• No results found

Wire Boom Deployment and Attitude of Spinning Free Falling Units in Sounding Rocket Experiments

N/A
N/A
Protected

Academic year: 2022

Share "Wire Boom Deployment and Attitude of Spinning Free Falling Units in Sounding Rocket Experiments"

Copied!
102
0
0

Loading.... (view fulltext now)

Full text

(1)

Wire Boom Deployment and

Attitude of Spinning Free Falling Units in Sounding Rocket

Experiments

MANIL DJOUADI

KTH ROYAL INSTITUTE OF TECHNOLOGY

SCHOOL OF ELECTRICAL ENGINEERING AND COMPUTER SCIENCE

(2)

R OYAL I NSTITUTE OF T ECHNOLOGY

EF233X D

EGREE

P

ROJECT IN

S

PACE

T

ECHNOLOGY

Wire Boom Deployment and Attitude of Spinning Free Falling

Units in Sounding Rocket Experiments

Author:

Manil D

JOUADI

Supervisor:

Nickolay I

VCHENKO

Examiner:

Tomas K

ARLSSON

A thesis submitted in fulfillment of the requirements for the degree of Master in Aerospace Engineering

in the

Department of Space and Plasma Physics

School of Electrical Engineering and Computer Science

September 19, 2018

(3)
(4)

iii

Abstract

The SPIDER experiment conducted in February 2016 made it possible to gather measurements in an aurora and is likely to bring a better understand- ing of instabilities occurring in it, especially of the Farley-Buneman instability.

To exploit properly these measurements, the attitude of the Free-Falling Units (FFUs) that gather the measurements must be well known. However the atti- tude reconstruction process is made harder as the camera included in the FFU for that purpose failed to provide usable data. Using raw GPS data and data from the Inertial Measurement Unit, and knowing the magnetic field’s behav- ior outside of the electrojet, the attitude of one FFU can be determined with an acceptable level of precision. This could extend to other scientific projects and provide an inexpensive alternative to the use of star trackers.

A re-flight by 2019 is scheduled for the SPIDER experiment. Improvements on the Boom Deployment Unit (BDU) are being investigated so their mecha- nism is more robust, as well as easier and faster to assemble. Getting a bet- ter understanding of it through functional analysis helped finding directions for improvement on the pin release mechanism, on the solutions to ensure the rotational guiding, on making the hardware resistant to vibrations during the launch, and on the connections between the measurement probes and the elec- tronics to process the measurements acquisition. A BDU including the modifica- tions proposed should be assembled and tested in the perspective of including the new design in the re-flight.

(5)
(6)

v

Sammanfattning

Uppskjutningen för SPIDER experimentet skedde i februari 2016. Mätningar från en Aurora/Norrsken/Polarsken samlades. Det är troligt att få en bättre förståelse för instabiliteter som förekommer i den, särskilt av Farley-Buneman instabilitet. För att kunna utnyttja denna mätning korrekt måste inställningen hos de frit-fallande detektorer kallades FFU -Free-Falling Units på engelska- som samlar mätningarna vara välkända. Men återställningsprocessen görs svårare genom en extra wobblingrörelse som gäller för enheterna. Genom att använda rå data från GPS-satelliter och data från tröghetsmätaren och genom att känna till magnetfältets beteende utanför elektrojet kan inställningen hos ett FFU avgöras med en acceptabel nivå av precision. Detta kan sträcka sig till andra veten- skapliga projekt och erbjuda ett billigt alternativ till användningen av stjärn- spårare.

Ett nytt flyg till 2019 är planerat för SPIDER-experimentet. Förbättringar av boom-implementeringsenhet -Boom Deployment Unit (BDU) på engelska- un- dersöks så att dess mekanism är robustare, så väl som enklare och snabbare att montera. Att få en bättre förståelse av den genom funktionell analys har hjälpt till att hitta riktningar för förbättringar på pinfrigörningsmekanismen, på lös- ningarna för att säkerställa rotationsstyrningen, avskärmning från vibrationer under uppskjutningen och på anslutningarna mellan mätproberna och elektron- iken för att bearbeta mätförvärv. Ett BDU inklusive de föreslagna ändringarna bör monteras och testas med tanke på att den nya designen ingår i omflyttnin- gen.

(7)
(8)

vii

Acknowledgements

I would like to thank people who helped be out throughout my thesis work.

First, I want to thank Juliette GAMOT for working along with me on most of the tasks concerning the attitude reconstruction of the Free-Falling spinning Units.

I wish to thank my supervisor, Nickolay IVCHENKO, for providing me oppor- tunity to carry out my Degree Project at the Department of Space and Plasma Physics (SPP) and for his advice and availability when I needed it. Advice and support from SPIDER team member Gabriel GIONO was also very helpful.

Per-Arne LINDQVIST forwarded my request when I was looking for a Degree Project to work on and I would like to acknowledge him for that. Finally, I thank Peter WEIJNITZ for providing me the Rocket Mounting Unit CAD files and for introducing me to them, Federico RORRO for advice he gave me while I was working in the SPP computer room, and Tomas KARLSSON for approving my request to produce thesis work for the KTH Shool of Electrical Engineering and Computer Science (EECS).

(9)
(10)

ix

Contents

Abstract iii

Sammanfattning v

Acknowledgements vii

1 General Introduction 1

1.1 Introduction . . . 1

1.2 Purpose of the report . . . 1

1.3 Description of the Hardware. . . 2

2 Attitude Reconstruction of the Free-Falling Units (FFU) 5 2.1 Problem to solve . . . 5

2.2 Processing the Recorded Data . . . 6

2.3 Retrieval of the FFU’s attitude pattern . . . 8

2.3.1 Description of the Process . . . 8

2.3.2 Results and Interpretations . . . 10

2.3.3 Error Factors . . . 12

Error Due to the Numerical Integration . . . 12

Measurement Uncertainty on the angular rate sensor . . . 14

2.4 Determination of Attitude Pattern’s Orientation . . . 18

2.4.1 Description of the Process . . . 18

2.4.2 Results and Interpretations . . . 22

2.5 Conclusions and Perspectives . . . 27

3 Redesign of the Boom Deployment Unit (BDU) Mechanism 29 3.1 Presentation of the System . . . 29

3.1.1 Description of the BDU Mechanism . . . 29

3.1.2 Functional Analysis . . . 30

3.2 Proposals for improvement . . . 34

3.2.1 Concerning the Pin Release Mechanism . . . 34

Current Design . . . 34

Proposal for improvement . . . 34

3.2.2 Concerning the Bearings . . . 36

Current Design . . . 36

Proposal for improvement . . . 36

3.2.3 Concerning the Spool . . . 37

(11)

