Following S2 the ions are further accelerated by a potential PA2 (-0.4 to -3 kV) applied between S2 and C2, to a minimum E/Q of approximately 1200eV/e over the remainder of their path. The second toroidal electrostatic analyzer (EA2), refocuses the ions in alpha at the MCP detector surface. In addition the combination of geometric parameters for EA1 and EA2 are chosen to produce energy focusing at the MCP. A general discussion of the theory and laboratory verification of the double focusing properties of toroidal ion optics systems is given in Ghielmetti and Young [1987] and Young et al. [1987]. The theory is extended to the poloidal geometry (i.e. the geometry of the TIMAS) in Ghielmetti and Shelley [1990]. Laboratory verification of TIMAS-like optics is found in Bratschi et al. [1993].
The combination of PA1 and EA1 voltages determines the mean ion energy
per charge (E/Q) and together with the S2 slit width determine the
energy per charge passband (delta E/Q). The internal energy
passband (delta Eie
Ions exiting from slit S3 following EA2 are momentum dispersed radially
outward in a toroidal magnetic field that is produced by 32 wedge
shaped magnets arranged in a spoke-like pattern. Each magnet subtends
an angle of approximately 3 degrees, thus obstructing somewhat more
than 25% of each of the 11.25 degree sectors. The samarium-cobalt
magnet material (RECOMA with BHmax >= 29 MGOe) was selected
for high inductance at the working point. Magnets are precisely
matched in total flux (<< 1% variation) to provide nearly equal magnetic
induction of ~ 0.26 Tesla in all 32 gaps. In addition the magnets
are arranged to minimize any residual dipole and higher order moments
of the stray field.
Ions that strike the polefaces of the magnets can be scattered into
the detector as either degraded ions or neutrals. This undesirable
background is suppressed by providing each magnet face with an anti-scattering
surface consisting of a set of parallel vertical slats. The slats,
separated by 1 mm, are 240 microns high and are less than 10 microns
wide at their peaks. They are produced by wire electron discharge
machining a thin (~ 280 micron) layer of copper which was chemically
deposited directly onto the magnets. The slatted surface produced
in this manner has proven highly effective in reducing scattering
background to less than 5 x 10^-4 of the ion flux impinging
on the magnet faces [Bratschi et al., 1993].
The image on the MCP detector for a given EA setting ideally consists
of a series of nearly concentric rings, each ring representing a different
M/Q ion. As the EA settings are changed to transmit higher or lower
E/Q ions, the radii of the family of rings decreases or increases
accordingly. Acceleration to a minimum E/Q of approximately 1200eV/e
within EA2 ensures that the radius of the lightest ion (H
Before leaving the discussion of the ion optics, a few specific details
are worth pointing out. Firstly, as discussed by Ghielmetti and Shelley
[1990], the electrostatic field in a poloidal toroid varies with position
along the principal path, resulting in non-circular paths and significant
aberrations that degrade the image. Furthermore, the ion accelerations
at C1 and C2, if not compensated for, result in beam refraction that
alters the effective focal lengths of the respective analyzers EA1
and EA2. These effects are mitigated by making small alterations
to the toroidal deflection plates and to the entrance and exit slit
regions. Some of these features, such as the off-set between the
starting angle of the inner and outer deflection plates of EA1 near
S1 and the serrations on the EA2 deflection plates, can be seen in
the details of
Figure 1. These modifications minimize imaging errors
while maintaining first order angle-energy focusing at the MCP. The
final calculated full line width (including 3-D aberrations) is approximately
3.3 mm at the detector surface.
A plan view of the grid and detector geometry is shown in
Figure 5.
Note the front side of the MCP detector faces downward in
Figure 1.
The 90% transmission grid assembly, placed approximately 1.5 mm in
front of the MCP
(see Figure 1),
not only terminates the PA2 equipotential
but also acts as an important collimator in the azimuthal direction.
