7. Analog Sensor Studies
Along with the DCOPS optical sensor
testing, we tested a prototype CMS
Z transfer (-X
survey) between a reference MAB surface and the Transfer plate
system (ET1/ES1
mechanical link), a dual axis inclinometer on the Transfer
plate (Applied
Geomechanics 900) to define its orientation with respect to
gravity (Q,F), plus a prototype
CMS R transfer (+Y survey) between the
Transfer plate system
and a mock ME2/2 outer boundary (CSC panel) plus
a prototype R
connection between a mock ME2/2 inner boundary and a
ME2/1 outer boundary.
ME2/2 was simulated by a 50 mm wide, 11.2 m
long strip of CSC
panel. It may not be a good representation of a CSC
assembly because of
thermal flexure and free expansion.
The MAB Z connection was a distance
measurement between an
OMRON Z4MW40 laser
distance sensor and a MAB surface about 45 mm apart.
The sensor was
mounted on the end of a 1168 mm (1175
mm july-sept)
Aluminum tube. The
other end of the tube contacted a spring loaded potentiometer
LCP8-10 mounted on
the Transfer plate and measuring the gap to the tube.
There were two
temperature sensors T1 and T2 on the two ends of the tube.
They tracked each
other extremely well and provided temperature monitoring
to 0.1 deg K.
The R connections were done with
wire extension potentiometers
LPXA-2 (50 mm
stroke). All of these devices are
voltage output devices.
Finally temperatures around the
setup were measured with Analog
Devices AD592 current
source devices (1 mamp/deg
K) which terminated
on voltage dividers
near the DAQ giving 10 mv/deg K. All of
the sensors
were read out as
analog voltages by an HP 34970A data acquisition/switch
system which scanned
the sensors through an ADC to a serial port of
the DAQ computer. It
also provided timing (gating) of the lasers in DCOPS
measurements. With the low pass filtering installed
(determined by tests
at Fermilab), we were
able to limit noise effects (ADC fluctuations) on our
distributed system
(15 m lines) to ± 0. 75 mv.
We had calibrated all these sensors at Fermilab and the
results are
available
@http://home.fnal.gov/~leer/calibration/ISR_Analog_Sensor.
The effective
calibration of these sensors is given in the following Table MM.
The Transfer plate inclinometers are (x
= q,
y = f).
The Z4MW40 is the laser
distance sensor to
the MAB, ZLCP8-10 is the gap potentiometer between the
Transfer plate and
the Z tube, RLXPA-2-1 is the R transfer potentiometer
between ME1/2 and
ME2/2 CSCs, and RLPA2-2 is the R transfer potentiometer
between the Transfer
plate and the ME2/2 CSC.
Typical temperature coefficients are
(for inclinometers) span +0.03%/deg C
and zero 10-20 arc
seconds/deg C, for LPX2A potentiometers +0.03%/deg C,
for LCP8-10
potentiometer 400 ppm/deg C, for Z4MW40
laser distance sensor
.03%Full Scale/deg C
(for sensor and for amplifier). For our three test periods
in I4, the
temperature varied at most 1.5-2 deg C overall and in each period.
All the analog
sensors were referenced by a very stable 12 V
DC power supply.
Table MM. Analog Sensor
Conversions:
qx(tilt)(arc deg) = 10.4739Vx 0.0963Vx2
0.1714Vx3 + 0.0025Vx4 + 0.0026Vx5
qy(tilt)(arc deg) = 10.5002Vx + 0.1549Vx2
0.1942Vx3 - 0.0037Vx4 + 0.0062Vx5
ZZ4mW40
(mm) = (VZ4M + 16.180)/(0.4038) + ZFace position
ZLCP8-10(mm)
= (VLCP8)/(1.0987) + Z free point position
RLX-PA-2-1
(mm) = (11.928 (VLX-PA-2-1))/(0.2202) + R Zero
Extension (ME2/1-ME2/2)
RLX-PA-2-2
(mm) = (11.928 (VLX-PA-2-2))/(0.2188) + R Zero
Extension (ME2/2-Tr Plate)
Tn (deg K) = (VTn
offset(n))*100 where o(n) = 0.007,
0.032, 0.034, 0.032,
0.033,0.035 for n = 1,..,6
They are all very linear except the
inclinometers which have some polynomial
corrections for fluid
vessel shape. All sensors have small thermal coefficients.