Current Design . . . 37

Proposal for improvement . . . 38

3.2.4 Other Improvements . . . 39

3.2.5 Conclusions . . . 40

4 Conclusion 41 Bibliography 43 A Components of the FFU 45 A.1 List of the Components of the FFU and their Function . . . 45

B Matlab Codes 51 B.1 Routine for Running the Codes . . . 51

B.2 nonGPSgyrlines.m, code to Section 2.2 . . . 55

B.3 GyroTimeClock.m, code to Section 2.2 . . . 57

B.4 TimeIntQuat.m, code to Section 2.3 . . . 62

B.5 IfRotDiff.m, code to Section 2.4 . . . 66

C Siemens NX Drawings 79 C.1 Drawing of the Cork, with Dimensions . . . 79

C.2 Exploded Drafting View of the Improvements Proposed for the Pin Release Mechanism and to Get Rid of the Bearings . . . 81 C.3 Drawing Highlighting the Modification Proposed on the Spool . 83

(12)

xi

List of Figures

1.1 The FFU . . . 2

1.2 The BU . . . 2

1.3 The CU . . . 3

1.4 The BDU . . . 4

1.5 Artistic view of the BDU . . . 4

2.1 IMU data . . . 6

2.2 Angular Rates . . . 7

2.3 Direction cosines . . . 9

2.4 Quaternions . . . 11

2.5 Body Vector Tips Plot . . . 12

2.6 Error on the Quaternions Integration . . . 13

2.7 Angular difference, with and without integration error . . . 14

2.8 Angular difference between the b1 axes with and without offsets 15 2.9 Angular difference in degrees between the~b1axes with and with- out sensitivity to temperature . . . 16

2.10 Histogram of the angular rate noise along the x-axis . . . 17

2.11 Angular difference in degrees between the~b1axes with and with- out adding the extra noise . . . 18

2.12 C/N0s for the strongest satellites . . . 19

2.13 Theoretical C/N0s, orientation unchanged. . . 21

2.14 Body vector tips for the correct pattern orientation . . . 22

2.15 C/N0s measured and modelized . . . 23

2.16 Colormaps for the correlation . . . 25

2.17 Resume of the pattern orientation strategy . . . 27

2.18 Resume of the attitude reconstruction strategy . . . 28

3.1 Expression of the need . . . 31

3.2 Constraints for the BDU . . . 31

3.3 SADT A-0 . . . 33

3.4 FAST Diagram . . . 34

3.5 Solution for the the Pin Release Mechanism . . . 35

3.6 Solution for the Bearings . . . 36

3.7 Solution for the Pin Release Mechanism and Bearings . . . 37

3.8 Slots replaced by holes on the spool . . . 38

3.9 Spool used for DICE . . . 39

(13)

3.10 Casing for the Springs . . . 40

(14)

xiii

List of Tables

2.1 Euler angle sequence (in radians) for the 6 portions between 113 s and 120 s, and between 142 s and 149 s . . . 24 3.1 Description of the constraints the BDU must obey as well as the

criteria used for assessing the solutions proposed . . . 32

(15)
(16)

xv

List of Abbreviations

SPIDER Small Payloads for Invesigation of Disturbances in Electrojet by Rockets FFU Free-Falling Unit

RMU Rocket Mounting Unit

CU Common Unit

BU Bottom Unit

BDU Boom Deployment Unit IMU Inertia Measurement Unit DCM Direction Cosine Matrix C/N0 Carrier-toNoise-density ratio

SADT Structured Analysis and Design Technique FAST Functional Analysis and System Technique w.r.t. with respect to

ie id est

dps degrees per second

(17)
(18)

1

Chapter 1

General Introduction

1.1 Introduction

The Small Payloads for Investigation of Disturbances in Electrojet by Rockets (SPIDER) experiment aims at studying particles’ behavior in auroras, more specif- ically at characterizing the Farley-Buneman instability through electric field mea- surements. To reach this goal, it was proposed to send 10 Free-Falling Units (FFUs) in different direction inside an aurora, all of them being packed with 8 measurements probes. Such multipoints measurement techniques are being used since they were already tried successfully in other projects in the past (Ivchenko, Giono, and Jensen, 2017). They make it possible to have measure- ments from multiple zones of the aurora on various spatial and temporal scales, in order to improve the understanding of the phenomena under study.

The FFUs were deployed from a sounding rocket launched from the Esrange Space Center at Kiruna, Sweden, on February 2nd, 2016.

1.2 Purpose of the report

This reports aims at describing the thesis work related to the SPIDER project and carried out from April 2018 till September 2018. It falls into two main tasks.

The major task resides in reconstructing the attitude of the FFUs only from data collected from the GPS antenna and Inertial Measurement Unit (IMU). In- deed, the camera used to get the reconstruction from post-flight analysis gave images that turned out to be difficult to analyze, which led the team to investi- gate other solutions to retrieve the attitude.

The other task is to improve the wire Boom Deployment Unit (BDU) mech- anism for the measurement probes of the FFUs by making it more robust and easier to assemble.

These two aspects of the thesis work are depicted more thoroughly in the report.

(19)

1.3 Description of the Hardware

The 10 Free Falling Units (FFU) deployed from the sounding rockets are identi- cal. In stowed configuration, they have cylindrical shape with a 90 mm height and a 240 mm diameter. A CAD 3D view is shown in Figure1.1

FIGURE1.1: FFU CAD 3D Model

Each FFU falls into three “pancake” layers that are 30 mm high. The one located at the bottom is the Bottom Unit (BU) - see Figure1.2. It contains the electrical connections to the measurement probes through a flat cable, and to the mechanism for the deployments of the probes. But most importantly, it contains the electronics driving the probes and saving the data to memory.

FIGURE1.2: Photo showing the content of a BU1

A SMILE magnetometer measuring magnetic fields along 3 axes, and a star camera used for determining the FFU attitude are also implemented in the BU.

The camera provided footage that is difficult to analyze for two reasons :

• the mirror was not covering the full field of view, so the camera image combined in one image plane both picture of the stars above, and to the side of the FFU;

1https://spiderrocket.wordpress.com/2014/12/04/bottom-pancake-layer/

(20)

1.3. Description of the Hardware 3

• the aurora light made the background bright, which reduced the contrast of the star/sky, resulting in only few stars observed.

The layer located at the top is called the common unit (CU) - see Figure1.3.

It contains the components for external communication and positioning - GPS, VHF beacon, etc... -, and the Inertial Measurement Unit (IMU), which is made of an accelerometer and an angular rate sensor recording measurements along three axes.. Also, a parachute is stowed in the CU an is to deploy when the FFU approaches ground. The GPS and IMU data recorded by the CU components are used for reconstructing the attitude of the FFU - see Chapter 2.

