22 November 2004
WSRT observes in the LOFAR (high)band
Ger de Bruyn - Bert Woestenburg - Hans van der Marel
As of late September 2004 we have started to commission a new suite of receivers for the Westerbork Synthesis Radio Telescope: they have been nicknamed the LFFE, which stands for Low Frequency Front Ends. The frequency range covered by the receivers runs from about 115 to 180 MHz. These boundaries are defined on the low side by the FM band and by a strong local TV transmitter (Smilde, Ned 1) on the upper end.
Preliminary results are presented below. They are giving Dutch astronomers a sensitive first look into the low frequency Universe which, within a few years, will be the playing ground of LOFAR. But LOFAR, however, will have 40 times better angular resolution, much better sensitivity and a decade in frequency coverage. Exciting times indeed are ahead for low-frequency radio astronomy.
Delivery and installation of the antennas and receivers proceeded on a tight but realistic schedule. Following a preliminary design study during 2002, funding to build 14 receivers was requested from NWO in the Spring of 2003. Following the decision to fund the project in June 2003, design and construction started in earnest in the Summer of 2003. The first prototype was tested on the telescope in November 2003 and was followed by PDR in Jan 2004. Delivery and installation of the full suite of antennas and receivers started in June and was completed by mid October.
>> The animation shows how the (2+2) feeds are moved into position in the prime focus of the WSRT. (komt later)
This must have been one of the fastest projects ever realized by ASTRON and congratulations are due to the various groups within ASTRON: the Technical Lab for the design and construction of the receivers, the mechanical workshop for the beautiful feed design and the Radio Observatory for the actual mounting of the feeds and the electrical and software coupling to the IF and backend.
Now it is up to the astronomers to start the technical and astronomical commissioning. Although this is still in full swing at the time of writing the WSRT Program Committee, prior to its Nov 10 meeting informed of very positive results, decided to allocate the first batch of LFFE proposals and scheduled observations will commence on November 29 2004.
Lots of aspects of the system still have to be tested and characterized. The large volume of data, about 10 GByte in raw MS (measurement sets) for a single 12h run and double that once the DZB-recirculation is made available (really soon now !) are clearly testing our software, CPU, and stamina.
Thusfar we have completed 12-hour runs on four different objects. Three of them were on sources that were previously found to be highly polarized at 350 MHz (two of which are ms-pulsars). They will allow us to test polarization performance and study the effects of ionospheric Faraday rotation.
The powerful and flexible WSRT backend (the DZB) was configured to cover the available RF band with 8 bands of each 2.5 MHz at frequencies centered around approximately 117, 123, 130, 139, 141, 147, 163 and 174 MHz. Each band has 64 channels x 4 pol. In all observations we used 2-bit sampling, 10sec integration and Hanning taper.
As was expected the LFFE observing window is shared by many other users. Dedicated filters have been installed to suppress some of the strongest sources. However, there is still quite a lot of RFI, some of which is very strong. It is usually narrowband and intermittent, and can be flagged. The RFI situation is considerably better in the early morning hours (midnight to ~ 6 AM, due to reduced airtraffic a.o.). We are still in the process of selecting the optimum set of frequency bands.
Observations in 'cold' areas in the North Galactic halo (on the sources 3C196, 3C295 and DA240) reveal that the system NEFD (Noise Equivalent Flux Density) of a telescope (sky + antenna + receiver) is about 7000 Jy, in agreement with results presented during the Jan 2004 design review. From this we calculate a thermal noise level of about 1 mJy after 12-hour observing assuming a useful total band of about 10 MHz. In many fields the noise will go up a factor 1.5 due to increased Galactic noise levels. The computed classical source confusion noise levels for the WSRT 3km-array, however, is about 2-5 mJy, depending on the observing frequency. In total intensity images we will therefore not reach the thermal noise level.
Good 12-hour observations were made on Oct 15, Oct 31 / Nov 1 and Nov 12. The images on 3C147 (with 13 working telescopes) and DA240 (12 telescopes) are shown in the accompanying panels. Ionospheric refraction was very benign in the 3C147 observations and allowed us to make diffraction limited images (with a resolution of 120"x150") over a field as big as 24 x 24 degrees ! However, in the outer parts of the images the source density is low, because the halfpower beam measures 'only' about 6-8 degrees. We are still working on the data but the noise level in the current image of 3C147 (itself a 56 Jy source) is less than 10 mJy in the inner parts of the image and reduces to less than 5 mJy at the edges of the field (and 3 mJy in Stokes Q). The 3C147 image includes data from bands at 139, 141, 147 and 163 MHz. The grating lobe and general sidelobe confusion in the images is enormous and the achievable noise in a single 12-hour observation will always be limited by grating sidelobe confusion, rather than source confusion. Multiple 12-hour syntheses will allow us to go the classical confusion noise. The image of DA240, at a frequency of 117 MHz, is still severely limited by grating lobe noise (note that only 27 different baselines were used to make this image !).
In all fields analysed to date the famous A-sources (Cas A, Cyg A, Tau A and Virgo A) have been detected. Their apparent flux densities range from a few to 10 Jy ! In addition, we also recorded a free Solar flare in the distant sidelobes in two of the datasets in early November.
All of these sources are of course highly instrumentally polarized. The polarization performance of the antennas is well-behaved and we may look forward to widefield astronomical polarization imaging at very low frequencies. As far as we know this has never been done before in the history of radio astronomy.
To our delight, although not unexpected, we have discovered that the ms pulsars PSRJ0218+4232, and PSR1937+21 are still highly polarized, permitting their use as ionospheric Faraday calibrators.
The standard software packages probably will come a long way in delivering useful images, as long as the ionosphere behaves not to wild. The images shown here were all processed within the NEWSTAR package. For the ultimate imaging, which included 'peeling' to take care of non-isoplanaticity problems in less friendly ionospheric conditions, we expect to need the new LOFAR-modeled software that is currently being developed and tested.
Stay tuned for more new in the next few months.
Ger de Bruyn, LFFE Project Scientist
Bert Woestenburg, LFFE Project Manager
Hans van der Marel, WSRT telescope physicist