The Electronic Multi Beam Radio Astronomy ConcEpt (EMBRACE) demonstrates the design readiness of the phased array technology for the Square Kilometre Array (SKA). There are two stations, one in Nançay, France, and the other one at the Westerbork Synthesis Radio Telescope (WSRT) in the Netherlands.
The front-end consists of the antenna array including a radio frequency transparent radôme and supporting mechanics for the array. The array is organized in tiles of approximately a square meter containing 144 antenna elements in dual polarization. Each tile is independently controlled by means of a local control unit (LCU), which is located in the back-end. All tiles will only listen and act as slave. The back-end contains all remaining electronics required for processing the signals from the tiles including the control subsystem. The back-end part is hosted in a small shelter near the array.
To continue along the path of cost reduction, the design of the antenna element is made of solid but fairly thin sheet of aluminium. The aluminium radiator is fed with a micro strip feed designed using Rogers 4003 material (εr = 3.38) which, although very small, is the most costly item on the antenna and hence there may be further opportunities for cost reduction in the future SKA design. The microwave material is necessary as the loss in the transmission line needs to be optimized from noise figure perspective of the system as a whole.
The antenna element has evolved during the EMBRACE project. For the first dual polarized version two vivaldis were cut, and folded, out of one piece of material. These elements clicked together form a tile. These antennae gave problems during assembly and also the electrical connection wasn't sufficient. In the next iteration EMBRACE went back to a single antenna approach where the antennae are connected by means of an extruded profile that slides in between at the crossing of four antennae. The Vivaldi has multiple features to align the feed board to the element, minimizing the gap between the antenna element and the feed board. The edges of the board are used for the alignment of the board. Two Grad Conn connectors are mounted to contact the hex board. A dowel pin is used to connect the feed board to the Vivaldi with a guaranteed electrical contact. The dowel pins used for connecting the feed board are made of aluminum. Only one dowel pin is used to firmly connect the two. This has the advantage that thermal expansion and mechanical tolerances are less critical on the antenna complete antenna element performance.
The structure to hold the antenna consists of two parts. The first part, called the subframe is integrated with the tile. It holds the centerboard, six hexboards and the antenna array. The second part of the structure is a large frame supporting the 144 tiles and holding the cable gutters for all 288 coaxial cables. It was a deliberate choice to divide the structure into two. This choice was made for assembly and testing reasons. For the SKA this option can be reconsidered. Integration of parts within a mainly on mass production based instrument can have a positive influence on cost reduction due to economy of scale. A third structure was build to protect the antenna. This is the radôme covering the complete front-end. For testing purposes and evaluation within the EMBRACE project a large radôme was preferred. It gave flexibility in the phase of the project where it was needed. Also this structure might be integrated with the two support structures in the future SKA design. The materials used for the large radôme are selected on radio transparency and could be used for the construction of smaller integrated radômes as well.
Throughout the design of EMBRACE mechanical tolerances plays a central role. In the end all tiles must fit and connect to its neighbouring tile. It has multiple connections on approximately 1m contact side, where it makes electrical conductive contact between the antenna elements and hex boards. All these connections should be reliable under mechanical stress and thermal expansion of the used materials. The chosen concept for EMBRACE uses an extruded profile to make the required contact. The profile will clamp the vivaldi but also allows the vivaldi to slide a bit into and out of the profile. This sliding is needed to deal with the production tolerances and the difference in shrinkage due to temperature fluctuations. The extruded profiles aren't only used between elements within one hex board but also between elements at the edge of a hex board within one tile.
The Vivaldi was made out of aluminium 1050A. The shape was cut by laser out of large plate. Several features, like for the feed board alignment, were shape stamped or bended into the plate. For larger quantities like for the SKA the complete Vivaldi could be stamped out of aluminium instead of using laser cutting.
The feed board is made out of Rogers 4003. The feed line and stub are etched and silver plated. One hole is applied to fix the feed board on the vivaldi. The edges of the board are used for the alignment of the board. The dowel pins used for connecting the feed board are made of aluminum. Only one dowel pin is used to firmly connect the two together. This has the advantage that thermal expansion and mechanical tolerances are less critical on the antenna complete antenna element performance.
The connection profile for contacting the Vivaldis is made by extrusion of aluminum 6060. Since the profile is very small several attempts were made together with the manufacturer, MIFA aluminum precision extrusion, Venlo. A few iterations were needed to find workable dimension for both parties, manufacturer and costumer.
The sub frame is built out of two main parts. The top part, a polypropylene laser cut plate assures the right position accuracy and mounting possibilities. The bottom part, a glass filled epoxy tubes based frame, assures the required stiffness. The two parts are aligned with an alignment tool during mounting.
The large frame was made out of glass filled reinforced epoxy beams (EXTREN®). It doesn't interfere with radio signals or add noise due to reflection also it is easy in assembly since it is low in weight. The frame consists of thirteen large beams standing one tile width apart. These beams are connected at three places and are placed on adjustable legs at 70cm above ground. The beams hold the cable gutters and support the tiles.
The radôme should protect the antennae, but should not interfere with the radio signal. Although EPP turned out to be the best radio transparent material, the base material of the radôme is EPS (expanded polystyrene) since this is less expensive and more common. Since EPS is used, a good cover should be used to protect it from environmental influences. A good polyurethane coating (POLUREA) was used to protect the EPS, but also takes care of the tension getting into the structure due to the innovative building procedure. A radôme with a thickness of 200mm was built by bending. Parts were glued were needed. The doors also were made out of EPS and the frames for these doors were made out of glass filled epoxy tubes. The total size of the radôme is 17mx17mx4.5m (L x W x H).