FIGURE1.3: Photo showing the content of a CU2

Between the BU and CU is located the Boom Deployment Unit (BDU) - see Figure1.4and1.5. It contains the eight measurement probes:

• four probes to get electric field measurements by measuring the potential difference between the probe and the FFU body

• four Langmuir probes measuring the plasma properties - density and tem- perature - by recording the current for different potential biases of the probe

It also contains the mechanism to deploy them. The probes are gold coated spheres with a 25 mm diameter, deployed on wire boom. The ones for the elec- trical field (resp. current) measurements should be 2m (resp. 1m) away from the FFU and linked to the measuring electronics by titanium conductive wires.

The BDU suffers from a relatively sophisticated design that makes it long to as- semble and that jeopardizes its robustness. Reviewing the design and proposing solutions to improve it is one of the topics tackled in this report - see Chapter 3.

2https://spiderrocket.wordpress.com/2014/12/05/top-pancake-layer/

(21)

FIGURE1.4: BDU CAD 3D Model

FIGURE1.5: Logo of the SPIDER project, with an artistic view of

the FFU with the measurement probes deployed by the BDU3

A list of all the components located in the three units as well as a description of their function was made. In can be found in AppendixA.

3https://spiderrocket.wordpress.com/about/

(22)

5

Chapter 2

Attitude Reconstruction of the Free-Falling Units (FFU)

2.1 Problem to solve

The FFUs make it possible to measure electric fields and plasma parameters - density and temperature - within the aurora thanks to the probes. The measure- ments are recorded to be processed after the flight, once the units are retrieved.

However, it is necessary to know for each measurement the location and ori- entation for the corresponding instruments in the aurora. This is achieved by reconstructing the trajectory and attitude of the FFU.

Positioning the FFU’s center of mass at each time does not cause any major trouble since it is directly obtained from the GPS data. It is supposed to be already known for the study. Reconstructing the attitude of the FFUs consists in finding the orientation of the spinning and wobbling FFU at each moment.

Investigating new ways the retrieve the attitude was necessary because of the failure to get it from the star camera.

The possibility to reconstruct accurately the attitude of the FFUs using only raw GPS data as well as data from the IMU was considered. This would give a simple and inexpensive way to do attitude reconstruction in comparison with the most common solutions that require star trackers.

6 FFUs out of 10 were successfully recovered, but not all of them recorded all the necessary data properly. The thesis work has been pursued using data from CU06 - ie the data recorded by the Common Unit of the FFU #06.

First, the transformation done from the raw data recorded by the FFU to readable and usable data for the attitude reconstruction process is depicted.

Then, focus is put on the integration of the parameter chosen to describe the attitude of the FFU for arbitrary initial conditions. After that, it is possible to use the raw GPS data to find its right orientation at any time.

(23)

2.2 Processing the Recorded Data

A log file gathers the data retrieved in the CU06. This file contains data pack- ets, each packet corresponding to the measurements at one given time instant of either the GPS, the angular rate sensor or the accelerometer1. The packets are written in the chronological order, but among the GPS, accelerometer and angular rate sensor only the GPS packets contain temporal information, which can be retrieved after processing. The GPS packets were only recorded starting after the FFU ejection from the rocket. The log file is displayed as a string of hexadecimal characters which cannot be exploited directly.

Juliette Gamot divided the packets into separate lines (Gamot, 2018). She also put the GPS, angular rate sensor and accelerometer lines in separate files, and converted the last two into readable decimal data. Figure 2.1 shows the angular rate sensor and accelerometer lines after the process.

FIGURE 2.1: Lines from the angular rate sensor (left) and ac- celerometer (right) data file while the GPS is active

The angular rates from the angular rate sensor are useful for the study. They correspond to the last three columns of the angular rate sensor lines. Columns 5 6 and 7 correspond to the angular rates about the x, y and z-axes of the IMU respectively. They were first converted into signed decimals converting the dec- imal values of the columns into binary, and then using the two’s complement operation2 before converting the signed binaries obtained into decimals again.

After that, they were converted into rad/s. The result was obtained applying Formula2.1. In the formula, the function called TwosComplement also includes the two conversions between decimal to binary.

wx(resp.y,z) =TwosComplement(Dx(resp.y,z))⇥s180p (2.1) where

1Some packets also correspond to the analog-to-digital converter (ADC) used for housekeep- ing the log file

2https://en.wikipedia.org/wiki/Two%27s_complement

(24)

2.2. Processing the Recorded Data 7

wx(resp.y,z) : Angular rate about the x, y or z-axis in rad/s Dx(resp.y,z) : Digital value corresponding the angular rate about

the x, y or z-axis

s=70⇥10 3dps/digit : Sensitivity of the angular rate sensor for 2000 degrees per second (dps)

The first column in the angular rate sensor and accelerometer lines of the processed files from Figure2.1correspond to the order in which the line is writ- ten as a packet in the initial log file. It is worth noticing that there is a large gap between each line for both files. This suggests that there are much more GPS lines while that one is active. Having these line numbers plus the clock from the GPS makes it possible to build a time clock for the angular rate sensor in the region of interest, ie the region where the measurement probes are totally deployed and the FFU has a periodic wobbling movement. That region was de- termined plotting the angular rates contained in the angular rate sensor data file - see Figure2.2.

(A) (B)

FIGURE 2.2: Angular rates in rad/s along the x, y and z axes of the IMU, with respect to the line number the angular rate sensor measurement; Close-up on the rates corresponding to the region of

interest (B)

The behavior of the FFU can be observed reading the plots. After a transition period until line 23 000, the FFU spins with a slightly oscillating angular rate of 6 rad/s about the z-axis. Afterward around line 25 000, the spin rate decreases until 2 rad/s as the probes deploy to keep the conservation of the angular mo- mentum. The FFU wobbles periodically until line 47 000, after which it probably

(25)

gets too affected by the friction from atmospheric particles. The last portion may correspond to an FFU that already landed on the ground. The region of interest was set between lines 27 481 and 46 442 in the angular rate sensor file, which correspond to lines 10 572 124 and 151 140 248 in the initial log file. This is the area where the angular rate sensor clock should be built. The closest GPS line preceding line 10 572 124 is line 10 570 320 and corresponds to 40s.

The following formulas were implemented into a Matlab code - see Ap- pendixB- to make the angular rate sensor clock:

Tgyro(1) = t⇤ (Lgyro(1) L40) 1) +Tstart;

8i=J2 , nK Tgyro(i) = Tgyro(i 1) + (Lgyro(i) Lgyro(i 1) p(i))⇤dt (2.2) where

Lgyro(i) : Number of the ithangular rate sensor line coming after 40s of measurements

Lgyro(1) = 10572124 : Initial value for Lgyro

Tstart =40s : Time for which the probes are completely deployed n=54446 : Number of angular rate sensor data lines located inside

the region of interest

L40 =10570320 : Line number in the log file corresponding to 40s of data for the GPS

p(i) : Number of lines that are not GPS lines located between angular rate sensor lines i-1 and i

dt = 2642581 s : GPS sample time when it is active

After building the clock it was observed that the region of interest contains measurements between 40s and 223s. Attitude reconstruction of the FFU should be made on that interval as it contains the two 20s portions where the FFU goes through the aurora, first upleg and then downleg (Ivchenko, Giono, and Jensen, 2017).

