A. Lobo Sagamihara, 13 November 2008 1
LTP diagnosticsLTP diagnostics
Alberto LoboICE-CSIC & IEEC
A. Lobo Sagamihara, 13 November 2008 2
Why is diagnostics analysisneeded in the LTP?
for . This is so very demanding a previousLPF mission is planned, with a relaxed sensitivity requirement:
LISA’s top level sensitivity requirement is:
1/ 24
1/ 2 15 -1/22
0.1mHz m( ) 3 10 1 1 Hz
8mHz sa
fS f
f
0.1 mHz 0.1 Hzf
2
1/ 2 14 -1/22
( ) 3 10 1 Hz ,3 mHza
f mS f
s
1 mHz 30 mHzf
LISALISA::
LPFLPF::
Even if LPF works to top perfection, a fundamental queston remains:
How do we make it to LISA’s sensitivity?
A. Lobo Sagamihara, 13 November 2008 3
DDS: Data Management& Diagnostics Subsystem
Diagnostics items:
• Purpose:– Noise split up
• Sensors for:– Temperature– Magnetic fields– Charged particles
• Calibration:– Heaters– Induction coils
DMU:
• Purpose:– LTP computer
• Hardware:
– Power Distribution Unit (PDU)– Data Acquisition Unit (DAU)– Data Processing Unit (DPU)
• Software:
– Boot SW– Application SW:
Diagnostics items Phase-meter Interfaces
A. Lobo Sagamihara, 13 November 2008 4
Diagnostics items use
The methodology:
1. Split up total noise readout into parts –e.g., thermal, magnetic,...2. Identify origin of excess noise of each kind3. Orient future research in appropriate direction for improvement
1. Sensitive diagnostics hardware needs to be designed and built
2. a) Suitable places for monitoring must be identified. b) Algorithms for information extraction need to be set up.
3. Creative research activity thereafter...
A. Lobo Sagamihara, 13 November 2008 5
Noise reduction philosophy
Problem: to assess the contribution of a given perturbation to the total noise force.
Approach: 1) Apply controlled perturbation to the system
2) Measure “feed-through” coefficient between force and perturbation:
int( )f
F
3) Measure actual with suitable sensors
4) Estimate contribution of by linear interpolation:
int ( ) ( )f F
6) Subtract out from total detected noise:
red int int ( )f f f
5) Check against a physical model of perturbations
A. Lobo Sagamihara, 13 November 2008 6
Location NTCs Comments Req. No
Optical Bench 4 Near base corners 3.1
Optical Windows 4 2 per OW 3.2
Inertial Sensors 82 on each of the outer x-faces of the IS EH
3.3
LCA mounting struts 6 1 per strut, near centre 3.4
Total 22 --- ---
Sensor sensitivity requirement: 10-5 K/Hz , 1 mHz < f < 30 mHz
Global LTP stability requirement: 10-4 K/Hz , 1 mHz < f < 30 mHz
Thermal and sensor requirements
A. Lobo Sagamihara, 13 November 2008 7
Thermal damper
1/ 2 1/ 2inside boundary, ,S r H r S
Transfer function
DMU test campaign
A. Lobo Sagamihara, 13 November 2008 8
DMU test campaign
DMU EQM
A. Lobo Sagamihara, 13 November 2008 9
DMU test campaign
Insulating anechoic chamber for the test
Open thermal damper,wiring and DMU
A. Lobo Sagamihara, 13 November 2008 10
DMU test campaign
S/H A/DIntegrator
N M Z-1 2
1/2
+
-
Ax(t)+
n(t)SN(f)
+m(t)
y (t)PSDy(f)
PSD1(f) PSD3(f) PSDdiff(f)
Analog processing Digital processing
PSDint(f)
nA/D(t)SNA/D(f)
+
+
FEE Readout electronics PC data acquisition program
FEE Analog PCB FEE A/D PCB
A. Lobo Sagamihara, 13 November 2008 11
DMU test campaign
Example run0.5 K/s
dT
dt
A. Lobo Sagamihara, 13 November 2008 12
DMU test campaign
0.5 K/sdT
dt
A. Lobo Sagamihara, 13 November 2008 13
DMU test campaign
0.5 K/sdT
dt
A. Lobo Sagamihara, 13 November 2008 14
NTCs magnetic properties
In IEEC we eventually discovered the NTCs, used both astemeperature sensors and heaters in the EH, have magneticproperties:
Nickel parts areinside the device
Damaged sample!
A. Lobo Sagamihara, 13 November 2008 15
NTCs magnetic properties
First magnetisation curves
Damaged sample!
