DE605 RFI

There is evidence for interfering equipment on a nearby cell tower that affects the low band. To investigate details, we use one TBB dump from 2023-10-26 13:34 UTC, and 16 dumps taken during a Jupiter observation (rcu mode 3) on 2023-12-27 20:35–22:26 UTC. Each dump used here has about 1 sec duration. In the standard analysis they get correlated, calibrated and imaged with the standard LOFAR frequency resolution (subbands of ca. 200 kHz) and with 0.1 sec time resolution. Visualisation of such a data set is not easy. My standard approach is to select a frequency (and maybe time) range, show the corresponding averaged allsky image, and potentially select a point on the image to plot the spectrum and time series.

Here is an example plot for a dump taken 2023-12-27 21:19 UTC:

(click on the image for full-size version)

The panels on the top right show the total dynamic spectrum and the integral over time and frequency. The red-shaded region is used for the imaging on the left. The maps show four polarisation products. Note that these are not exactly the Stokes parameters, but combinations of X and Y. The sky is shown in an orthographic projection, which compresses the elevation range near the horizon. Outside of the horizon circle the plots show nearfield imaging of the ground level, where the distance goes with $1/(r-1)$, it thus starts at infinity on the circle (consistent with the sky image at the horizon) and then decreases further away from the circle.

The lower right shows the spectrum in the direction of CasA (relative to the mean sky spectrum).

In this example the signal is dominated by the sky, but some RFI is seen near the horizon.

Here we use the same dump, but image the frequency range 54–60 MHz. The RFI source is seen at the horizon near azimuth -33 deg (north towards east). The lower right panels are for this direction. Note that the spectrum looks very different from CasA, but is stable with time.

This source seems to be there at least most of the time, with a pretty stable spectrum.

To get an overview, we have to reduce the number of dimensions. Because many RFI sources appear at the horizon at large distance, it seems appropriate to only plot the signal around the horizon.

This PDF shows such plots for all dumps from 2023-12-27, subdivided into 0.1-sec chunks (11 per dump, the last one is always shorter): tbb_image_horizon_spec_de605_2023-12-27_jupiter.pdf Values for each frequency are normalised by the median around the horizon for this frequency. The dump shown above is on pages 67–77.

Let us show integration 5 of the same snapshot as before:

The RFI source shown above is the bright blob below the centre of the image. It is quite prominent, but not the only obvious RFI. When going through the PDF pages, we see that some of the features are very variable with time. Also note (A) the very bright big blob below 20 MHz around azimuth -70 deg. Also very special are (B) the narrow (in frequency) spikes below -100 deg, and (C) the bright broadband signal near 140 deg. We will come back to these below.

To investigate the time dependence better, we plot the same dump with now much shorter subintegrations of 2 msec on 511 pages (note large file): tbb_image_horizon_spec_de605_2023-12-27_jupiter_2msec.pdf The celltower RFI is constant at this time resolution, but many other sources are extremely variable, sometimes periodic.

To study the time variability better, we plot the same data differently as one dynamic spectrum per azimuth and page: tbb_image_horizon_spec_de605_2023-12-27_jupiter_2msec_dynspecs_fft.pdf The dynamic spectra are in the left panels, the right panels shows a Fourier transform along the time axis to show periodic power variations. The celltower azimuth is strongest on page 74:

There are traces of 100-Hz periodicity over a wide frequency range, but the strong signal below 60 MHz does not show this at all.

The bright blob called (A) above does not show interesting behaviour in the time domain. The transient spikes (B) can be seen around page 26:

It is actually worth looking at a few other pages around this number, because these features vary a lot also with azimuth. These signals are very short spikes that at 10-msec intervals, but not all intervals are actually active. This shows as strong periodic signals with 100 Hz (but also 50 and 200). Maybe this is related to sparking? There is a 380 kV line in that azimuth range, and maybe the strongest azimuth values correspond to individual masts.

Finally we have (C) the broadband stuff near 140 deg, e.g. on page 160:

This does again show strong periods of 50 and 100 Hz and multiples thereof, probably related to the grid frequency. The azimuth range corresponds exactly with the direction towards the transformer station Oberzier

From about 13:15 UTC of 2nd to about 14:30 UTC on 3rd Dec we ran the station in rcu modes 3/5 (10–90/110–190 MHz filters), recorded SST files continuously, and took TBB dumps in LBA mode, and three types of HBA settings: single-element-per tile, beamformed towards CasA and beamformed towards CygA. The beamformed ones are skipped whenever the elevation of the target is below 25 deg. Dumps had durations of 1.024 sec for the beamformed HBA parts and 1.536 sec for the LBA and single-element-per tile HBA parts. The latter are meant for the RFI studies, and they are a bit longer to extend beyond most electric fence periodicities. The LBA and the beamformed HBA dumps will also be used to compute calibration tables.

Here are azimuth/frequency plots for all TBB dumps, subdivided in 0.1-sec subdumps per page: tbb_image_horizon_spec_de605_2024-12-02.pdf
And here the corresponding dynamic spectra with one azimuth (2-deg steps) per page: tbb_image_horizon_spec_de605_2024-12-02_dynspec.pdf
The labels on the horizontal axis are a bit confusing. This is actually in seconds if we concatenate all dumps by skipping the gaps between them. It is about 1.5 sec per dump. The potential celltower RFI is strongest on page 74:

It is unclear what causes the dips in frequency and intensity below times of 10 and between 45 and 65.

The other features (A-C above) are either absent or less obvious now. The weird structure that can be seen on page 20 (second dump in the left) is actually the Sun, which happens to be near the horizon and active at this time:

After the TBB dumps, we also recorded a beam towards an azimuth of -33.5 deg at the horizon for 1.5 minutes starting at 14:45 on 3rd Dec. That is after the dip of the celltower signal shown above, but at a lower level than the day before.