This experiment was originally proposed as a commissioning observation to be carried out in ILT mode. Details can be found in the original proposal. Because cycle 0 observations have higher priority for the ILT, we decided to use locally recorded data instead. After important system and calibration updates in LWA1 in the second half of March, we started planning the details. First we wanted to use Onsala and Effelsberg and maybe Jülich (with reduced bandwidth), but at the Dalfsen II meeting, Derek McKay and Olaf Wucknitz decided to include KAIRA as well. That plan emerged on 22nd March, and the observations were planned for 26th March 2013 for one hour at 2-3 h UTC. As target we chose B0809+74, which is a bright pulsar with period 1.2922 sec that suffers only from very little scattering so that it should be pointlike even on the transatlantic baselines.

Even though the LWA observations have slightly less than 40 MHz bandwidth, we decided to record the maximum of 244 subbands in 16-bit mode, which corresponds to 48 MHz total bandwidth. We selected subbands 166-409 (32.42-79.88 MHz). The beam is formed with the following command:

beamctl --antennaset=LBA_OUTER --rcus=0:191 --rcumode=4 --subbands=166:409 --beamlets=0:243 --digdir=2.1598086,1.30000704,J2000 &

To maximise robustness and uniformity, we recorded raw UDP packets that will be interpreted offline.

I (Olaf Wucknitz) recorded one hour of beam-formed data with read_udp_gen2.c on the lofarB1-4 machines. The total data volume is about 1.4 TB. In addition, local correlations were recorded for subband 388 (75.78 MHz), as well as subband statistics. Before the actual observation (at 1-2h UTC), some tests were added. The day before, LBA observations of CasA and CygA were used to create a calibration table for RCU mode 4. This table was used for the beam-formed observations and should increase the SNR by about a factor of two.

The main files are:

327G Mar 26 04:00 read_udp_gen2.lofarb1.2013-03-26T01:58:59
326G Mar 26 04:00 read_udp_gen2.lofarb2.2013-03-26T01:59:00
327G Mar 26 04:00 read_udp_gen2.lofarb3.2013-03-26T01:59:01
326G Mar 26 04:00 read_udp_gen2.lofarb4.2013-03-26T01:59:02

When doing the calibration observations, I also made short beam-formed test observations in the low band. Just before the main observation I did the same in the high band. One HBA target was B1919+21 at rather low elevation. Interestingly I see strong periodic signals that are not related to the pulsar. Individual pulses are extremely sharp and not dispersed. See the time series plot and periodogram (with harmonic summing):

The main period is about 1.1175 sec. This period seems to be rather stable over a few minutes, but there is noticable jitter of the individual pulses. In addition there are some weaker periodic signals with different frequencies. The period of B1919+21 itself is 1.3373, but it cannot be seen here, partly as result of the dispersion that is not corrected. The plots show 1/4 of the total bandwidth. This looks like signals from (an) electric fence(s). It could well be that the strong RFI that we are seeing in the low band coming from the direction of the 100m telescope is also reflected electric fence pulses, as already suspected. If this can be confirmed, there should be ways to solve the problem.

As additional test I observed out main target B0809+74, also in the high band. The dispersion is relatively low, and individual pulses can be seen easily.

A dynamic spectrum, the time series and periodogram look like this:

The fact that we see prominent pulses is very promising for the main experiment. The low band has less sensitivity, but there should still be enough signal for a clear detection.

Originally we had planned to include Jülich as well, but because things would have to be done manually, Olaf decided that juggling with DE601 and DE605 increases the risk of failure more than it would help. Therefore Jülich was not included this time. We would have been able to record with only 1/4 of the bandwidth anyway.

Onsala (Tobia Carozzi) recorded raw UDP packets with their own system. We have the following files, one for each lane, each of which has 320 GB:

343957154496 Mar 26 11:39 SE607-Obs36_0.lbp
344019144048 Mar 26 11:41 SE607-Obs36_1.lbp
344010279456 Mar 26 12:36 SE607-Obs36_2.lbp
344019144048 Mar 26 12:36 SE607-Obs36_3.lbp

The download to Bonn took between 5h39m and 7h41m per file, which corresponds to about 15 MB/s. We can use dedicated e-VLBI connections for future experiments.

Derek McKay observed with KAIRA, using a modified version of Olaf's simple recording software. The modifications were necessary, because one machines is used to receive all four lanes via separate network connections. The KAIRA software would have modified the data format. Unfortunately it turned out that full-bandwidth recording was overflowing the buffers, possibly because of limited disk write speed. Therefore we decided to continue with only the two central lanes. These are the 122 subbands 227-348 (44.34-67.97 MHz).

The observations are featured by a KAIRA blog entry.

Transfer to Bonn is not entirely easy, because the station does not have a high-speed link to the institute. As first test, Derek recorded another few minutes with one lane after the main observations:

read_udp_gen2.kaira4347.2013-03-26T03:00:40

Those 4 GB could be transferd to Bonn with 2-4 MB/s in about half an hour. With the same speed the full 700 GB will take about 2-4 days, which is acceptable.

Frank Schinzel took care of the observations at LWA1. Two tunings with 19.6 MHz bandwidth (and a bit of overlap) were used:

tuning 1:  66.9921875 MHz = 196MHz/2^32*n with n=1468006400
tuning 2:  47.8515625 MHz                      n=1048576000    in data: n=1048576064 (47.851565420627594 MHz)

The total frequency coverage is thus 38.05-76.79 MHz, slightly less than LOFAR.

The main file is:

310G 056377_000293256

It was downloaded from here. Frank Schinzel made an analysis of the LWA signal. Result files can be found here. Most important are the PRESTO plots that show clear detections in both tunings:

The original PS versions are also available locally: tuning 1 and tuning 2

The data format is described here, more info about the LWA in general here. The LWA software suite is publicly available, see here. For test we also use the Pulsar extensions and psrfits_utils.

There is also a short tutorial.

For some consistency checks I am using standard pulsar software tools.