A. Lobo Sagamihara, 13 November 2008 16
NTCs magnetic properties
Summary of magnetic moments of the screened NTCs (Am2):
A. Lobo Sagamihara, 13 November 2008 17
NTCs magnetic properties
Assessment of magnetic noise generated by NTCs:
background
1/ 2 1/ 2
0Bx
F x
VS B S B B
B B
, background
1/ 2 1/ 2
0B xx
F BS S B BV
BB
These two terms add, and are due to the quadratic couplingof the magnetic field due to TM susceptibility. Independentcalculations done at IEEC and ESA reach similar conclusionsthat the noise is not very significant for the LTP budget,
A. Lobo Sagamihara, 13 November 2008 18
NTCs magnetic properties
Excerpt from S2-EST-TN-2026 (2-Oct-2008):
A. Lobo Sagamihara, 13 November 2008 19
Thermal sensing summary
• Fully compliant with requirements
• Thermal gradients slightly distort performance at low frequency
• Can be improved by dithering the bridge voltage with a saw-tooth signal --not applicable to LPF, but useful for LISA.
• Magnetic properties of NTCs: not an issue of major concern, though some precautions are recommended.
• Research ongoing for improvements for LISA, encouraging first results
A. Lobo Sagamihara, 13 November 2008 20
Magnetic disturbances in the LTP
Main problem is magnetic noise. This is due to various causes:
• Random fluctuations of magnetic field and its gradient• DC values of magnetic field and its gradient• Remnant magnetic moment of TM and its fluctuations• Residual high frequency magnetic fields
Test masses are a AuPt alloy
70% Au + 30% Pt
of low susceptibility
510
and low remnant magnetic moment:
8 20 10 A mm
a = 46 mmm = 1,96 kg
A. Lobo Sagamihara, 13 November 2008 21
If a magnetic field B acts on a small volume d3x with remnant magnetisationM, and susceptibility , the force on this small volume is:
30 02
d
d x
FM B B M B B
Total force requires integration over TM volume, or:
00
V
BF Bm
Quantification of magnetic effects
Most salient feature is non-linearity of force dependence on B.
A. Lobo Sagamihara, 13 November 2008 22
Magnitude Value
DC magnetic field T
DC magnetic field gradient T/m
Magnetic field fluctuation rms PSD 650 nT/Hz
Magnetic field gradient rms PSD 250 (nT/m) /Hz
Magnetic susceptibility 10-5
Remnant magnetic moment 10-8 Am2
Summary of magnetic requirements
A. Lobo Sagamihara, 13 November 2008 23
Magnetic sensor requirements
Req 2: A minimum 4 magnetometers.Req 2-1: Resolution of 10 nT/sqrt(Hz) within MBW.Req 2-2: Two magnetometers located along x-axis, each as close as possible to the centre of one of the TM, not farther than 120 mm.Req 2-3: The other two may be offset from the x-axis by an amount not larger than 120 mm (TBC), their x coordinate should fall between the IS's at distances TBC.Req 2-4: Operation of magnetometers compatible with full science performance.Req 2-5: Final exact choice of magnetometer locations depends on final configuration of magnetic sources. Limited adjustment of magnetometer positions to within +/- 10 cm along x, y and z must be allowed until system CDR.
A. Lobo Sagamihara, 13 November 2008 24
Magnetometers type:
• 4 Flux-gate, 3-axis magnetometers
Model is TFM100G2 fromBillingsley Magnetics.
Magnetometer layout
Areas for magnetometeraccommodation
A. Lobo Sagamihara, 13 November 2008 25
Magnetometers’ accommodation
A. Lobo Sagamihara, 13 November 2008 26
Magnetic field reconstruction
• Exact reconstruction not possible with 4 magnetometers and around 50 dipole sources (+solar panel)
• Tentative approaches attempted so far:
• Linear interpolation• Weighted interpolation –various schemes• Statistical simulation (“equivalent sources”)
• Best possible: Multipole field structure estimation:
• Only feasible up to quadrupole approximation, though, given only four magnetometers in LCA.
A. Lobo Sagamihara, 13 November 2008 27
Multipole reconstruction theory
B x x
In vacuum,
A multipole expansion of B follows that corresponding to (x):
,
, , 1lm lml m
a r Y r x n x n n
The coefficients alm(r) depend on the magnetisation M(x). In anobvious notation, structure is:
)
,
lm
l m
(B x xB
A. Lobo Sagamihara, 13 November 2008 28
Multipole reconstruction theory
Evaluation of multipole terms is based on some assumptions:
