The LWA1 station consists of 256 dipole antennas spread over a 100 m diameter area. The minimum distance between each antenna is 5 m, and the position of each antenna is pseudo-random. The geometry is determined by minimizing the station side lobes. One additional dipole antenna is located on a 300 m baseline, initially for calibration purposes.
The path for a signal through the system is illustrated in Figure
1. The signal enters the antenna where a first gain stage
is located (Sect. 2.3) and is then transferred to an
analog processor for a second gain stage and filtering options (Sect.
2.4). The signal then enters the digital processor, where it is
digitized (Sect. 2.5) and distributed in parallel to the
transient buffers (Sect. 2.8) and to the beamformers where an
electronical delay is entered into the signal path to correct for
pathlength differences (Sect. 2.6). The digital receiver
then furthers the signal by determining the tuning frequencies and
bandwidths (Sect. 2.7), and eventually the signal is recorded.
The following subsections of this document contains more details of
each step in this signal path, see also [Craig (2009)].
A single antenna element is a orthogonally and linearly polarized ’tied-fork’ dipole (Fig. 2), with the arms drooping downward at about 45∘ [Craig (2009)]. They are placed on top of a ground screen to stabilize the antenna impedance and improve the collecting area.
The first gain stage is placed at the antenna, consisting of an active balun providing about 36 dB gain.
The analog processor provides a second gain step of about 68 dB gain, of which 60 dB can be adjustable in 2 dB steps. The adjustment level will depend on the signal strength and will be determined at a system level rather than by the user.
To allow some flexibility in the rejection of strong RFI signals, the analog receiver provides three filter configurations (Fig. 3; [Craig (2008)], [Craig (2009)]) that can be set by the user:
The full bandwidth option implies a passband between 10-88 MHz, while the reduced bandwidth option suppresses signals outside the 28-54 MHz frequency range (both options with a response within 3 dB across the given frequency range).
The split bandwidth filter option is introduced to allow a stronger suppression in the low frequency part of the band, especially below 28 MHz where the RFI signal power is highly variable in time. Below 28 MHz, the attenuation is adjustable in steps of 2 dB up to 30 dB maximum. If the gain of this part of the band needs to be adjusted during observations, it should be noted that the gain changes may require the station array to be recalibrated.
Figure 3: The response of the analog receiver across the passband for each of the three configurations: Full (blue), reduced (green) and split (red). The three red lines correspond to to different attenuator settings of 6, 12 and 20 dB respectively.
After the first stages of gain the signal enters the digital processor, where it is first direct sampled with 12 bits at 196 Msps. This sampling rate has been chosen for an alias-free digitization of the whole 10-88 MHz frequency range.
After digitization, the signal is distributed to 5 parallel processors. There are 4 beamformers (Sect. 2.6), and one transient buffer (Sect. 2.8).
The beamformer is a part of the digital processor system, and is responsible for adding delays to each dipole signal to form beams in the desired pointing direction. The delays are automatically calculated dependent on the pointing position specified, and are not directly accessible to observers. The bandwidth of the sampled signal at this stage is the full digitized bandwidth. Four different beams can be formed that can be run separately and simultaneously, effectively giving 4 telescopes.
Following the beamformer, the digital receiver processes the data to the desired tunings. The signal from each beam can be tuned to two different center frequencies ν1 and ν2 between 10-88 MHz. The center frequency settings are on a grid with spacings of 0.046 Hz.
Each tuning will have an associated passband of width B1 and B2 that do not need to be the same. B1 and B2 can be set to values between 0.250 MHz to 19.6 MHz (see Table 2.7). The actual usable bandwidth is somewhat smaller due to the band edges, and a conservative number is 0.8. Observations have shown that observing with a 19.6 MHz bandwidth gives an effective bandwidth of approximately 16 MHz.
Currently, the default output is the raw time series data. There are two options for the user to get spectral data, via either a software or a hardware spectrometer. Using the software option, you can use LSL to set an FFT of any length (with the potential of up to tens of thousands of channels). The time resolution for the FFTs will be limited by the length of the window. For example, a 1024 point FFT will have a spectral resolution of 19.1 kHz (assuming 19.6 MHz bandwidth observed), and the maximum time resolution will be 1024 times 51.2 ns or 5.22µs. The post processing software part of LSL can control spectral resolution and process to the Stokes parameters. The cost of this is that the software is slower and of course another step in the data reduction process. The second option is to use the hardware spectrometer (which still is under development). The spectrometer gives 32 channels, so if using the spectrometer together with the smallest bandwidth of 0.250 MHz the minimum channel width that can be achieved is 7.8125 kHz. The spectrometer averages 6144 spectra, which at 19.6 MHz will give a read out every 10 ms. This will successively increase as the bandwidth is decreased.
Filter number Bandwidth Usable bandwidth1 Spectrometer Δν2 Spectrometer Δ t3 MHz MHz kHz ms 1 0.25 0.20 7.8125 784 2 0.50 0.40 15.625 392 3 1.0 0.80 31.25 196 4 2.0 1.60 62.5 98 5 4.0 3.20 125.0 49 6 9.8 7.84 306.25 20 7 19.6 15.68 612.5 10
Table 1: The available LWA1 DRX filter bandwidths and estimated actual observed bandwidths.
As an example, an observer could point a beam to a source, and define 2 individual frequencies. With a 9.8 MHz bandwidth, about 7.8 MHz would be achieved per frequency and polarization. In total, a bandwidth of 16 MHz could thus be covered. A second beam could be pointed to a different source, or tuned to a different set of frequencies.
Parallel to the processing of the beamformer and digital receiver, the signal from each dipole is also fed into the transient buffer system after digitization. The transient buffers record the raw digitized signal directly for each polarization and antenna.
The Transient Buffer Wideband (TBW) records the full bandwidth output from the digitizer for as long as possible, which currently have two options available. Per trigger you can observe either 12× 106 12-bit samples with a capture of 61.2 ms (giving Δν=16 Hz) or 36× 106 4-bit samples with a capture time of 183.7 ms (giving Δν=5 Hz). There is approximatly 60 s between each trigger, resulting in ∼ 0.1% duty cycle.
Science uses of TBW includes long duration total power transients, solar observations and riometry.
A Transient Buffer Narrowband (TBN) will be available, which will collect data continuously. The TBN records data without any beamforming applied, and records all data streams, both polarizations, for each individual dipole. User settings for this buffer are the frequency (10-88 MHz) and sample rate of (1-100 ksps, see Table 2.8.2).
Science uses of TBN includes all-sky transient searches, and perhaps recombination lines.
Filter number Sample rate (ksps) Effective bandwidth (kHz) 1 1 0.67 2 3.125 2.09 3 6.25 4.19 4 12.5 8.38 5 25 26.75 6 50 33.5 7 100 67
Table 2: The available LWA1 TBN filter bandwidths
For LWA1, the data recorder consists of a LAN and a network connector. The four beams processed through the digital receiver are being recorded on one station computer each, and a fifth station computer records the output from the TBN/TBW.
The storage unit is a RAID array, consisting of 5 streaming-tuned hard disk drives. For more information and how to assemble a data recorder storage unit see [Wolfe, Ellingson & Patterson (2009)].