2.3 Retrieval of the FFU’s attitude pattern

2.3.1 Description of the Process

A way to describe the orientation of the FFU at any is to watch the trajectory of axes fixed to the FFU’s body in the inertial frame. This has the advantage to be easy to visualize and to be directly exploitable when it comes to match the attitude with the measurements from the probes. Hence, this is what is targeted for end of the process. The vectors considered are the axes of the IMU. They are used as the base vectors b~1, ~b2 and b~3 of the body-fixed frame B. They can be described w.r.t. the inertial frame N as shown in Figure2.3:

(26)

2.3. Retrieval of the FFU’s attitude pattern 9

FIGURE 2.3: Figure showing the direction cosines and axes from

body and inertial frames (Schaub and Junkins,2009)

Using the parameters of Figure 2.3 the vectors b~1, ~b2 and ~b3 can be written with respect to the inertia-frame-fixed vectors n~1, n~2 and n~3 by the following equation:

8i =J1 , 3K ~bi =cos(ai1)~n1+cos(ai2)~n2+cos(ai3)~n3 (2.3) The Direction Cosine Matrix (DCM) CBN makes it possible to describe the orientation of the FFU. An expression of its coefficient that uses the parameters defined above can be written (Schaub and Junkins,2009):

CBNij =cos(aij) = ~bi.~nj (2.4) The orientation of the body axes can be found as soon as the DCM is known.

Hence, one must find an method to get back to the DCM using the data available.

The DCM at each moment can be found directly from the angular rate given by the angular rate sensor. However, the quaternion notation is preferred for de- scribing the attitude as it is more compact, and because is makes the kinematics equations simpler. The quaternion notation used is described in the following equation (Schaub and Junkins,2009):

b= b0+ib1+jb2+kb3 (2.5)

(27)

where

b0=cos f ; b1=e1sinf2 ; b2 =e2sinf2 ; b3 =e3sin f2 where

f, e1, e2and e3being the Euler eigenaxis rotation parameters defined by:

cos f= (CBN11 +CBN22 +CBN33 1); e = 2 sin f1 2

4CBN23 CBN32 CBN31 CBN13 CBN12 CBN21

3 5

A change of orientation corresponds to a rotation of the body frame with angle faround e.

Using these notations, the DCM can be written as follows (Schaub and Junk- ins,2009):

CBN = 0

@b20+b21 b22 b23 2(b1b2+b0b3) 2(b1b3 b0b2) 2(b1b2 b0b3) b20 b21+b22 b23 2(b2b3+b0b1) 2(b1b3+b0b2) 2(b2b3 b0b1) b20 b21 b22+b23

1

A (2.6)

Hence the orientation can be retrieved.

The quaternions can be integrated from the angular rates w1, w2and w3 by solving the following matrix differential equation (Schaub and Junkins,2009):

2 66 4

˙b0

˙b1

˙b2

˙b3 3 77 5 = 1

2 0 BB

@

0 w1 w2 w3

w1 0 w3 w2

w2 w3 0 w1 w3 w2 w1 0

1 CC A

2 66 4

b0

b1

b2 b3

3 77

5 (2.7)

This was done numerically using the Matlab function ode45 that applies a Runge-Kutta solving method. The Matlab code used for the integration can be found in AppendixB. The initial conditions used for the integrations are b0=1, and bi = 0 for i = J1 , 3K. Such a set of quaternions correspond to CBN being the identity matrix, hence to body axes equal the the base vectors of the inertial frame. The FFU after deployment of the probes is expected to have an attitude close to that one. Only a small orientation change would then result in a correct attitude. This is useful for Section 2.4.

2.3.2 Results and Interpretations

Figure2.4shows the evolution of the quaternions with time on the whole region of interest.

The periodicity read on the angular rate plots - Figure2.2 - can be observed here. b3varies with a high amplitude compared to b1and b2, which means that the FFU wobbles close to its spin axis. This is a logical behavior for a wobbling oblate body.

(28)

2.3. Retrieval of the FFU’s attitude pattern 11

FIGURE2.4: Evolution of the quaternions with time in the region of interest

Using Equations 2.6 and 2.4, it is possible to possible to get the Cartesian coordinates of the body axes ~b1, b~2 and b~3 in the inertial frame for the given initial conditions. Figure2.5 shows the trajectories of the tips of vectors ~b1, ~b2 and~b3in that frame.

The behavior of the FFU body wobbling close to its spin axis can be observed.

(29)

FIGURE2.5: Trajectory of the tips of vectors~b1, ~b2 and~b3(blue) in

the inertial frame for initial conditions b0=1, and bi =0 for i=1:3

2.3.3 Error Factors

The integration of the quaternions is subject to several errors and uncertainties.

Error Due to the Numerical Integration

There is a slight error for each quadruplet of quaternions integrated. Summing the square of each of them should result in finding 1, which is not the case.

Hence, the error can be defined as:

e =1 b20 b21 b22 b20 (2.8) Figure2.6shows the error factor epsilon with time.

(30)

2.3. Retrieval of the FFU’s attitude pattern 13

FIGURE2.6: Evolution of the quaternions integration error factor with time

This results in a drifting effect as the integrations go on. It is important to quantify this effect to evaluate its impact on attitude reconstruction final results.

Let’s assume all the error is contained in beta1. beta1 should be then replaced by sign(b1)⇥qb21+e. The integration of the quaternions has be simulated again after including such an error. Figure2.7 shows the angular difference between the x body axes with and without the error with time.

The drifting can be clearly observed, with a reset around 140s that still needs an explanation. A few peak values can also be observed, but they do not out- reach 3⇥10 6degrees.

According to the scientists exploiting the results of the attitude reconstruc- tion data, 10 5degrees are not noticeable in electric field measurements. Hence, such errors can be considered as neglectable for the purpose of their research.

(31)

FIGURE2.7: Evolution with time of the angular difference between the x body axes including the error due to integration and the ones

not including it, in degrees

Measurement Uncertainty on the angular rate sensor

The angular rate sensor measurements are subject to two types of uncertainties3:

• The sensitivity due to the temperature variations, which is defined as a per- centage of the measured value. For the model used in the FFU (L3G4200D), it is equal to +/- 2% ;

• The constant offset that applies on all the measurements, which is called

“Digital zero-rate level” on the technical sheet of the angular rate sensor.

