Space Sciences Laboratory  University of            note        pages  4    
     California at Berkeley ASTROPHYSICS NOTE        number  478       plus      
                                                         v1.0       appendices   
                                                                   and figures   
keywords GEM instrument; setup; site preparation;    author Giovanni De Amici   
local support                                                                   
           title GEM setup requirements              date july 1995             


The physical requirements for the site where the GEM instrument will be installed are presented, together with suggestions on how to meet them. Other needs (power, shelter, etc.. ) are also discussed.

The GEM project

The Galactic Emission Measurement (GEM) project is a multi-year effort at precisely measuring the absolute intensity of the sky at several frequencies (ranging from 408 MHz to 10 GHz, with the option of extending upward to 15 GHz). The set of calibrated maps will allow a redetermination of the value and spatial distribution of the spectral index of galactic radiation as a function of frequency.

The low end of the frequency range is set by limits in the size of the antennas that can be transported to the observing sites (because of the long observing time, a dedicated antenna is required), while the high end is determined by the rising atmospheric foreground signal, which quickly makes Galactic emission undetectable from ground based receivers.

Description of the hardware

The GEM instrument consists of a small parabolic antenna (5.5 m diameter), which has been modified to be fully transportable, its support and control hardware, a set of feed antennas and microwave receivers (each one optimized for a specific frequency).

The advantage of a parabolic antenna is that it can be used as primary mirror for all the frequencies in the project, provided the surface is within the tolerance limit; only the feed antenna and the receiver need to be changed when the observing frequency is changed. We purchased a 5.5 meter f/0.32 parabolic reflector antenna and built and field-tested an altzimuth mount (Figure NN) which is capable of pointing the antenna up to 42deg. from the zenith and continuously rotating the antenna 360deg. in azimuth. The data acquisition and pointing electronics is mounted on the rotating platform; only digital signals and power pass through a slip ring.

The reflector disassembles into small parts, for ease of transportation, making the system suitable for shipment to remote observing sites. The entire system fits inside a standard 20-foot shipping container. Because of the relatively small diameter of the reflector, we have added a set of ground screens around the perimeter of the dish; these screens are made of lightweight aluminum sheet, and extend the diameter of the reflector to about 10 m. The extension panels undercut the reflector opening angle and do not contribute to the definition of the main beam of the antenna; however, tests have showed that these extensions effectively reduce the contribution from the ground around the observing site, allowing us to almost fully illuminate the main reflector and maximize the angular resolution attainable from it.

The data are processed and digitized on-board the GEM instrument, and provided to the data-acquisition system as a serial stream (RS232 interface) of 16-bit words. The first word is a synchronization character, the last word is the least significative bit in the time stamp. The present configuration of a data frame consists of 21 such words at 300 baud.

Observing strategy

No ground-based instrument can observe the whole sky from just one location on Earth; the measurements must be made from several sites. Experience and practical considerations suggest limiting the zenith angle of the receiver's beam, to keep the contribution of the ground surrounding the observation site to a manageable level. To minimize the corrections to the data, it is however desirable that the same instrument, or a very close duplicate, be used from each site.

The minimum number of sites/instruments is defined by the local horizon and by the mechanical quality of the dish antennas used for the measurements. If the zenith angle is limited to ~30 degrees, mechanical deformation of the parabolic surface is small to negligible, sidelobe contribution can be controlled via ground shields, and the antenna mount is of relatively simple construction.

Observations are conducted by tilting the antenna to an angle from zenith no greater than 40deg., and setting the instrument in rotation at 1 rpm. The field of view describes a circle in the sky, and the diurnal rotation of Earth brings different parts of the celestial sphere within sight. A single 24 hour observation could, in principle, provide full and redundant coverage of the entire sky visible from any one location. In practise, repeated observations will be needed to achieve a complete sampling (especially at the higher frequencies, which have narrow beamwidth) of the sky with the desired sensitivity, and to correct for instrumental and weather effects.