This collimation is achieved by leaving unetched spoke regions approximately
3.3 degrees wide centered over each magnet and aligned with a second
set of similar spokes (without grids) placed approximately 4.5 mm
from the MCP. These spokes, combined with other stops at S3 and S2/C2,
are designed to prevent ions from crossing over from one 11.25 degree
sector to adjacent sectors and in effect limit the azimuth (beta)
acceptance of ions from any given position at S1 to ~ +/ 2.5 degrees
in the absence of preacceleration. The PA1 and PA2
preaccelerations lead to refraction at S1 and S2 respectively and
expand the external acceptance in beta to as much as +/-
10 degrees at the very lowest energy. As a result there is some
overlap in beta between sectors below about 1 keV/e.
The input side of the MCP is maintained at a fixed -3.5 kV to ensure
a high quantum efficiency for all ions and to repell any secondary
electrons produced on surfaces other than the MCP. This results in
a potential difference of between 0.5 and 3.5 kV across the 1.5 mm
gap between the grid and the MCP. While accelerating the ions into
the MCP, this potential accelerates electrons produced on the interstitial
surfaces between micropores away from the MCP. This eliminates image
spreading that would otherwise result if these electrons were repelled
back to the MCP and were detected at a position other than where the
ion struck the surface. However it has the undesirable effect of
reducing the overall MCP efficiency by approximately 50%.
A two dimensional wedge and strip anode [Anger, 1966; Martin et al.,
1981; Schwarz and Lapington, 1985] is placed approximately 3 mm behind
the MCP. Its bias voltage relative to the MCP can be adjusted for
optimum performance. The wedge pattern consists of 28 discrete steps
that are oriented along concentric circles to uniquely identify each
of the 28 azimuthal sectors, independent of radial distance. The
strip pattern, interleaved with the wedge pattern, decreases nonlinearly
with radius so that the relative signal S(R) = S(Ro) +
log(Ro/R), where Ro is the inner radius; thus
providing approximately constant delta R/R resolution. A higher
positional resolution is required at smaller radii because both the
mass line width and dispersion decrease with decreasing radius. The
anode area not occupied by either the strip (A) or the wedge (B) pattern
is filled in by a third electrode (C). Following preamplification
of signals from each electrode, three signals are generated; A, B,
and A+B+C
(see Figure 6). These signals are
shaped using a six pole
critically damped LCR, RC, LCR, RC network which produces a pulse
that peaks at 650 ns from the start of the pulse and recovers to 0.5%
of this peak in 3.5 us. They are fed to two ratioing analog
to digital converters (ADC) which generate an 8-bit A/(A+B+C) radial
position output code and a 6-bit B/(A+B+C) sector, or azimuth, output
code. In order to accommodate higher event rates, the positive signal
from the MCP contact anodes (i.e. the positive terminals of the MCPs)
is shaped using a 4 pole RC network with an RC time constant of 32
ns. The pulse peaks at 40 ns from the start of the pulse and crosses
zero at 100 ns from the start. The default discriminator level for
this pulse is set at 25% of nominal amplitude, thus it will respond
to another pulse about 125 ns after the start of the first pulse.
The precise response time depends on the amplitude of the first pulse
and the actual setting of the programmable discriminator, but will
be within 100 to 150 ns for the expected range of outputs of the MCP.
Each fast pulse blocks the imaging system for 3.5 us to avoid
pile-up. In addition, a singles counter, driven by a one shot whose
pulse width is set to 140 ns is used to normalize the imaged events
at high rates. In an alternate (calibration) mode the sum signal
(A+B+C) is fed to an ADC with a 4-bit output code proportional to
the MCP gain. Gain measurements will be performed periodically to
determine whether the threshold discriminators or the MCP bias need
to be adjusted.
Ideally, all ions with the same M/Q and E/Q are detected at the same
radial position on the MCP. However, noncircularities in the deflection
plates, misalignments of the analyzers or the detector relative to
the optics axis, or differences in magnetic induction among magnets
could result in a slight variation of radial position with azimuth.