It varies between +/- 75 dps, which cannot be neglected.

It is rather important to take these uncertainties into account, particularly the digital zero-rate level that can impact the measurements very importantly.

About the Zero-Rate Level:

Even though the technical sheet of the angular rate sensor shows a large range for the zero-rate level values, the variations of the angular rates suggest that the offset is very low. The mean values of the rates generating the wobbling move- ment during the periodic movements under study are 2.02 dps for the x-rate and -8.78 dps for the y-rate. For an ideal wobbling movement, these two rates should be equal to zero. It can thus be inferred that the zero-rate level has an order of magnitude similar to the mean values given above.

3https://www.parallax.com/sites/default/files/downloads/

27911-angularratesensor-3-Axis-L3G4200D-Guide-v1.1.pdf

(32)

2.3. Retrieval of the FFU’s attitude pattern 15

angular rate sensor and accelerometer measurements are supposed to start before the launch. This means that the IMU is taking measurements while the FFU is immobile for a certain portion of time. If a look is taken at the very beginning of the angular rate sensor data - first 16728 lines of angular rate sensor data processed file, first 200 001 lines of the log file containing GPS and IMU data, oscillations very close to a certain angular rate can be observed - see Figure 2.2. This rate is:

• 0.0189 dps for the x-axis

• 0.2466 dps for the y-axis

• -0.1781 dps for z-axis

It can be inferred that they correspond to a time slot during which the FFU is actually immobile, hence that these rates are the offsets - or Digital zero-rate level - for the angular rate sensor.

If this is the case, they must be injected in the quaternion integration algo- rithm to get accurate results. Indeed, Figure2.8shows important difference - up to 32 - for the~b1axis whether the offsets are injected or not, 3.6 on a 20s aurora region.

FIGURE2.8: Angular difference in degrees between theb~1axis with and without adding the inferred digital zero-rate level

The inferred digital zero-rates level were injected in the measurements for the rest of the study.

(33)

However, the sensitivity issues related to the temperature are still to be taken care of.

Sensitivity due to Temperature Variations:

To evaluate the influence of the temperature on the attitude determination of the FFU, the quaternions were integrated with a 2% decrease added on the angular rate along the x-axis. Figure2.9shows the angle between the vector~b1with and without that decrease.

FIGURE2.9: Angular difference between the~b1axis with and with- out adding the 2% decrease on the angular rate along the x-axis of

the angular rate sensor

A variation of 12 can be observed on the duration of the interval of study.

For a 20s aurora region, variation is about 1.4 . This cannot be neglected. How- ever, that duration, as well as the duration of flight, could be considered as short enough so that the angular rate sensor is not subject to important temperature variations. Hence, the sensitivity due to temperature variations was neglected for the rest of the study.

About the Measurement Noise:

Seeking for the offsets to apply on the angular rates measurements draw made it possible to spot another potential issue for the accuracy of the results : the angular rate sensor data is subject to noise. This corresponds to the fluctua- tions observed on the angular rates while it is supposed to be immobile before the launch. That effect can be observed if the zone corresponding to the early

(34)

2.3. Retrieval of the FFU’s attitude pattern 17

measurements in Figure2.2is zoomed. Analyzing the way it affects the results makes it possible to know more about the potential error factors.

The array containing the early measurements for the angular rates along the x-axis of the angular rate sensor has a standard deviation of 0.0083 rad/s. A noise signal was built by making a normal distribution - or Gaussian, see Figure 2.10- out of it, and adding it to the angular rate along the x-axis of the angular rate sensor which is already subject to noise.

FIGURE2.10: Histogram showing the probability distribution con- sidered for the angular rate noise along the x-axis of the angular

rate sensor

Figure 2.11 shows the angle between the vector b1 with and without that extra noise.

The injection of the extra noise in the signal induced angular variations rang- ing between 0 and 0.94 on the time interval of study. There is no solution to clear this error factor but picking a angular rate sensor that is less subject to noise. However, this factor can be neglected depending on the precision re- quired for the measurement needed. It has been considered for the study that the angular rate sensor’s noise could be neglected.

The two time intervals where the FFU seems to be immobile do not give the same mean values for the angular rates. However, along the z-axis of the angular rate sensor they are relatively close - -0.0027 rad/s against -0.0031 rad/s at the beginning.

Indeed the damages consecutive to the FFU’s landing as well as the expo- sition to the extreme environment - Esrange Space Center during winter - may

(35)

FIGURE 2.11: Angular difference between the ~b1 axis with and without adding the extra noise on the angular rate along the x-axis

of the angular rate sensor

have triggered a change in the supply voltage level, while the zero-rate level is known for being quite sensitive to it (Vágner and Beneš, 2013). Hence, it has been decided to keep the offsets calculated from the data preceding the flight that are more reliable. One should also consider the possibility that the vibra- tion occurring during the launch might have an influence on the voltage.

After the quaternions are integrated, the attitude reconstruction process remains incomplete. Indeed, the initial orientation of the FFU still needs to be deter- mined. This is the purpose of Section 2.4.

2.4 Determination of Attitude Pattern’s Orientation

2.4.1 Description of the Process

Using time history of the carrier to noise ratio of the GPS signals received by the FFU’s commercial antenna from several GPS satellites makes it possible to get information on how the unit is oriented, as long as the positions of the satellites are known at each moment. It is the case for the study.

The FFU recorded raw GPS data in the L1 frequency band - 1575.42 MHz - within the region of interest. From the data available, Juliette Gamot could build the carrier-to-noise-density ratio - called C/N0- in decibels for the five satellites

(36)

2.4. Determination of Attitude Pattern’s Orientation 19

for which the antenna received the most intense signals. These satellites are labeled as #2, #6, #12 #25 and #31. Figure 2.12 shows the evolution of these ratios with time.

FIGURE2.12: Evolution of the C/N0s in decibels for the satellites

#2, #6, #12 #25 and #31 between 40 s and 60 s

A periodicity can be observed for each signal maximums correspond to situa- tion when the antenna approaches the orientation of the satellite, then gets away from it. The signals reach a maximum every 3-4s approximately. This is close to the spin velocity of 2 rad/s found in Section 2.2 for the FFU. An asymmetry can be observed on the peaks. For the five signals the ascending slope decreases as the peak is being reached, whereas such a decrease cannot be found after the signal passes the peak. The reconstruction strategy consists in building a model that gives an estimation of the C/N0 signal for any orientation of the attitude pattern determined in Section 2.3, and in comparing it to the one acquired by the antenna. The orientation changes of the pattern are obtained by applying 3-2-1, or yaw-pitch-roll, Euler angle sequences to it. The corresponding rotation matrix R(gamma,phi,rho) was applied to vectors~b1, ~b2and~b3to obtain the new vectorsb1~

new,b2~

new andb3~

new as shown in the following equation:

8i=J1 , 3K binew~ =R(y, f, r)~bi b~inew =

0

@cos q cos y sin f sin q cos y cos f sin y cos f sin q cos y+sin f sin y cos q sin y sin f sin q sin y+cos f cos y cos f sin q sin y sin f cos y

sin q sin f cos q cos f cos q

1 A~bi (2.9)

(37)

220 samples located between 0 and 2p were taken for yaw values For pitch and roll angles, 51 samples between p6 and p6 were considered. Indeed, the orientation of the FFU’s spin axis is not expected to vary a lot w.r.t. to the z- axis of the inertial frame, and the initial conditions chosen for the quaternion integration, gave an attitude pattern close to that axis.