Observing site

The observing strategy for GEM places unusual requirements on the observing site, and they are summarized here.

horizon: the GEM antenna scans the entire 360deg. of azimuth, comparing the signal from each one of the pixels observed with the signal from all the others. In order to keep systematic uncertainties low, it is important to reduce to a minimum and maintain constant the contribution (through the antenna sidelobes) of thermal emission from the ground. Minimization of the absolute value is achieved by observing at (relatively) small angles from zenith, and by erecting a set of ground screens around the antenna (full details on the construction and assembly of the ground screens are given in appendix A). Minimization of spatial variations is achieved by selecting a site with a low, flat horizon. The top of an isolated mountain, or the middle of a large valley are ideal examples. An accurate survey with an optical transit instrument, to precisely measure the profile of the horizon from the potential site, is important to avoid time-consuming massaging of the data during analysis.

radio-frequency quiet: in order to push the sensitivity of the microwave receivers to the limit of their theoretical capabilities, the GEM uses direct-RFI amplification of the signal and bandpass of up to 10% of the center frequency. This setup allows for a simple architecture in the receiver and good sensitivity, but leaves us susceptible to interfering signals from man-made sources. The narrow beamwidth of the GEM antenna will provide some additional rejection of ground-based interference, but it should not advisable to depend on it for complete suppression of RFI. Before installation, the site should be checked for RFI by running a survey with an antenna (preferably omnidirectional), a spectrum analyzer and a low-noise (comparable to the noise of the GEM receivers) amplifier, or other suitable setup.

wind conditions: while the parabolic antenna and the support structure of GEM is built to easily withstand winds of up to 160 Km/h (when properly assembled), the extension panels used by GEM and their supports are (relatively) lightweight structures. Winds of 40-50 mph have been known to detach the panels from the supports and cause damage. A replacement panel can be easily manufactured and installed, and occasional downtime for repairs should be included in the observing schedule, but sites with a repeating pattern of frequent high winds should be avoided.

high altitude and dryness: even though radioastronomy takes advantage of one of the few clean windows of electromagnetic transparency of the Earth's atmosphere, molecular oxygen and water vapour produce a foreground signal which varies with weather conditions and cannot be eliminated. In order to reduce this variable, it is desirable to chose a site that sits at high altitude, above the bulk of the atmospheric gasses, and in a dry area, to further reduce the amount of water vapour. At the frequencies of interest for GEM (0.4 to 10 GHz), the minimum of atmospheric emission ranges from ~0.8 K to 2 K at sea level and from ~0.7 K to 1 K at 3600 m altitude (Costales et al. 1986). Under adverse (light clouds) conditions the high-frequency limits become 5 and 2 K, stressing the advantages of a dry site at high altitude.

cloud conditions: at the low-frequency end of the intended spectrum the foreground signal from the atmosphere is low and only slowly variable, and the Galactic background signal is so large (between 10 and 300 times as large as the atmospheric one), that almost any weather condition, short of heavy thunder clouds, is suitable for observations. At the high end of the frequency spectrum, however, the Galactic signal is very faint and the atmospheric signal speed and amount of variation can drown out the Galactic signature. A site with extended, predictable spells of dry, clear weather should be preferred.

Supporting structures

The GEM instrument is not self-contained and self-sufficient, and it necessitates several supporting elements that must be provided at the site. A non-exclusive list is provided here:

concrete pad: the pointing of the GEM antenna is determined via optical encoder in the elevation and azimuth axis. The encoders should be checked and calibrated at frequent, regular intervals, but consistency between calibrations relays on the stability of the GEM platform. Given the weight of the GEM antenna, about 2500 Kg, it is recommended that a concrete platform be constructed at the site. The GEM antenna will then be placed on the platform and leveled before being secured to it. The minimum exact size for this structure and other construction details are given in Appendix B.

electrical power: startup of the GEM antenna and control electronics, and related receiver requires approx 3 KW of electrical power at 110 VAC. Normal operation will require much less. This values do not include the power drawn by the computer(s) used for data acquisition. Reliable line power, or stabilized and buffered generators are needed.

