Available on CMS information server CMS NOTE 1997/0xx
November 20, 1997
Prototype ATLAS muon system transparent Si optical
sensors have undergone further testing at Fermilab. A detailed
study of the long term behavior of a multi sensor straight line
monitor has been made.
* Student from Purdue University, West Lafayette, IN 47907, USA
I. Introduction
Our initial studies of the ATLAS Muon system transparent
amorphous Si sensors [1] during the summer of 1996 were restricted
to measurements of the short term performance, resolution, refraction,
and absorption effects. More recent studies, in the summer and
fall of 1997, have made detailed measurements of the sensor performance,
stability, and resolution over extended periods of time (>
9 days). We also studied thermal effects on the simple ( non thermocooled)
laser diode module sources.
In the CMS Endcap muon position monitoring system
(EMPMS), critical phi reference planes will be transferred to
the Endcaps at the CMS outer radial boundaries from the tracking/barrel
muon systems. See references [2, 3]. This is to be achieved by
the LINK laser beam tracker coordinate system transfer to the
Barrel end MABs, Endcap Rasnik lines across the Barrel/Endcaps
defined by these MABs in Endcap Rasnik reference plates, and
by local Endcap straight line monitors (SLMs). We plan to use
MPA sensors in these straight line monitors (SLMs) to define radially
opposite link points. An Endcap Rasnik reference plate on the
Z+ end MAB is shown in Reference [3]. Across each of the cathode
strip chamber planes of the Endcap system, there will be laser
diode beamlines defined by the radially opposite link points.
These are the layer reference phi plane SLMs. There are three
to six of these lines across a detector plane. One linkpoint MPA
sensor on a Link transfer plate with precision inclinometers defines
the laser diode source position while the opposite radial end
linkpoint/linkplate defines the position of a MPA sensor used
to monitor the optical beam motions (i. e., defines the instantaneous
optical straight line within short term sensor resolution).
Two MPA detectors mounted on each phi plane reference
cathode strip chamber (CSC) will define their chamber positions
(Rphi, z) relative to these crossed laser beams. This requires
eight chamber position sensors and an endpoint definition sensor
along each local phi straight line monitor (SLM) for a total of
eight independent measurements. The laser beam must pass through
all the detectors. The resolution requirement for the position
measurements is 50-100 µm. A schematic of the ME1/2, ME1/3
cathode strip chamber layer (inner and outer rings) and phi boundary
straight line monitors is shown in Reference [3].
II. Methodology
The ALMY MPA sensors were mounted on a heavy steel magnet assembly stand in a 'half CSC layer' arrangement. That is, sensors were positioned at distances and orientations mimicking their positions on chambers in one half of a CSC layer of inner and outer ring chambers. Available undisturbed laboratory space precluded a 13.5m setup. The full layer setup is to be implemented in future integrated system tests
in the CERN ISR tunnel. The steel magnet stand
was found to move slightly with the shifts in temperature in the
laboratory (>±2C°) as well as transmit small vibrations
from outside the immediate area.
The COHERENT systems 3 mW, laser diode fixed optics
module (LDM) used in most tests had a nearly constant beam profile
and size (5 mm) over our test length to 13m. The module provided
a stable, but non-Gaussian, beam shape which was measured with
a simple photodiode and pinhole aperture. Beam profiles at Z
= 1.52m and 8.85m are shown in Figures 1, 2. We did not observe
significant diffraction secondary peaks and the short term output
was stable. The module was held in a fixed mount. We did establish
that the source and MPA sensor are adequate at 13m with a temporary
setup.
Measurements of the fixed laser beam position were
made in a manner consistent with the summer of 1996 measurements
[1]. The sensors were first readout with the laser diode off
to make a baseline background measurement and then again with
the laser diode pulsed on to obtain a beam position measurement.
The two measurements were subtracted from each other to produce
a beam profile free from background distributions. The resulting
profile was cut with a .1V threshold (about 2% of the maximum
possible voltage) to eliminate tails outside the main distribution.
The spectrum was then fit with a Gaussian curve, from which the
mean and chisqr were determined (cycle). Datapoints of beam position
spectra on the sensors were not determined from the averaging
of several events (no stochastic averaging improvement) but rather
taken as the mean of the fit Gaussian distribution of single cycles.
Estimations of sensor resolution and performance were done by
taking the standard deviations of large samples of individual
measurements of beam position over time. These represent the worst
case resolutions that can be improved by averaging a group of
repeated measurements.
The laser diode source of the SLM included a filter
so as to fully saturate the beam center strips of the first ALMY
MPA detector (ALMY S2, .40m downstream). Full saturation yielded
a beam peak which overflowed near the center and corresponded
to a maximum ADC value (Vmax) of +5V in both the X and Y strips.
During most of the trials, the LDM was powerful enough to just
about saturate the second sensor in the SLM line as well (ALMY
S3, 2.55m from LDM). There appeared to be no degradation in performance
of either ALMY S2 and S3 due to operation in these conditions.