Periodically we did special burst
runs of rapid multi measurements of
analog sensors only.
These define the short term (instantaneous) resolution of these
sensors. These runs are time stamped, so they can be
used for long term
studies. Tracking the
burst measurements throughout the running periods
provided a
redundant high statistics test on long
term stability/drift to compare
to normal data
samples. Typically, we took two hundred measurements in a burst run.
For short term resolution
estimates we took the worst standard deviations found in any
Run as the guaranteed
short term resolutions. The instantaneous resolution results are
given in the
following Table NN.
Table NN - Short Term
Resolutions:
sQx(Xtilt)= sfy (Ytilt) = 0.52 mv = 19.54
arc sec 12 V Analog Reference s
= 0.567 mv
sZ4M-MAB = 0.778 mv = 1.93 mm sLCP8-10 Z-Trans Plate =
0.517 mv = 0.5 mm
sME2/2-Trans Plate R= 0.523 mv = 2.4 mm sME2/2-ME2/1 R
=1.039 mv = 4.7 mm
AD592
Temperature sensors s = 1.032 mv = 0.10 deg
K (typical 0.05 deg K)
We have taken measurements of the
analog sensors each time we took any
measurements (any
laser combination) of the DCOPS sensor lines. Typically,
eight analog
measurements are made in a DCOPS cluster measurement. These
measurements were
used in each test period (about one week) to determine
Intermediate sensor
resolutions (with/without thermal drift corrections). All
sensors show some
small thermal dependence. Also our mechanical rulers;
the Z tube and the
ME2/2 CSC strip expand/contract in di urnal thermal cycling.
The Raw analog (no
thermal drift corrections) standard deviations of measurements
for the three test
periods are given in Table PP. below:
Table PP - Intermediate Term
Resolutions: (one run period ~week)
Use
data from normal cluster runs (cycles 8 measurements every 15 min); no
corrections, all points are raw data (LM302 off 4 measurements
/cycle):
Period sXtilt (Q) sYtilt (F) Z4M-MAB 12V
Ref
June s = 116.75 arc sec 58.31 arc sec 18.04
mm 1.06 mv
(3.11 mv)
(1.55 mv) (7.25
mv) *
July/Aug s = 97.23 arc sec 58.31 arc sec 32.22 mm 0.50 mv
(2.59 mv) (1.55 mv) (12.95
mv)
Sept s = 97.23 arc sec 19.56 arc sec 46.19 mm
4.17
mv
(2.59 mv) (0.52mv) (59.58 mv)
Period LCP8-10 (ZTP) ME2/2-TP
R ME2/2 ME2/1 R
June s = 23.60 mm 14.03 mm 68.25 mm**
(25.90 mv) (3.07 mv) (15.03 mv)
July/Aug s = 28.80 mm 18.92
mm 44.70
mm**
(31.60 mv) (4.14 mv) (9.843 mv)
Sept s = 11.80
mm 14.21 mm 75.29 mm**
(12.95 mv) (3.11 mv) (16.58
mv)
* Partial period data
** Connected to free floating end of3.2m
long CSC panel strip (50mm wide)
Finally, we have combined the analog
sensor measurements (LM302 laser off) of
the three separated
test periods in real time. From the combined file, we determine the
raw standard
deviations (resolutions), the long term
drifts, comparisons to temperature
sensors, and drift
corrected standard deviations. The resolutions are given in Table QQ
below:
Table QQ -Long term Resolution: June-July-August-Sept Data
fit across 8.51M sec
All
data; no corrections, no drift
removal; other than DV shifts for LM302on data
Xtilt
sQx = 0.018835 arc deg (1.13
arc minutes) 12 V
Ref sLV = 2.61 mv
Ytilt
sFy = 0.040467 arc deg (2.43 arc
minutes)
Z4M-MAB(JulyAugSept)
sZ4M-MAB
=
101.44 mm
LCP8-10
Z-Transfer Plate(JulyAugSept) sLCP8-10
(ZTP) = 15.27 mm
ME2/2-Transfer
Plate R s ME2/2-TP R = 40.27 mm
ME2/2
Inner ME2/1 Outer R s
ME2/2-ME2/1 R = 112.94 mm
Looking at all the sensors, we can see
them track/react to the average temperature
changes
in the different test running periods. Unfortunately, we did not have a MAB in
June,
so we only have comparable Z4M-MAB and LCP8-10 data for two periods, so
we
dont have tracking proof positive in these cases as for sensors in three
periods.
