FUVITA - Detector Calibration
1. Overview
All three FUVITA detectors were submitted to the same procedure :
after selection and conditioning of a proper micro-channel plate (MCP)
triplet the MCPs were mounted in the detector and the general
behaviour (background, hotspots, electronic non-linearity,
deadtime, gain, flatfield) was tested.
Specific calibration measurements were then started by mounting
a pinhole mask (10 micron precision pinholes at 2 mm spacing)
directly on the front MCP and illumination by a Hg lamp.
Analysis of the pinhole measurements yielded information
on the spatial resolution as well as on the image distortions of
the detectors.Also, the optimum highvoltage setting for the MCPs
was selected in order to provide the best resolution possible.
Next, the pinhole mask was removed and flatfield measurements were
performed with a Krypton lamp (123,6 nm) in order to study the
overall pulseheight and relative efficiency response. This was done
with high statistics to search for small-scale efficiency variations.
In between, measurements without illumination served to assess the
background behaviour.
In the last step each detector was illuminated locally (~ 1 cm
diameter of the beam spot) by a beam of monochromated light at
several wavelength between 49 nm and 121.6 nm and the quantum
efficiency was determined absolutely. The quantum efficiency over
the whole detector open area is then obtained by combining this
measurement with the flatfield measurement.
Finally the transmission of the Indium filters was determined at
8 discrete wavelengths between 53 nm and 121.6 nm.
2. Results
(a) Spatial Resolution
In order to determine the spatial resolution the pinhole mask measurements
were analyzed using a pinhole searching routine. Each pinhole intensity
response was fitted in x/y by one-dimensional Gaussians and the
square-root of the product of the x/y-fwhm values was defined as the
intrinsic detector resolution at the position of the pinhole.
Fig. 1 shows the distribution of all FM1
pinhole fwhm values (covering the whole field-of-view). The low-resolution
tail is clearly correlated with pinholes close to the detector edge.
The average resolutions (over the full area) are 66, 75 and 79 microns for
FM1, FM2 and PFM respectively. Here we have corrected for the MCP channel
spacing of 15 microns, because in the most cases two or even three channels
will be illuminated by a 10 micron diameter pinhole.
Since 68 micron correspond to 10 arcsec, the specific goal of 10 arcsec
angular resolution has been essentially achieved.
(b) Image distortions
Analysis of and correction for the image distortions are also based on the
pinhole measurements. Fig. 2 shows the raw, uncorrected
image for FM1, revealing the fact, that the most significant distortions
mainly appear close to the detector edge. This is also demonstrated
by the curves labelled "before Corr." in Fig. 3,
which displays the locally averaged distortions as a function of the
detector radius.
In order to correct for the distortions we proceeded as follows :
- The pinhole centers were compared to an ideal grid with 2 mm spacing
and the deviation vectors for each pinhole were determined.
- The discrete deviation map obtained in this way was then fitted
with a 2-dimensional spline function.
- Thus, arbitrary photon locations can be corrected, either photon-
by-photon (most accurate, but slower) or using pixel maps with one common
correction value for all photons located within a certain pixel.
The latter can be done "globally" over the whole area with 512x512 pixels
or "locally" within a zoom area with pixels as small as 1x1 microns.
The curves labelled "after corr." in Fig. 3 show
the result of a global correction applied to the original pinhole images
and re-analyzing the center deviations after the correction. The residual
distortios have been suppressed to a level which is below the 10 micron
positioning accurancy of the pinhole mask itself.
(c) Detector efficiency
The quantum efficiency (qe) of the bare front MCP (averaged over a
central spot of about 5 mm) was measured at various fixed wavelengths
in the range of 49 nm to 120 nm, using a hollow cathode gas discharge
lamp and a monochromator. Normalization of the incident beam was obtained
from current measurements with a photo-emissive Al-diode. Fig. 4
shows the results. For the qe-measurements the Indium filter to be mounted in space,
was replaced by a wire mesh, which was biased with respect to the front MCP potential.
With a negative bias of about -20 V the results were compatible with corresponding
measurements in the literature. With a positive bias the qe is about 40 % lower,
roughly consistent with the idea, that electrons create on the MCP surface are pulled
away from the MCP and do not contribute to the qe. In this configuration,
photoelectrons created in the Indium filter (around a wavelength of 100 nm on a level
comparable to the number of photons passing the filter and being detected !) cannot reach
the MCP and produce "halo" events.
All 3 detectors measured, show roughly the same efficiency. The absolute
normalization uncertainity is estimated to be about 15 %.
The relative efficiency over the whole MCP was measured in high-statistics
flatfield measurements, which can then be used to correct from small efficiency
variations ( with amplitude below 10 %) on length scales less than 3 mm.
(d) Indium filter transmission
The transmissivity of the Indium filters was measured at 5 discrete wavelength between
49 nm and 120 nm by comparing the intensity of a beam of the selected emission line with
and without filter by means of a MCP detector. Fig. 5 shows the results
obtained for 3 flight filters. The filter manufacturer quoted the thickness of the filters as 174 +/- 4 nm.
With this thickness a theoretical prediction gives 50 % higher values for the transmission as measured.
Part of the difference is certainly due to the fact, that the density of a thin film is known to be about
15 % higher than that of a bulk material used for the prediction. The left-over discrepancy may be
explained by the notorious uncertainity of the optical parameters especially in this wavelength
domain, combined with some "ageing", i.e. exposure to small amounts of water and oxigen between
filter production and measurements.