Furthermore, Gaussian curves placed on distributions of the resulting,
if somewhat attenuated peaks, did not at any time yield results
inconsistent with previous measurements of unattenuated peaks
in sensors below saturation. Any potential damage to the sensor
from operating in saturated conditions was most likely precluded
by the quick OFF/ON cycling in a low duty cycle of the LDM during
position measurements. We will treat the laser and MPA sensor
optical stability issue in detail further on in the paper.
III. Long Term Resolutions
With the ALMY MPA sensors placed on the magnet assembly
stand in their half-CSC layer positions, we took several long
series of repeated single measurements for 9-10 days (approx.
900 single sample measurements at 12-15 minute intervals established
by the computer clock). Optical tube shields were employed during
the run to minimize external fluctuations in the ambient lighting.
However, the system was subject to electrical, thermal, and vibrational
disturbances in a normal simple concrete slab building. Measured
RAW resolutions (including optical beam, thermal, and mechanical
motions) varied from 15-100 µm in the different MPA sensors
along the 6m SLM. For two 9 day trials (RData 27 & 47) and
one 10 day trial (RData 43); the average, RAW position
resolutions are summarized below in Table 1:
The structure of the observed beam position fluctuations in all sensors was very similar in all long term runs. All sensor measurement spectra showed some long term drift (different in the runs) and several characteristic oscillation peaks of varying magnitude increasing with distance from the laser source. These peaks contained a single maximum and minimum for each day that data was acquired (diurnal cycles generally reflecting the temperature variation). Furthermore, each set of X or Y measurement beam position oscillation peaks occurred in phase for all MPA sensors. Although both axes (X horizontal, Y vertical) on all sensors displayed these periodic oscillations in beam position measurement, the two axes (X, Y) differed by a phase constant indicating different mechanical and vibrational contributions in the horizontal and vertical planes. Each plane was also out of phase with direct ambient temperature fluctuations.
Close inspection of the MPA sensor raw data distributions in
the different long term stability runs (Rdata 27, 43, 47) revealed
that the long term drifts in the MPA sensor measured laser beam
positions were reasonably linearly related; the drift was proportional
to the distance between individual sensors and the laser diode
module. In the RData47 run, the long term drifts were much smaller
and the derived drift corrections (linear fits of the raw data
spectra) were somewhat more variable reflecting minor differences
in the diurnal fluctuations. Additionally, individual X, Y beam
position measurements in the different SLM MPA sensors, after
the long term drift correction described above, were typically
related by the same distance relationship. This indicated that
the dominant variation of the beam position measurements was generated
by long and short term variation of the position and pointing
of the laser diode source. In the CMS Endcap SLMs, the final MPA
sensor beam position measurement (at the opposite linkpoint)
is combined with the laser diode module source to define the instantaneous
reference line.
To evaluate the effective long term resolution of the MPA sensors
in such a SLM, the final sensor 6 in the test SLM setup was taken
as the endpoint reference sensor. Preceding MPA sensors (2,3,4,5)
had their position measurements (point by point) corrected by
the fractional ratio of distance to the laser diode times the
Sensor 6 fluctuation of the beam position measurement. For example,
a MPA sensor 4 (S4) measurement of position was modified in the
following manner: (S4)corrected=(S4)
raw - [(S4 distance to LDM)/(S6 distance to
LDM)]*(S6) raw
Figures 3,4 illustrate the horizontal (X) and vertical (Y) MPA sensor 2, 3, 3, 5, 6 beam position measurement oscillations (after correction for long term drifts) compared to the T3 measured temperature fluctuations.
Changes in the temperature of the magnet stands, sensor mounting plates, Z coordinate transfer structures, and sensor casings were recorded. Limited correlation between shifts in the MPA beam position and the variation in temperature was observed. The beam position measurements in the long term runs show variable and generally weak statistical correlation to simultaneous measurements of temperature variation in the MPA sensor environment. A good correlation (narrow linear distribution ) is shown in Figure 5 while a poor correlation (broad distribution only showing a weak correlation of the major axis) is shown in Figure 6. Good correlations are observed mostly in the distant sensors MPA 5, 6 in both the horizontal and vertical planes. Our conclusion was that a direct beam position measurement correction using temperature sensor data would not improve the results. Test trials on RData 27 proved this to be the case.
Figure 7(a), (b) shows the independence of position
measurements of the closest MPA sensor 2 to the laser diode (separation
of .40 m) with respect to the temperature on the sensor's casing.
This indicates that substantial variations in position measurements
are dominated by laser beam motion and external mechanical effects.
After the long term drift and preceding short term
oscillation corrections, all the MPA sensors consistently exhibited
resolutions between 10-25 m over the extended trials. The corrected
results of the two 9 day trials (RData 27 & 47) and one 10
day trial (RData 43) are shown below in Table 2.
The averaged data sets (in all extended trials) were very similar,
with all resolutions within 10% of the calculated averages shown
in Table 2. The vertical plane contains more mechanical, thermal,
and vibrational effects of the support beam that are not corrected
by beam position tracking (using the endpoint reference sensor
6 correction). These resolutions do not include any geometrical
X, Y refraction or optical gain variation corrections. Here, the
MPA sensors are treated as simple, uniform detectors.