The sampling makes 572 220 orientations available for the study. The num- ber of samples can be increased in order to achieve more accurate results. This would increase the complexity of the algorithm, hence the calculation time.

Since no direction diagram was available for the commercial antenna, it was inferred for the study that the b3~new body axis of the FFU coincides with the direction of the antenna. Therefore , a theoretical model was inferred to move from the angles between the orientation of the satellites and theb3~new axis to the shape of the C/N0signals. This is achieved applying the following equation:

8i =J1 , 5K C/N0,thi(t) = 10 log10

sinc(ai(t) 180 )2

; (2.10)

where

C/N0,thi(t): Theoretical shape of the C/N0signal for satellite i w.r.t time ai(t) : Angular distance between the unit vector oriented towards

satellite i and the direction of the FFU antenna w.r.t time

Figure2.13shows the C/N0s obtained if no orientation change is applied to the body axes.

The asymmetry observed in the measured C/N0s (see Figure 2.12) cannot be seen here. This suggests that it is not due to the dynamics of the FFU body, but to an asymmetry of the antenna diagram, since the model for the theoretical C/N0s have been chosen symmetric. The optimization strategy still takes the general variations of the C/N0s into account, but the direction antenna diagram should be measured experimentally in order to achieve a better model.

Thanks to the steps above, a model that gives an estimation of the C/N0

signal shapes for all the orientation samples is available. Reconstructing the correct attitude can be made by comparing the signals for each orientation with the one acquired by the antenna. This is achieved by doing cross-correlation on the derivatives of each. A discretized version of the following equation is applied get the cross-correlation coefficients:

(C/N0,measiC/N0,thi)(t) =

Z t1+6s

t1 C/N0,meas i(t)C/N0,thi(t+t)dt (2.11) where

C/N0,thi : Theoretical shape of the C/N0signal for satellite i C/N0,measi : C/N0signal measured from satellite i

t : Lag

t1 : Start time for the correlation process

(38)

2.4. Determination of Attitude Pattern’s Orientation 21

FIGURE2.13: Evolution of the theoretical C/N0s if the orientation of the body axes remains unchanged, between 107 s and 126 s

A maximum lag of +/-1s is allowed. The highest coefficient reached within the corresponding lag interval is kept for each orientation.

The orientation that gives the highest correlation coefficient is considered to be the right one for the FFU.

As depicted in Section 2.3, there are a few possible factors that may cause drift for the results as the time goes by. Mainly, they are the temperature varia- tions and the zero-rate level variations due to a possibly unstable supply voltage for the angular rate sensor. Hence, the comparison was not made on the whole region of interest to not include that drift. The comparison was then carried on portions with length equal to 1/30th of the 183 s of measurements. The spin velocity being close to 2 rad/s, each portion on which the correlation process is applied contains about 2 complete spin rotations.

Correlation coefficients are calculated for three consecutive portions each time to get a comparison. As a consequence, the comparison algorithm was carried on three times on time intervals of 18.3s. This is handy as it is close to 20 s seconds which corresponds to the regions where the FFU goes through an aurora. By choosing the right time interval, it is possible to focus the attitude reconstruction on the aurora regions.

(39)

2.4.2 Results and Interpretations

Figure2.14gives the trajectory of the body frame vector-tips with the right ori- entations found for the three portions between 107 s and 126 s. The view is normal to the inertial frame’s z-axis. Note : Matlab could not support the level of graphics required for plotting the figure. It switched to the OpenGL graphic interface, which renders lower quality figures.

FIGURE2.14: Plot showing the trajectory of the body frame vector- tips with the right orientations found for the three portions be-

tween 107s and 126s

It is shown that the three orientation found are close to each other and that the~b3axes patterns are slightly off the z-axis of the inertial frame.

Plotting the C/N0s corresponding to the right orientation helps to evaluate the accuracy of the algorithm. They are shown on Figure2.15, along with the C/N0s measured by the GPS satellites:

(40)

2.4. Determination of Attitude Pattern’s Orientation 23

(A) (B)

(C)

FIGURE2.15: Plot showing the C/N0s measured by the GPS satel-

lites (B) and the C/N0s from the model with the right orientation

(A and C), between 113 s and 120 s. (C) is a zoom of (A) for satellite

#25.

Peaks coincide for plots corresponding to satellites #02, #12, #25 and #31. The plots for satellite #06 coincide for all except the last peak.

The algorithm seems to give accurate results. Yet, the angles applied to the initial pattern to get the three orientations are not the same. Table2.1shows the 3 sets of Euler angles found between 113 s and 120 s, as well as between 142 s and 149 s.

(41)

TABLE 2.1: Euler angle sequence (in radians) for the 6 portions between 113 s and 120 s, and between 142 s and 149 s

Angle 113 s-120 s 142 s-149 s

Yaw 3.9710 4.9637

Pitch -0.2094 -0.1885

Roll 0.0419 -0.2723

The triplets of angles are different for each portion. Between 107 s and 126 s, they slightly differ but the comparison algorithm gives more contrasted results between 136 s and 155 s.

Colormaps were plotted to observe the local maxima in the volume of pa- rameters formed by the yaw, pitch and roll of the Euler angles sequences. Figure 2.16shows these colormaps for the orientation found between 113 s and 120 s.

(42)

2.4. Determination of Attitude Pattern’s Orientation 25

(A) (B)

(C) (D)

FIGURE2.16: Colormaps showing the cross-correlation coefficients across the parameter space shaped by the Euler angles of the 3-2-1 sequence, between 113 s and 120 s. (B), (C) and (D) are cuts from (A). The red color corresponds to the case when the correlation be-

tween the two signals is maximum.

Three local maxima can be observed. Two are observable for a pitch around -30 , a roll around 0 and a yaw either around 0 or 360 . The other one seems close to a yaw of 240 and low values for pitch and roll.

