LOFAR Signal Path

  This page provides  brief overview of the LOFAR system and introduces a lot of technical
(and hardware) terms used throughout the rest of the documentation.
 
Figure 1 gives an overview of the LOFAR signal path. The radio signals enter the system through the antennae. The antennae (see Antennae Description) contain the first amplifier and, in the case of the High Band Antennae (HBA), an analogue real-time delay beamformer. The signal is then transported via coaxial cables to the station processing cabinet, which contains the digital Receiver Units (RCUs), the Remote Station Processing (RSP) boards, the Transient Buffer Boards (TBBs), and the local control unit (LCU). The LCU is a standard Linux PC that monitors and controls the local station hardware.  

 

LOFAR Signal Path

Figure 1 LOFAR signal path. On the left-hand side the station processing, on the right-hand side the processing at the supercomputer centre in Groningen.

 

The RCUs have three single polarization inputs to connect either the HBAs or one of the LBAs  to the RCU. There are analogue filters directly after the inputs with 10 MHz and 30 MHz high pass filters on the LBA inputs and 110-190, 170-230, and 210-250 MHz bandpass filters on the HBA input. The LBA high pass filters are used to include (10 MHz) or suppress (30 MHz) the strong Earth-bound short wave radio signals below 30 MHz. The HBA band pass filters are used to select an appropriate Nyquist zone for a certain clock and observing frequency combination. Finally, the signal is digitized by a 12 bit analogue to digital converter (ADC) at a sampling frequency of 160 MHz (80 MHz total band width) or 200 MHz (100 MHz total band width).  

The digitized signals can now proceed in parallel to two different pieces of hardware: the TBBs and the RSP boards. The TBBs contain memory buffers which store the most recent 1.3 seconds of data from every RCU. The TBBs freeze and dump their contents if they receive a trigger, which may come from an algorithm running in real time on an FPGA on the TBB. Alternatively, an explicit command to dump its contents right away can be given to the TBBs.  

The RSP boards first split the input signals from the RCUs into 512 subbands via a polyphase filter (PPF) followed by a 1024 point fast Fourier transform (FFT). Further processing is done per sub-band. The most common step after channel separation is the digital, phase rotation based beamformer, which sums the signals of all selected RCUs after multiplication with complex constants that involve a phase rotation based on the geometrical delay required for beamforming towards a certain direction, and any phase/amplitude calibration from the station calibration sub-system. The beam formed signals are subsequently sent to the BlueGene/P over the wide area network (WAN).

The RSP boards also implement a station correlator that can be used to obtain RCU-RCU visibilities for a single sub-band with integration times of multiples of one second. These visibilities can be used for imaging the entire sky at once and very large angular scale structures. This mode is also used extensively in the station calibration procedure.

The BlueGene/P (BG/P) is the real-time processor of LOFAR. It can perform delay compensation and fringe stopping on the incoming data streams from the stations, separate each sub-band into 256 channels using a polyphase filter (PPF) and and FFT, compensate for the bandpass of the station PPF, and correlate (interferometry) or add the signals of (subsets of) stations to form tied-array beams (Beam forming).

The minimum real time integration time is approximately 0.25 seconds. Normal observations will be done at a time resolution of multiples of 1.006633 seconds.

Note that it is also possible to bypass all these operations and directly pass the input data streams to the storage cluster connected to the BG/P. This is the current method for storing time series data, for example pulsar work.  

The BG/P is the final stage of the real time system. The output of the BG/P and any TBB dumps are stored on the post-processing cluster.

More detailed information on each of the stages of the signal path can be found in van Haarlem et al (2012).

 

Design: Kuenst Development: Dripl   © copyright 2012 Astron