Monday, 3 June 2013

Observing with 190.6 MHz bandwidth

Almost as soon as KAIRA was switched on, we worked out a way to observing with both the High-Band Antenna (HBA) and Low-Band Antenna (LBA) arrays. This is the so-called Mode-357 Observing. This method of observing has been extremely useful, not just to our own observers, but also for other members of the LOFAR community (of course, our scripts can be easily adapted to work on other LOFAR stations). However, the (initial) limitation of this form of observing is that your are still limited to a maximum of 244 beamlets. Each beamlet is a pointing direction and a frequency channel (~195 kHz wide), which means the total bandwidth you that can be observed is about 47.6 MHz. Still, we have been using that successfully for various projects (such as ionospheric scintillation).

In April, we upgraded the KAIRA station with the new LOFAR software. This included some new signal processing modes, specifically: 8-bit and 4-bit mode.

By default, the LOFAR samplers in the signal processing hardware operate at 12-bit. However, by reducing the number of bits, it is possible to increase the number of beamlets. LOFAR 8-bit mode can have 488 beamlets, and LOFAR 4-bit mode can have 976 beamlets. This additional number allows either extra pointing directions or extra bandwidth to be obtained.

KAIRA mode-357, 4-bit observation of Cas A from 22 to 249 MHz.
In the plot, you can see that the frequency range (left axis) is from 22 to 249 MHz. The dark blue horizontal bands are regions that we did not observer (due to radio-frequency interference, RFI). The coloured vertical strips are (mostly), ionospheric scintillation. However, you may also notice a pale blue band that runs from 50 to 70 MHz. This is where the signal is being clipped due to a limited number of sampling bits. In other words, there is not enough dynamic range.

If you look at the spectra of the three different RCU modes side-by-side, you see that the peak response of the LBA amplifiers is actually quite high.


The radio spectrum as measured at KAIRA.

When you observing in LOFAR 4-bit mode, the data is effectively clipped. But there is a way around this. What you need to do is attenuate the signal. The system has the ability to apply up to 31-levels of attenuation in 0.26 dB steps.

Yesterday, we carried out an experiment where we set the attenuation from 31 down to 0 at approximately 60 second intervals. The level-31 attenuation was applied at about second 350.

Attenuation being applied to the signal from KAIRA's LBA aerial L08.

If normalised and plotted as a cross section, you can see how the dynamic range is lost in the next figure.

Normalised cross-sections through the data. Red is 0 dB attenuation, blue is 7.75 dB.

In the same way that KAIRA selectively configures the receiver units (RCUs) to use different modes, and hence antennas, it is possible to do the same thing with the attenuation settings. This means that the attenuation can be selectively increased for only those RCUs with LBA antennas. Although it doesn't completely fix the dynamic range issue, it does result in a huge improvement.
A small sample of a KAIRA Mode-357, 4-bit mode observation.

Although not perfect, it does make this mode usable for certain experiments. While loss of some dynamic range is a problem in some cases, the availability of 190.6 MHz total bandwidth is a very attractive prospect indeed!


No comments:

Post a Comment