Furthermore, some electronic distortion results from cross talk between
electrodes A, B, and C. The combined optical and electronic distortions
are removed by transforming the 6x8 bit code to a 5-bit (actually
0-27) detector sector identification and a 6-bit psudo-radial code
that is dependent only on mass and energy. This is accomplished using
ROM look up tables (LUT) as shown in
Figure 7, which is a block diagram
of the instrument control and data processing system showing the flow
of data from the detector system through the various stages of processing.
These LUTs were defined on the basis of the detailed calibration
of the complete optical system and will be installed into the instrument
during the refurbishment period shortly before launch.
A graphic representation of the ROM LUTs themselves is shown in
Figure 8,
combining the radial (vertical bands) and azimuthal (horizontal
bands) into a single two-dimensional grid pattern. Each radial mass
step typically alternates in width between two and three radial detector
bins, whereas the sector number spans either one or two azimuthal
bins. Only 60 of the 64 available mass steps are shown here, running
from MS = 2 on the far left to MS = 61 on the far right. Mass steps
0, 1, 62, and 63 have been assigned to the MCP edges and beyond,
and are not expected to have any signal. It should be noted that
the rightmost mass steps in
Figure 8 are near the inner edge of the
MCP, where the spatial density of detector radial bins is the highest.
This accounts in large part for the greater undulation of the constant-rigidity
loci on the right. As far as the field-of-view sectors (numbered
0 through 27) are concerned, it is worth noting that sectors 13 and
14 (midway up) are physically separated by a 22.5 degrees blind sector.
The flight data products actually use a reversed sector numbering,
running from top to bottom in Fiugre 8, in order to provide right-handed
external coordinates. Data accumulation beyond this point will be
discussed in a later section.
To facilitate ground testing at atmospheric pressure, when the MCP
is non-functional, and to serve as a standard end-to-end data systems
test, there are two independent stimulus points incorporated into
the detector anode. When enabled, a fixed pattern of pulses are injected
into these stimulus points, thus producing a unique image pattern.
This stimulus calibration allows for testing of the gains of the
analog components as well as the digital processing and telemetry
formatting.
An optical regulator can be described as a light sensitive high voltage
diode, an LED, and a light transparent medium between the two. The
principle of operation is control of the leakage current through the
high voltage diode by the amount of LED light incident on it. The
device essentially behaves as a current controlled current source.
As an added advantage, the LED and high voltage diode have high voltage
isolation, making a four-leaded device which can control high voltages
from an isolated potential. Arranging the regulators in a series-shunt
configuration results in equivalent positive and negative going slew
rates. This is particularly important for the two pre-acceleration
potentials, PA1 and PA2, which do not follow a monotonic stepping
pattern. In addition, this arrangement, which requires no passive
pull-down to achieve the negative going slew rates, permits the use
of very high impedance high voltage sensing resistors.
Running multiple regulators from a single DC high voltage supply allows
for independent regulated outputs with the overhead of only one high
voltage power supply. This method is particularly efficient in the
case of capacitive loads since the optical regulators are using significant
power only while slewing. Furthermore, since the loss of any one
of the analyzer voltages would represent an instrument failure, the
use of fewer supplies actually increases reliability.
The series-shunt configuration is achieved by placing the regulator's
LEDs inside of the feedback loop and closing the loop by sensing the
forward current through these devices. This arrangement inherently
protects the series and shunt regulators from both ever being "on"
simultaneously. In addition, the current sensing scheme strongly
reduces any crossover distortion when switching between the series
and shunt devices. Since the error amplifier is contained within
the compensated feedback loop, it is able to swing through this region
in an open loop manner.
The initial calibration consisted of measuring a combination of single
parameter cuts through the energy-angle-mass response functions over
a wide range of parameter space and more detailed multiple parameter
matrix measurements over more limited ranges of parameter space.
Figure 9
shows a typical energy passband for the TIMAS instrument.