The three regions observed on the colormaps for the [107s,126s] region can be found again running simulations on several 18.3s intervals located at the be- ginning, middle and end of the whole time line. 5 local maxima areas appear regularly in the triplets maximizing the correlation:

• One region around [yaw,pitch,roll]=[4,-0.15,0] (in radians)

• One region around [yaw,pitch,roll]=[5,-0.2,-0.3]

• One region around [yaw,pitch,roll]=[3,0,0.15]

(43)

• One region around [yaw,pitch,roll]=[0.7,0.3,-0.4]

• One region around [yaw,pitch,roll]=[6,0.02,-0.4]

The latter two are very close to each other and it may be inferred that they ac- tually correspond to the same region, as they is a yaw gap of 2p between the two. However the first three ones are different. It gives close orientation for the z-axis, but the x and y axes are in differ. The 4 regions spotted correspond to three distinct orientations because of that. A solution should be proposed to de- termine which one is correct. Using the magnetic field measurements recorded outside of the aurora, where the field can be predicted, could make it possible to eliminate the two wrong solutions. Once this is done, simulations can be run on the whole timeline, and the orientation giving the right attitude for the FFU can be interpolated on the triplets found.

Figure2.17summarizes the strategy applied to find the right orientation for the attitude pattern.

(44)

2.5. Conclusions and Perspectives 27

FIGURE 2.17: Drawing resuming the orientation determination strategy of the attitude pattern found in Section 2.3

2.5 Conclusions and Perspectives

A recap of the strategy used to find the right orientation from the logfile is shown in Figure2.18.

A routine that explain the steps to follow to get the correct attitude of CU06 was written. It is available in AppendixB.

The attitude reconstruction algorithm detailed in this report would offer an innovative, cheap and simple alternative for researchers willing to conduct ex- periments in space or close to space. Improving it by getting better knowledge of the FFU’s antenna and by optimizing the time intervals for comparison could make it possible to approach or top the precision accessible by the solutions us- ing star trackers.

(45)

FIGURE2.18: Recap of the attitude reconstruction strategy

Finally, the WOLF - Wobbling Control system for spinning Free falling unit - experiment was conducted to show it was possible to prevent the FFUs from wobbling by using a reaction wheel properly oriented and controlled. The solu- tion proposed was proven to be efficient, but it does not make it possible to com- pletely suppress the wobbling. The thesis work presented in the report would then also be useful for the re-flight of the SPIDER experiment that is scheduled for 2019 and that should include a reaction wheel to minimize the wobbling.

(46)

29

Chapter 3

Redesign of the Boom Deployment Unit (BDU) Mechanism

3.1 Presentation of the System

3.1.1 Description of the BDU Mechanism

The Boom Deployment Mechanism, or BDU, aims at deploying 4 of the mea- surement probes 2 m away from the spool, and the 4 others 1 m away from it.

The probes are attached to conductive titanium wires and through springs that adapt to the tension in the wires. These wires are winded up on a spool that contains eight gutters on its surface - one for each wire. The electrical con- nections between them and the measurements electronics are made inside the spool. Access to the inside is allowed thanks to holes dug on the area were the gutters are.

The spool’s rotation is initially blocked by a disk with pins, the pins entering holes dug on the spool’s surface. The disk is spring loaded and maintained in position by a fishing line.

To release the mechanism, a thermal cutter cuts the line. The springs push the disk away from the spool which can then rotate. The rotation is controlled by a non-magnetic piezo-motor. As it rotates, the wires deploy to the outside with the probes. Contact with the hull and lid of the BDU is maintained through bearings all along the deployment. Also, the electronics located on the rotating spool are connected to the BU thanks to a flat cable winded up below the piezo- motor.

Five rotations are necessary for complete deployment. At the end, the probes are fully deployed, and tension in the wires is ensured thanks to the mechanical energy form centrifugal forces as the FFU is spinning.

Experience showed that the design could be improved as it took time to as- semble for the SPIDER team members and as deployment was not completed on some of the FFUs.

(47)

3.1.2 Functional Analysis

This part is made using a functional analysis approach. It aims at providing a better understanding of the BDU and of the criteria leading to the technical solu- tions chosen for the mechanical design. It is likely to help identify the potential issues of the design, therefore it is useful to seek for improvements.

The Boom Deployment Unit is designed as an answer to a specification the SPIDER team members had to deal with: How to deploy the measurement probes at a certain distance from the FFU - see Figure3.1?

(48)

3.1. Presentation of the System 31

FIGURE3.1: Expression of the need

The main function satisfied by the BDU is the one expressed in Figure 3.1.

However, several external factor also influence the way it should be designed.

They are constraints the design must comply with so that the BDU is conform to the specifications. Figure3.2and Table3.1describe these constraints and the criteria used to evaluate how they are observed.

FIGURE3.2: Diagram representing the external factors that apply constrains to the BDU

(49)

TABLE 3.1: Description of the constraints the BDU must obey as well as the criteria used for assessing the solutions proposed

Function Description Criterion Level

P(rincipal)

F(unction) Deploys the measurements probes of the FFU

Boom

Deployment Tests

Test passed

C(onstraint)1 Being easy and quick to

assemble

Assembler assessment, duration for assembling

Less than XX hours

C2 Complies with

the Swedish Space

Corporation (SSC) regulations

Review from the

SSC Approval from

the SSC

C3 Can be properly

integrated into the FFU

Design N/A

C4 Stays within the

allocated budget for the FFU

Set by the

allocated budget N/A

C5 Works with the

energy supply available

Electrical energy supply,

mechanical energy form centrifugal forces

48V ,Works with rotation of at least 1Hz

C6 Can be properly

integrated into the RMU

Design N/A

C7 Resists launch

conditions Vibrations tests on the shaker table

Tests passed

C8 Is robust under

climatic and particle

conditions inside the aurora

Deployment in close-to-vacuum conditions, robust for the temperature range the BDU is likely to meet

Deployment tests passed, robust for Temperatures between -43.3 C and 31.6 C

C9 Does not interact

with the outer magnetic field

No

ferromagnetic materials

Measurements for the SMILE magnetometer unchanged

(50)

3.1. Presentation of the System 33

Figures 3.1 and 3.2 give a better insight concerning the needs the engineer must comply with when designing the BDU. A more technically-oriented anal- ysis is necessary for him to find solutions fulfilling all the functions and con- straints.

Structured Analysis and Design Technique (SADT) helps identifying the in- puts, energy and information available for the BDU to fulfill its principal func- tion. Figure3.3shows a zero-level SADT that gives the global inputs available to obtain the desired output:

FIGURE3.3: SADT A-0 for the BDU

All the previous analyses make it easier to seek for technical solutions. The FAST (Functional Analysis System Technique) method helps identifying the tech- nical requirements for the design. Figure3.4 shows a FAST diagram detailing these requirements and the solutions that were found for the former design.