Two stations were built. One station at la Station de Radioastronomie de Nançay which is part of de l'Observatoire de Paris, in Nançay, France and the other one close to the Westerbork Radio Synthesis Radio Telescope (WSRT), which is part of ASTRON in the Netherlands. The EMBRACE demonstrator in the Netherlands has been placed between the dishes nr.4 and nr.5 of the WSRT.
To protect the electronics from the outside environment, such as sun, wind, rain, snow, a radôme is needed. The radôme should be capable of handling extreme weather conditions in the Netherlands. When building outdoor constructions for the Dutch climate one has to take the wind force into account. The maximum wind intensity measured in the Netherlands is 12 on Beaufort scale; 36,11m/s (130km/h). The wind intensity varies with the distance towards the ground and the roughness of the ground. So a large radôme will take more pressure by the wind than a small radôme on tile level at the same wind intensity, specified according to the Beaufort wind force scale. When a wind intensity of 36,1m/s has been specified (12 Beaufort), the large radôme will have to deal with 30,3m/s (at 4.7 m). The highest solar radiation of 800W/m2 has been taken into account as this is the common maximum for the Netherlands. Also the radôme should withstand 0.5m of snow.
Embrace considered to apply multiple small tile like radôme of about 1-2 meter size or a single huge radôme of about 20 m typical size. As Embrace is a test bed also and experiments with different antennae configurations are required, a single overall radôme is chosen for this instrument. Future likewise instruments we will go back again to multiple smaller radômes.
Since EMBRACE is a demonstrator for SKA and an instrument to learn about production, assembly, installation and processing obviously maintenance should be taken in to account. For EMBRACE the choice was made for only maintaining the instrument for one year. Thus the design was done in such a way it fits this requirement. It is not easy to replace components and it should only be done if it is really required.
EMBRACE uses coaxial cables to transport the RF signal of the two beams to the backend. DC and control are transported towards the tiles using the same cables. This minimizes additional cost for power and control cabling.
The cables are led in metal wired cable gutters placed along the large 12 m beams of the frame. Except for the middle beam every beam has a gutter. Twelve gutters were placed and every gutter holds the cables for twelve tiles. With two cables per tile a total of 288 cables are present in the cable gutters.
For SKA demonstration reasons, costs have been a key design driver throughout the design process. In a balanced SKA design, front-end bandwidth needs to match with the back-end processing capabilities of the central correlator, to obtain a cost effective implementation. EMBRACE demonstrates how FoV can be tailored to bandwidth requirements by RF beam forming near the elements to reduce costs. Also EMBRACE shows the influence of mass production. The price per element for EMBRACE was enormously reduced compared to previously demonstrators like THEA.
For EMBRACE the numbers of some parts are relatively low, making initial cost still count. For the ultimate SKA initial cost is not an issue due to economy of scale, clearing the path for cheap manufacturing processes (with high initial costs) and low component cost.
For EMBRACE already a nice step towards cost reduction by mass production was made. A Vivaldi antenna, with the required process and tools, cost €1.55. While one element for THEA was about €45. For production of the vivaldis a cutting laser was required and a special mold for bending. For larger scale it will benefit to invest in a stamping tool. Also extrusion is a production process suitable for mass production. This process was used for production of the profiles for connecting the antennae. Also here an investment for a mold was needed. The costs of this mold are spread over the amount of profiles produced. Including this mould and the required labor the costs for a profile are €1.60 When a larger amount of profiles is required the costs of the mold are spread over a larger amount of products, reducing the cost per product. The sub frame which was build out of two part also is about €1,50 per antenna. The large plate required for mounting the printed circuit boards and alignment was about €40 and the glued tubes frame for support €175. Together with all small parts required for assembling of the subframe the total cost of the mechanics of one tile for EMBRACE is €577. This means €4 per antenna.
The platform at the WSRT was installed for €14000 (€0.68 per antenna) and the large frame for €10867 (€0.52 per antenna). The radôme had a total price of €66000 (€3.18 per antenna), which included material, coating, lifting, assembly etc.
All mechanics together EMBRACE has a tile price of €1209, which also means 1209 €/m2.
The SKA will operate over a range of frequencies from about 100 MHz to 25 GHz. The radio telescope array will deploy a collecting area of 1 million square meters and will be composed of a very large number of elements.
The array configuration will include a compact core with about 50% of the collecting area within 5 km, an extended array containing about 75% of the collecting area within 150 km, and the rest in various distant stations out to at least 3000 km. The SKA must be located at a site where there is the ability to do the best science at an affordable cost.
For good antenna performance the Vivaldi antennae that form the aperture array should be electrically connected. Obviously not the complete square kilometer can be connected. The minimal connected amount of square meters is under discussion. It is not unthinkable that it would be unnecessary to connect for example a complete station. For production, assembly and maintenance is would be preferable to have the required connected area as small as possible. From performance point of view the largest possible size is preferable, whether this is realistic, and more important, whether it is required, is to be sorted out. Clear knowledge about this issue could have a tremendously good influence on the development and cost of the SKA.
One of the drivers for the connected area is the edge effect. If an edge effect of less than 10% is required this results in connected modules in the order of 3 times 4 meter.
The edge effect is described as the antenna element at the edges of a connected array in relation to the total amount of elements of this array. As shown in the table one EMBRACE tile has an edge effect of 31%. If all tiles are connected within EMBRACE (144m2) the edge effect is about 3%.