It was actually measured using a large area, uniform, monoenergetic
ion beam incident normal to S1 and varying the EA1 plate potentials
while holding the EA2 plate potentials constant of their nominal settings.
Since the EA2 energy passband is much wider than that of EA1, it
does not contribute to the total system passband limits for normal
incidence ions. The mean value of the bandwidths in the 28 sectors
was 8.06% with a maximum variation of +/- 0.35%. This is completely
consistent with the optics design parameters. However; the RMS deviation
of the passband centers was 0.75%. This is interpreted as variations
in the plate spacings resulting from eccentricities and small distortions
in the optics. The small error in defining the energy of the ion
<<10% of the instantaneous bandwidth) is negligible in terms of defining
the velocity distributions. However, these deviations, taken together
with uncorrelated deviations of comparable magnitude in the EA2 passband,
result in some fluctuations in the overall transmission as a function
of sector. These fluctuations, combined with other instrumental effects,
are corrected for in the detailed, end-to-end sensitivity measurements
as a function of sector (cf. Figure 8).
Figure 10
displays characteristic elevation angle response curves
for a single sector. The solid symbols were measured with a monoenergetic
ion beam and display an almost ideal flat response with sharp cut-offs
at both the upper and lower limits as expected for angle focusing
optics
(See Figure 11).
The mean value of the cone angle, in this
case measured from the horizontal, is approximately 16.9 degrees,
within measurement error of the theoretical 17 degrees. Likewise,
the measured 9.9 degrees FWHM is consistent with the design passband.
The open symbols in
Figure 10 were measured with a beam having an
energy spread (~13%) significantly broader than the instrument energy
passband (~8%,
see Figure 9).
As seen from
Figure 11, the ray tracing
predictions show both a low energy, low angle and a high energy, high
angle cut off. These result from extreme rays striking respectively
the inner and outer of EA2 plates. This should lead to a softening
of both the upper and lower edges of the elevation angle response.
By contrast the low angle cut-off is nearly identical (within measurement
error) to that for a monoenergetic beam and the high angle cut-off
is somewhat softer than predicted. This effect is a direct result
of the small misalignments of the analyzer plates discussed above.
In this particular sector the deflection in EA2 is insufficiently
strong relative to EA1, so that too many rays are cut off by the outer
EA2 plate. The primary effect is a small variation in total instrument
sensitivity as a function of sector. The effective changes in look
direction are negligible compared to the overall angular resolution.
Finally, a short sample of the azimuthal response of the TIMAS is
shown in
Figure 12.
The look direction angle is defined in terms
of the incident ion velocity vector relative to the spacecraft spin
axis. Thus sector 13, which overlaps the 180 degrees look direction,
views along the spin axis. Note that the response drops to zero at
angles greater than about 181 degrees. This area corresponds to
one of the "dead zones" that is masked off by the grid frame.
By contract, the transmission does not go to background levels between
sectors as it should if the internal collimation elements at S2/C2
plus the grid assembly between the magnets and the MCP performed exactly
as designed. The details of the finite transmission between sectors
is best seen near 160 degrees. As the beam direction approached
a magnet, the transmission first drops nearly to background as expected,
but then increases to about 15%-20% over a region of a few tenths
of a degree. The transmission again drops to near background directly
over the magnet and is symmetrical about the magnet meridian. The
net result is that approximately one percent of the ions reaching
any given sector are leaked from the adjacent sectors in this calibration
set-up.
The azimuthal transmission band at each sector is approximately 6.5
degrees FWHM resulting in an azimuthal transmission of about 60%.
This is somewhat less than would be expected based only on the widths
of the magnets and grid spokes (~70%); however, a parallel beam incident
on S1 does not remain parallel, but crosses over approximately mid-way
through EA2. Thus the ~3.5 degrees wide stops at S2 will first cut
out rays that are closer to the meridian of the collimating slits
than the local meridional ray while the ~3.5 degrees stops at S3
and the 3.3 degrees stops at the MCP grid spokes will first cut out
rays that are toward the center of the sector from the local meridional
ray. The net result is that when the local meridional ray is coincident
with the meridian of the collimating stops at S2 and S3, parallel
rays have been cut off. Thus the full width, not the full width at
half maximum, is about 7.8 degrees as observed in Figure 12.