Experience made it possible to identify weaknesses in the former design. The functional analysis approach brings out the requirements the design failed to fulfill. Indeed, constraint C1 - see Figure3.2- emphasizes an issue when inter- acting with an external factor which is the SPIDER team, more specifically the members in charge of assembling the FFUs. The FAST diagram on Figure 3.4 highlights the technical solutions that caused failure for the deployment mecha- nism. FT11 and FT122 are the requirements that should be reviewed to increase the robustness of the design.

In Part 3.2, emphasis is put on the causes that make the deployment mecha- nism tough to assemble and likely to fail, and on the solutions proposed to solve these issues.

(51)

FIGURE3.4: FAST Diagram

3.2 Proposals for improvement

3.2.1 Concerning the Pin Release Mechanism

Current Design

The pin release mechanism is absolutely necessary for the deployment of the probes to start. Therefore is must be designed in a way so that its robustness is maximized.

However, the current design has disadvantages that jeopardize its reliability.

Experience proved that the mechanism activation relies on the way the assem- bling is made. On the one hand, the operator has to not push the pins too deep inside the spool’s holes otherwise they do not come out once the mechanism is activated. On the other hand, if the pins are not inserted deep enough they might pop out allowing the spool to rotate before or during the launch.

Proposal for improvement

The new design should minimize the impact the operator can have on the ro- bustness while assembling the BDU. An exploded view of the solution proposal is shown in Figure3.5.

The disk carrying the pins remains, but the pins themselves are replaced by 10mm diameter corks - see AppendixC. They have a spherical tip with diameter

(52)

3.2. Proposals for improvement 35

FIGURE3.5: Exploded 3D CAD view of the solution proposal for the pin release mechanism (without the springs)

8mm and they can be screwed in threaded holes manufactured on the disk. The spools holes are spherical with diameter 8mm and are 2 mm deep. This is lower than to the ones of the former design that were 4mm deep.

The operator positions the disk and corks pre-screwed on it so that the corks get inside the spool holes. Then, he attaches the disk the same way he used to for the former design. Finally, he screws the corks with higher torque to maintain the disk in its position. Such a design may improve the pin released mecha- nism’s robustness significantly.

First, it relies not only on the contact between the pins and the cylindrical sur- faces of the spool holes but on the contact with the spherical hole. The rotation prevention relied only on the shear stress applied to the pins’ cylindrical sur- faces, which could make it harder for them the get out once the mechanism is activated. With the spherical contact, the shear stress should be minor compared to the normal force applied by the corks at the bottom of the holes. It arises only once the release mechanism is activated, and the spherical contact should make the release smoother, therefore more reliable.

If experience shows that the spherical contact makes the mechanism not re- sistant enough to vibrations during the launch, it is still possible to move to cylindrical holes. The decrease of their deepness as well as the normal force ap- plied at the bottom and decreasing shear stress should still make the mechanism more robust.

Another advantage of such a design is that the bearings to ensure the contact between the upper part of the spool and the lid of the CU may no longer be necessary.

(53)

3.2.2 Concerning the Bearings

Current Design

The bearings have appeared to be a major issue concerning the difficulty in as- sembling the BDU. It takes time to position them, and several parts of the BDU assembling requires them to be positioned already. Hence, any unwanted move- ment for the operator may result in the bearing getting out of their position. He would have to retrieve all of them wherever they got spilled, and the assem- bling process would be sent back several steps backwards. Also, the contacts ensured with the bearings result in resisting torques that oppose the rotation of the piezo-motor. These efforts are likely to block the rotation when a small angle appears between the spool axis the axes of the hull and lid of the BDU.

Proposal for improvement

The solution proposal aims at totally getting rid the bearings. For the bearings located in the upper part of the spool, The improvements concerning the pin release mechanism make them unnecessary. A simple solution for preventing the spool to the slip out of the motor shaft could be considered. The bearings securing the contact between the spool and the BDU’s hull are replaced by a solution using a thrust washer and a bushing. The thrust washer is screwed on the spool while the bushing is only positioned in a housing manufactured on the hull. The surface of the normal contact between the washer and the bushing is minimized. The cylindrical contact between the outer part of the bushing and the inner part of the washer should have loose tolerancing. Figure3.6shows an exploded view of the solution proposal for the bearings.

FIGURE3.6: Exploded 3D CAD view of the solution proposal for replacing the bearings

(54)

3.2. Proposals for improvement 37

Such a design makes it possible to get rid of the bearings. The cylindrical contact is designed to host the non-normal forces in case a small angle appears.

Which no cylindrical contact, such an angle would result in an effort completely transferred to the plane contact with the washers and would be more likely to oppose to the piezo-motor’s rotation. Here, the cylinders carry some of the in- duced efforts, reducing the risks of gripping. The bushing and thrust washer should be made of non-magnetic materials.

Figure3.7 shows an exploded drafting view of the improvements proposed for the pin release mechanism and to get rid of the bearings. A larger version is available in AppendixC.

FIGURE 3.7: Exploded drafting view of the improvements pro- posed for the pin release mechanism and to get rid of the bearings

3.2.3 Concerning the Spool

Current Design

Winding up the titanium wire on the spool is one of the most fastidious steps for the persons in charge of assembling the BDU. Some decide to do it when after positioning the spool inside the hull, and some others prefer to do it before mounting the spool. But the process is long and tough in both cases for several reasons. Each gutter corresponds to one wire, and it is easy for one of them to slip into the wrong one while and after being winded up. A system using a small green component made of plastic is supposed the minimize that risk but it was found to lack efficiency. Also, tape is used to maintain tension inside the wires after they are mounted. There is also a risk for the wires to tangle up

References

Related documents

46 Konkreta exempel skulle kunna vara främjandeinsatser för affärsänglar/affärsängelnätverk, skapa arenor där aktörer från utbuds- och efterfrågesidan kan mötas eller

Generally, a transition from primary raw materials to recycled materials, along with a change to renewable energy, are the most important actions to reduce greenhouse gas emissions

För att uppskatta den totala effekten av reformerna måste dock hänsyn tas till såväl samt- liga priseffekter som sammansättningseffekter, till följd av ökad försäljningsandel

The increasing availability of data and attention to services has increased the understanding of the contribution of services to innovation and productivity in

Generella styrmedel kan ha varit mindre verksamma än man har trott De generella styrmedlen, till skillnad från de specifika styrmedlen, har kommit att användas i större

Närmare 90 procent av de statliga medlen (intäkter och utgifter) för näringslivets klimatomställning går till generella styrmedel, det vill säga styrmedel som påverkar

På många små orter i gles- och landsbygder, där varken några nya apotek eller försälj- ningsställen för receptfria läkemedel har tillkommit, är nätet av

Det har inte varit möjligt att skapa en tydlig överblick över hur FoI-verksamheten på Energimyndigheten bidrar till målet, det vill säga hur målen påverkar resursprioriteringar