offset measurements from two separate passes, epoch needs to be mapped to ground time

• Necessary for science modes

– Coarse Time Mapping: to the units of seconds, epoch needs to be mapped to ground time

– Emergency Time Mapping: to the units of minutes, no time offset measurements, only for Health and Status files

22

Time Mapping

• Verify epoch needed for processing has recorded ground start time – No entry: Run EpochStartCalculator.m– Enter epoch information into epochstart.csv

• epochStart.csv has ground starting times of all epochs

• Run TimeOffsetMeasurementChecker.m– Tells us how many T.O.M.’s accumulated for a given epoch, will recommend fine or

coarse time mapping– Also use operator’s discretion on fine, coarse, or emergency – depends on which

subsystem you are processing data for

• FINE/ COARSE– Execute ProcessedDownlinkedFiles command on InControl

23

Time Mapping

• EMERGENCY:– Used when data processing (H&S files) is crucial during a pass, and no time offset

measurements have been accumulated• Two options for mapping an epoch start time:

• Error Logs record DANDE time right before a reset– Requires a manual calculation:

(DANDE ground time from last T.O.M.)–(Ground Time right before DANDE reset/ Epoch Start Time) = DANDE present time

• Health and Status files record epoch and DANDE time – Epoch start time requires a manual calculation:

(Ground time) - (DANDE time) = Epoch Start Time (ground)

• Once epoch start time is recorded, can proceed to processing data

24

25

ACC Processing

ACC Data Inversion

26

• Back out the filtering and amplification. This has to be done assuming an ideal band pass

• The signal is then converted to an acceleration using honey-well provided temperature curves

• Temperature compensation is also applied to each analog filter string.

ACC Data Inversion

27

• The actual data product of interest in the density of the environment

• Blue labeled parameters are either modeled or measured prior to launch

• Green parameters are either modeled, or measured by NMS

• The force is the desired product from ACC, being manifested as the amplitude of the drag signal

221

WD VVACam

ACC Data Characteristics

28

• We have previously used the signals periodicity for analog processing.

• How else can we use this to better improve the data resolution?

ACC Data Characteristics

29

• Potential algorithms utilizing this feature

ACC Direct LSF Fit

30

• Treats each accelerometer independently

• Allows for each string to be compared

• Simple and relatively easy

• Well defined fit uncertainty

0 1 2 3 4 5 6

1.0

0.5

0.0

0.5

1.0

Tim e sD

rag

Sig

nalunitless

LSF Fit AC C D ata

ACC Coupled LSF Fit

31

• Takes each string, and applies a phase shift such that each signal is the same

• Signals are then summed, giving a single sinusoid, with amplitude 6 times the original signal

• In certain situations (very low signal to noise ratio) improves the co-efficient uncertainty from the LSF

0 2 4 6 8 10 12

4

2

0

2

4

6

Tim e sDra

gSig

nalunitless Singular LS F Fit

0 1 2 3 4 5 6 7 15

10

5

0

5

10

Tim e sDra

gSig

nalunitless C oupled LSF Fit

ACC Coupled LSF Fit

32

• This data was simulated with added Gaussian noise of the same amplitude, and variance 50% of the signal amplitude.

• The singular fit gives a 42% uncertainty in amplitude

• The coupled fit provides a 17% uncertainty in fit parameter

0 2 4 6 8 10 12

4

2

0

2

4

6

Tim e sDra

gSig

nalunitless Singular LS F Fit

0 1 2 3 4 5 6 7 15

10

5

0

5

10

Tim e sDra

gSig

nalunitless C oupled LSF Fit

ACC Very Noisy Data

33

• With the current orbit DANDE will be outside the designed density for a substantial amount of time

• To account for this a developed system for pulling parameters from extremely noisy data

• The simulated data has a noise variance twice that of the signal amplitude. The amplitude here is 2.

0 10 20 30 40 50 60

10

5

0

5

10

Tim e sDra

gS

igna

lunitless

Nois y AC C D ata

ACC Very Noisy Data

34

• The LSF programs, coupled or singular, are unable to fit to this data

0 10 20 30 40 50 60

10

5

0

5

10

Tim e sDra

gS

igna

lunitless

Nois y AC C D ata

ACC Fourier Based Filtering

35

• When filtered through an FFT algorithm, we can see that the signal frequencies are still dominant in the signal

• We can take advantage of this by isolating just the signal we want, which will have a well defined frequency based on the spin rate of the satellite.

0 10 20 30 40 50 60

10

5

0

5

10

Tim e sDra

gSig

nalunitless

Nois y AC C D ata

0 10 20 30 40 50 600

5

10

15

20

Pow er Spec trum

Freq

uenc

y

Pow er Spec trum of N oisy D ata

ACC Fourier Based Filtering

36

• Filter is applied digitally, simply setting all frequency components outside of a window around the dominant frequency to zero

• When passed back through an inverse transform we obtain the original signal, with aberrations around the edges from the distortion of the filtering

0 10 20 30 40 50 600

5

10

15

20

Pow er Spec trum

Freq

uenc

y

Filtered Pow er Spec trum of N oisy D ata

0 10 20 30 40 50 60 2

1

0

1

2

3

4

Tim e sDra

gSig

nalunitless

Filtered AC C D ata

Conclusion

37

• Provides a novel approach to characterization of the atmosphere of low earth orbit

• TOM correlates DANDE’s system time to the local ground time which allows time mapping to varying degrees of accuracy

• Forced drag signal modulation enables enhanced on board filtering, and improved data fitting techniques.

38

Drag and Atmospheric Neutral Density Explorer

Winner of University Nanosat V Competition

Questions?