shelter for operators and support electronics: permanent housing and protection from rain and bad weather for the control electronics and possibly for the data acquisition devices, and temporary shelter for the operator(s) should be available within sight of the antenna site, and preferably no more than 30 meters away, to avoid long runs of the power and signal cables. An insulated metal hut or trailer is adequate. The hut can also be used as a staging area when switching receivers, and at any other time when work must be performed on the antenna.

access: since daily (or hourly) checks of the proper operation of the instrument must be scheduled, and the operator(s) must be able to reach the hut and the antenna without undue hardship. A hardtop road reaching within a few meters of the site is highly desirable.

data acquisition: hardware and software (using National Instrument's LabView package) have been developed for an Apple MacIntosh platform of the Centris 610 (with mathematical coprocessor) family, and it runs on any higher performance Apple machine. The computer is not considered part of the GEM instrument, and it is not shipped with it. It is expected that either a suitable MacIntosh, or a compatible replacement will be available at the site. If N.I. LabView's routines are not available, new software will have to be written.

electronics and machine shop: an available, well stocked electronics shop is necessary to quickly diagnose and repair the instrument, to avoid prolonged down times and effect small changes in the instrumentation. Similarly a machine shop with at least the most common tools is needed for last minute adjustments.

appropriate safety measures (e.g. a telephone connection, fences, warning signs and lights) should also be considered.


this work draws heavily on the effort of many people, but especially Marc Bensadoun's and Michele Limon's, who contributed to the design and first field installation of GEM and helped define these requirements, and Camilo Tello's, who "wrote the book" on assembling the ground screens.

Appendix A

GEM system;

Ground preparation

Platform construction

Prepare a concrete pad: it will have a flat horizontal surface and 12 threaded rods of diameter 5/8" (1.6 cm) sticking out about 10 cm. It is important that the rods are placed EXACTLY as indicated in figure A1a , so as to match the holes in the base of the GEM pedestal.

Minimum dimensions of the pedestal are 155 cm by 155 cm. I recommend maximum dimensions of 180 cm by 180 cm, compatible with the safe and strong attachment of the threaded rods (i.e. smaller is better, but make it as large as it is needed).

Make sure the pad does not stick out of the ground by more than 30 cm, and it is recommended that the pad be flush with the ground (this will make future tasks easier).

The easier way to secure the threaded rods in place is to assemble a jig as in figure A1b ,, place the jig in place and pour the concrete on top of it.

The weight of the GEM instrument is less than 3000 Kg; since unreinforced concrete has very low tensile strenght, it is important that some kind of metal netting or reinforcement bars be added to the platform before pouring the concrete. Check with a civil engineer for the local codes, given the weight of GEM, and its surface area (10 meter diameter at 40deg. angle) with respect to the wind strenght.

Although the GEM pedestal can be shimmed later, it is important that the platform surface be as close as possible to horizontal; insist on this requirement, because it will simplify later tasks.

Preparing the ground for ground screens

Clear out a space about 7 meter radius centered on the pad; this will be the space enclosed by ground screens. With theodolite, identify and peg 12 places, equally spaced along circle of diameter 6.65 meters (+/-0.1 m), centered on the center of area defined by the threaded rods (i.e. the center of the concrete pad).

Prepare supports for ground screen (see write-up dedicated to this task): remember that all vertical measurements are given for flat terrain and assuming that the base is lying on ground; if the concrete pad is raised, or the ground has a slope, then all values must be changed accordingly.

Prepare ground stake(s) for electronics and drive it in the ground as close as possible to concrete pad; better if the stakes are driven 1.5 meters into the ground.

Appendix B

GEM system;

Ground Screen Construction and Assembly

The construction of the ground screen that we shall describe here under the item procedure is meant to serve more as a useful guide than as a strict way of how to do it. Available material sizes and facilities should dictate rather the most practical approach, and only the specified dimensions for the individual ground screen panels must conform to the ones given in the drawings This ground screen was assembled at the site of the observations, although some preliminary work had to be done in a machine shop. A cross section of the overall set-up can be seen in figure B1 .

i - material :

1"(or smaller) galvanized steel mesh

conduit aluminum pipes (2 sizes)

conduit aluminum T-connector


cotter pins

aereoseal clamps

iron stakes


(iron T-posts)

ii - panel procedure :

The ground screen is made up of 12 trapezoidal panels whose dimensions are given in figure B2 . The trapezoid frame is built using the conduit aluminum pipes. The corners of such frames are to be shaped from conduit pipes bent to the specified angles in figure B3 (we recommend using a pattern layout to make all bends as uniform as possible).