IV. Laser Diode Module Warm-up Effects and MPA Short Term Resolution
Short term resolution was determined by making a series of 100
complete measurements (cycles) and finding the standard deviations
of the resulting set of measurements. The readout cycles were
spaced without any time delay and typically took about 20 seconds
to record a single measurement, analyze, and reset the system.
The resulting set of beam position measurements showed an initial
smooth change in both the x and y beam axis positions. Figures
8 and 9 illustrate these drifts.
Investigations with a pinhole photodiode detector (pinhole aperture << LDM beam diameter) at the end of the SLM line revealed that the intensity of the LDM did not change through the measurements. This implies that the initial shifts in the sensor positions are caused by small shifts in the LDM junction-optics (as the LDM is warmed up) rather than some initial warm-up (photocurrent) effect in the MPA sensors. The MPA have been powered over a period of months. This warm-up effect of the LDM can also be seen as a rapid drift in the first 20-40 MPA X, Y beam position measurements in the long term runs; after which beam position measurement spectra drifts taper off and the periodic diurnal oscillation structure dominates.
The corrected short term resolutions on the steel magnet stand
were found to be slightly higher than the resolutions recorded
on the more stable granite table during the summer of 1996; but
no sensor was found to show substantial deviations from the previous
measurements. The resultant MPA sensor resolutions for two separate
100 short cycle runs is presented in average in Table 3. All
MPA sensors (X, Y) yielded resolutions of 13 m without geometric
refraction or optical gain variation corrections of the individual
MPA sensors.
V. Optical Long term stability of the Laser Diode Modules and
the MPA sensors
In our mode of operation, we generally pulse the laser on only
briefly (20 sec) in a low duty cycle (10-15 minute for long runs).
While there is a warm-up effect to thermal equilibrium, our measurement
experience over a several month period indicates that the output
of the simple Laser diode module is very stable. We conclude this
from the stability of the signal level in the MPA sensors throughout
and across the different long term runs which began in June and
ended in October. The manufacturer suggests that the source output
should be stable for several thousands of hours of operation.
In each of the long term stability runs, the peak signal amplitude
spectra in all the non-saturated MPA sensors show the same general
fluctuation structure. Comparing this structure with the diurnal
temperature fluctuation (see Figure 10 comparing MPA sensors 5,
6 and the S3 temperature sensor) suggests some correlation. However,
as for the beam position measurement spectra, the correlation
varies greatly. A few amplitude spectra correlate very well.
Most of the signal amplitudes show variable and sometimes weak
statistical correlation to simultaneous measurements of
temperature variation in the sensors environment.
Looking at the variations in the MPA measured beam positions X, Y and the variations in the MPA signal amplitudes, we do not observe significant correlation except for MPA sensors 5, 6 in the
vertical plane (Y). Like for the beam position measurement distributions, it is a statistical correlation
of a broad distribution (band) of values and the quality of the correlation varies significantly. There does
not seem to be any correlation in the horizontal plane (X).
VI. Conclusions
A. MPA Sensor Performance
Our evaluations suggest that our SLM configuration of MPA sensors, laser diode source, and fitting methodology provide adequate beam profiles, sensor resolution, Gaussian fits, and stability to make consistent local sensor position measurements well within a 25 m error over periods of 10 days. Errors in position measurements are dominated by external mechanical/thermal factors and the 'wiggling' of the laser diode optics and not by the MPA sensors. Averaging a group of repeated measurements would improve
the resolution.
As noted in our initial evaluation of the sensors [1], there are still several problems with the sensors including optical reflections and electrical termination problems for extremely short and long cable connections, hot channels in the readout (single channels spontaneously reading +5V; these were corrected by software control), variations in the thickness of the glass substrate which affect downstream beam profiles and positions, variation in the aSi layer which give the sensors a non-uniform optical response and
transmission affecting the given sensor resolution and those downstream.
B. MPA Sensor Stability
The six ALMY MPA sensors were extensively tested at Fermilab
over approximately 18 months and during that time did not show
any apparent degradation in performance or resolution. In particular,
the sensors operated in the SLM configuration for 3 months.
Position measurements obtained in that particular span of time
were consistent with each other as well as with results obtained
during initial evaluations in 1996. Possible degradation of
the sensors' performance was perhaps mitigated by the short, pulsed
LDM exposure (< 20sec) and the low duty cycle (10-15 minute
cycle for the 9-10 day trials).
References
(1) CMS Endcap Muon System Evaluation of Max Planck Institute
Transparent Amorphous Silicon- X, Y Strip Readout Optical Beam
Position Sensors; David P. Eartly, Robert H. Lee, Adam Bujak,
Denis O. Prokofiev, CMS IN 1997-005.
(2) CMS Muon Alignment - Progress Report, Muon Alignment Group,
CMS TN/96-050
(3) CMS Muon System Technical Design Report