It should be noted that the calibration data presented above did not
involve any preacceleration at PA1 or PA2. Preacceleration has several
effects. First, acceleration preceding S1 results in an increased
azimuthal acceptance per sector due to refraction at S1. It simultaneously
increases the area factor for a parallel beam for the same reason
(i.e. the incident rays are bent toward the local meridian). Finally,
the energy bandpass relative to the external ion energy increases
in direct proportion to the increased energy. Preacceleration at
S2 has much less effect on the geometric factor, but since it results
in a smaller spread in angles within EA2, fewer rays are cut off by
the analyzer plates and by the stops at S3. In particular it suppresses
the cross-over between sectors seen in
Figure 12.
3.2 Detector System
The active area of the MCP detector
(see Figure 5) extends from approximately
26 mm inner radius to 71 mm outer radius and is divided into two independent
arc sectors of 157.5 degrees each. Each sector is made up of two
quadrants which have different micropore bias angle orientations so
that the incident ions never enter parallel (within ~ 7 degrees)
to the micropore axis. This minimizes the variation of sensitivity
over the surface. The two inactive 22.5 degrees sectors of the optics
are used to transfer high voltage to the inner electrodes and to provide
mechanical support across the various apertures. This gap also facilitates
the independent operation of the two MCP halves. The symmetry line
between the two MCP halves is rotated by one and a quarter sectors
(i.e. 1.25 x 11.25 degrees = 14.06 degrees) with respect to the
spacecraft spin axis. This offset ensures that all solid angles in
the field of view, including those near the spin axis, are covered
at least once per spin. It furthermore provides for a more uniform
coverage of view angles near the spin plane because the look directions
shadowed by the magnets of one detector half fall into the field of
view centers of the other detector half approximately one half spin
later. The above configuration results in the fraction of the solid
angle within 17 degrees +/- 5 degrees of either spin axis
direction (~5% of 4 pi) being sampled only once per spin, while
the major part (~93%) is sampled twice per spin. The remaining 2%
of the 4 pi solid angle, within 12 degrees of either spin axis
direction, is never sampled due to the conical shape of the field
of view.
3.3 Analyzer High Voltage System
As described in the ion optics section, the TIMAS analyzer system
consists of tandem toroidal electrostatic analyzers, each floating
at a controllable acceleration voltage of up to -3 kV
(see Figure 6).
The two analyzers, EA1 and EA2, must be matched in energy transmission
to a few tenths of one percent in order to achieve the desired instrument
response characteristics. Thus the two preacceleration voltages,
(PA1 and PA2), and the four analyzer plate voltages ( EA1 and EA2)
must accurately track each other to ~0.1%. Furthermore, these analyzers
represent relatively large capacitive loads (several hundred pf).
Since the analyzer voltages are stepped rather than swept, they must
settle to within a few tenths of a percent of their prescribed values
within a few ms before data acquisition starts. In order to meet
these requirements the TIMAS high voltage power system uses optically
controlled linear regulators (HV601, manufactured by AMPTEK, Inc.)
to supply regulated voltages to multiple loads from a common high
voltage converter.
3.4 Calibration
Initial calibration of the TIMAS instrument, using the University
of Bern ion beam test and calibration facility [Ghielmetti et al.,
1983], verified most of the design characteristics. However, system
noise coupled through the MCP reference high voltage supply resulted
in significantly reduced imaging resolution and precluded the verification
of the detailed mass peak shapes and thus the mass resolution. The
noise coupling has subsequently been eliminated; however, a recalibration
will not be possible until the instrument recalibration and refurbishment
which is scheduled to occur after spacecraft integration and test.
Last modified February 1996 by Bill Peterson bill.peterson@lasp.colorado.edu