If the dimensions given in figure B3 are used, the frame will be ready to assemble with precisely 2 top bends, 2 bottom bends and 2 sides, since the conduit available to us in Berkeley came in units 10 feet long each. A bending tool, appropriate for bending conduit was also used. Individual conduit sections (straight ones for the sides and bent ones for the corners) were then joined to each other to produce the frame using short conduit pieces, whose diameter were small enough to make them slide tightly into adjacent sections. Cotter pins were used to secure these conduit pieces inside the conduit sections as depicted in figure B4 . Remember, that the top conduit section of each panel has a T-connector for the long conduit support that is going to hold the panel inclined above the ground as shown in figure B5 , and this connector should move freely around the top conduit of the panel.

The layout of the steel mesh (if possible the spacing of the mesh holes should be smaller than the suggested 1 inch size above) depends on how wide the rolls of mesh can be found in the store. Nevertheless, the adopted procedure consisted of running strips of mesh from top to bottom of the panels so as to fill in first the rectangular center portion of the trapezoid shape and to cut in half one additional strip to cover the missing triangular shapes at the wings of each trapezoid (see figure B5). Tiewraps were used to knit together adjacent strips and those in contact with the frame to themselves after wrapping them around the conduit. Some care should be taken, so that there are no long slits of open space between adjacent strips.

iii - assembly :

At this point the ground around the cement pad should already have been cleared. And if the base of the turntable is not leveled with this ground, iron T-posts must be used to raise the bottom of the fence to the level of the turntable base. The layout for these posts is found by using a theodolite to space them evenly along a circle of 6.63m in diameter and centered on the cement pad (also where the vertical alignment of the turntable shaft intercepts the ground). Once in place, all 12 panels should be laid down in a circle behind the posts, so that the bottom corners of adjacent trapezoids meet at the center of a T-post. If no T-posts are needed, figure B5 shows how far (6.4 m) from the center of the cement pad the midpoint of the bottom side of each trapezoid should be located.

Next, if the T-posts are required, aereoseal clamps should be used to secure the bottom corners to them (2 clamps for each trapezoid corner, or 4 altogether for each post). These clamps should not be tighten until the screen has been fully erected and secured. To do this, the long conduit supports are used to raise the top of the panels until neighboring trapezoids are most nearly aligned along their adjacent sides. Then, using a forklift, we secured these sides with 3 aereoseal clamps (one near to the top, one halfways along the adjacent sides and a third one near the bottom). A string was then attached around the common top corner of each pair of panels and stretched all the way to the ground. Extending radially outward along the projected adjacent sides of neighboring panels, a stake should be pounded into the ground to provide a support for tying the strings. The location of these stakes should be around the same radial distance as the ground contact of the long conduit supports.

One issue that is not addressed above is the problem of accessing to the dish: if you use the "T-posts", and make those posts at least 80 cm tall (above ground) then access will be possible, just as it was in Bishop, by bending your back and passing under the lower edge of the ground screen.

But if the posts are any shorter than 80 cm, than instead of bending you back (or knees) you will have to crawl in the dirt: not an option if you are carrying the receiver or the feed. In this case, I recommend that one of the panels described above and in figure B2 be modified as figure B6 .

Simply: slip two additional T joints on both the top and bottom horizontal sections, add two vertical sections between those joints, and then another horizontal section about 2 meter off the ground (since the panel is slanted to 50 deg, that will be about 2.5 meter from the bottom of the panel).

The new sections of conduit and the bottom of the panel define a rectangular space; leave this space free of chicken wire since it will become your door opening. Build, out of lighter conduit, a rectangular frame, slightly larger than the section left open. Cover this rectangle with chicken wire: it will be your door. Make sure it is lightweight, so that it can be moved easily by one person.