Also
the ME2/2-ME2/1 R connection was changed in September.Disregarding any
of the drifts as temperature correlated, sensor
Drift Limits are established by
a
linear fit of June-July-Aug-Sept Data plotted in real time across 8.51M sec and
are
given
in Table RR below:
Table RR -Long Term Drift
Limits:
Xtilt
Qx = -0.21259 arc deg 1.1995e-08
arc deg/sec (0.170 arc deg in 8.51 M sec)
Ytilt
Fy = 5.3311 arc deg 1.5176e-08 arc deg/sec (0.129 arc deg in 8.51 M sec)
Where
local T3 (shift) as a drift is T3 =
289.78 9.2574e-08 deg K/sec
Z4M-MAB*
= 47.387 mm + 1.9962e-08 mm/sec ( 93 mm in 4.64 M sec)
ZLCP8-TP*
= 8.6263 mm 3.527e-09
mm/sec ( 16.4 mm in 4.64 M sec)
ME2/2-TP
R = 11.036 mm + 8.859e-09 mm/sec ( 75.3 mm in 8.51 M sec)
ME2/1-ME2/2 R = 38.954 mm+ -6.2823e-08 mm/sec ** (251 mm in 4.1 M sec)
*
July-Aug-Sept periods
**
There is thermal motion (wiggle/expansion) of the 3.2 m long free panel connected
to the sensor
+ June-July-Aug periods
All of the
temperature sensors across the ISR test periods can be fit with a variation like
1e-07 deg K/sec and all
of the analog sensors that are fixed to stable objects show a variation
like 1e-08 unit/sec.
A simplest correction for thermal
effects on the analog sensor readouts would be to take
the average of sensor
readings and temperatures in each test period; choose one as a
reference
and see if the
variances are consistent with the sense
and magnitude of temperature differences.
We can try this
procedure for three sensors that have data and invariant conditions in the
three
test periods; Xtilt Qx, Ytilt Fy, ME2/2-TP R.
Table SS gives the results below:
Table SS -Long Term Drift
Limits with Simple Temperature Correction:
Xtilt
Qx = -0.23747 arc deg 9.1024e-10
arc deg/sec (0.0077
arc deg in 8.51 M sec)
Ytilt
Fy = 5.3091 arc deg 5.2444e-09 arc deg/sec (0.0446 arc deg in 8.51 M sec)
ME2/2-TP
R = 11.056 mm -7.2693e-11
mm/sec ( 0.62 mm in 8.51 M sec)
With the resulting Long
term resolutions
Xtilt
sQx = 0.0367 arc deg (2.202 arc
minutes)
Ytilt
sFy = 0.0097 arc deg (0.582 arc
minutes)
ME2/2-Transfer
Plate R s
ME2/2-TP R = 4.70 mm
Even with the simple temperature
correction, long term dirfts are dramatically reduced
and the effective
long term resolution is improved significantly. In the future, we will calibrate
the temperature
dependence of our sensors through detailed temperature cycling. Then we can
apply a real time
correction according our T sensors which will improve the resolution and
stability even more.
10. Acknowledgements
We especially thank our collaborators of Northeastern
University
(J.
Moromisato, Eb vonGoeler, R. Terry) who built the mechanical parts, calibrated
and filtered the DCOPS
sensors, and worked with us in the
development, setup,
and some execution of the tests.
We also thank S. Hanson,
E. Cowden, J. Lofgrin, R. Martin (Fermilab PPD
Electronics group) and V. Sknar, E.
Orischin (Petersburg Nuclear Physics Institute)
for their development, construction,
testing,DSP programming, etc. of the DCOPS
sensor and electronics for the ISR I4
tests.
We thank the CERN Laboratory
EP/CMS (A. Ball, D. Arevalo, R. Riberio)
for establishing the ISR I4 test area, utilities, supports;
for survey services and
Photogrammetry (C.Lasseur, C.
Humbertclaude, D. Mergelkuhl, R. Goudard,
J.Noel. Joux and crew- CERN Geodesy).
We also thank the Fermilab Geodesy
(J. Greenwood and crew) for
Photogrammetry measurements of our sensor assemblies;
and the Fermilab Technical Division
for CMM measurements of brackets and the transfer
plate.