1. Magnetisation is due to magnetic dipoles only
2. Such dipoles are outside the LCA.
Somewhat legthy calculations lead to:
with
) 0
4 lll
mm
lmM r Y
(B x n
*
31
4
2 1lm
lm l
YM d x
l r
n
M x
A. Lobo Sagamihara, 13 November 2008 29
Multipole reconstruction theory
Idea is now to fit measured field values to a limited multipoleexpansion model. Arithmetic sets such limit to quadrupole:
)
1model
lml
l m l
(B xB xL
Fit criterion is to minimise squared error:
4
2measur
2l
1m eed ods s
slmM
B xB x
2
0lmM
A. Lobo Sagamihara, 13 November 2008 30
Multipole reconstruction theory
Arithmetics of reconstruction algorithm:
Data channels: 12
Mlm dipole: 3
Mlm quadrupole: 5
Mlm octupole: 7
3+5 = 8 < 12 => some redundancy, OKthese 3 are uniform field components
3+5+7 = 15, 3 unknowns too many!
Summing up:
• Full multipole structure up to quadrupole level.
• This is poor, only first order polynomial approximation
• Errors large --easily 300%, and rather unpredictable
• Weighting procedure seems better, but also unpredictable
A. Lobo Sagamihara, 13 November 2008 31
Multipole reconstruction theoryS
imu
lati
on
A. Lobo Sagamihara, 13 November 2008 32
Multipole reconstruction theory
Bo
tto
mli
ne
(qu
alit
ativ
e)
A. Lobo Sagamihara, 13 November 2008 33
Ways to improve magnetic diagnostics accuracy:
• Fluxgates are:• Only 4• Far from TMs• Large size –long sensor heads, ~2 cm
We have recently started preliminary activites at IEECto assess feasibility of using AMR’s as an alternative tofluxgates. This kind of research is intended for LISA.
Multipole reconstruction theory
Summary:
A. Lobo Sagamihara, 13 November 2008 34
Magnetometers tests
-metal enclosure
Billingsley TFM
A. Lobo Sagamihara, 13 November 2008 35
Magnetometers tests
Billingsley TFM
LTP req.
A. Lobo Sagamihara, 13 November 2008 36
Magnetometers comparative
A. Lobo Sagamihara, 13 November 2008 37
SQUID test of AMR device
A. Lobo Sagamihara, 13 November 2008 38
Magnetic diagnostics summary
• Fluxgate magnetometers are extremely sensitive• Sensor core is large => space resolution limits• Sensor core is permalloy => distance to TM constraints• Box is large and somewhat heavy => few sampling positions
• AMRs indicate good sensitivity• They are very tiny• Weakly magnetic –due to small mass• More thorough investigation needed –underway at IEEC
A. Lobo Sagamihara, 13 November 2008 39
Ionising particles will hit the LTP, causing spurious signals in the IS.
These are mostly protons (~90%), but there are also He ions (~8%)and heavier nuclei (~2%).
Charging rates vary depending on whether
• Galactic Cosmic Rays (GCR), or• Solar Energetic Particles (SEP)
hit the detector, as they present different energy spectra.This has been shown by extensive simulation work at ICL.
Therefore a Radiation Monitor should provide the ability to distinguishGCR from SEP events.
This means RM needs to determine energies of detected particles.
Use of RM is to obtain frequent determinations of charging rates, andcorrelate them with IS charge management system.
Radiation Monitor
A. Lobo Sagamihara, 13 November 2008 40
Radiation Monitor
ICL simulations,based on GEANT-4.
(Peter Wass andHenrique Araujo)
A. Lobo Sagamihara, 13 November 2008 41
Radiation Monitor
It is a particle counter with somespecific capabilities:
• It counts particle hits
• Retrieves spectral information (coincident counts)
• Can (statistically) tell GCR from SEP events
• Electronics is space qualified
A. Lobo Sagamihara, 13 November 2008 42
RM accommoation in S/C
Sun
Earth
A. Lobo Sagamihara, 13 November 2008 43
In place for test at PSI, Nov-2005
A. Lobo Sagamihara, 13 November 2008 44
In place for test at PSI, Nov-2005
A. Lobo Sagamihara, 13 November 2008 45
Radiation Monitor data
Radiation Monitor data are formatted in a histogram-like form.A histogram is generated and sent (to OBC) every 614.4 sec.
A. Lobo Sagamihara, 13 November 2008 46
Summary
Thermal diagnostics:• Sensors fully in place• Heaters: in place, and integrating in EMP• Improved sensitivity towards LISA in progress
Magnetic diagnostics:• Sensors fully in place, sub-optimum expected performance• Research in progress towards LISA• Coils fully in place, working on EMP
Radiation Monitor:• EQM built, green light to FM, PSI tests coming up• More on RM in Tim’s talk
A. Lobo Sagamihara, 13 November 2008 47
End of PresentationEnd of Presentation