Notes from the August PPTA observations ... (all observations carried out using the 10/50 receiver with CPSR2 observing at 50cm and the WBCORR/pDFB at 10cm). In order to compare the WBCORR and pDFB easily (and to ease cable changes), the WBCORR was run with only 256 MHz of bandwidth even though we were observing at 10cm. All the results shown below are obtained without doing any offline calibration.

Simultaneous observations of 0437 with identical set-up for WBCORR and pDFB

Observed PSR J0437-4715 simultaneously with pDFB/WBCORR with 256 MHz bandwidth, 128 channels and 256 bins.

The following plots show the resulting WBCORR (top plot) and pDFB (lower plot) profiles after the complete observation. The WBCORR is affected by the "2-bin" problem.

In the figure below, the green line (lower) indicates the signal-to-noise ratio obtained using pav for different numbers of subints included in the analysis for the WBCORR. The red line is the SNR for the pDFB observation and and blue line is for the WBCORR after the 2-bin artificats are removed using paz. The purple line indicates the expected slope (sqrt(t)) for the SNR. Clearly the artificats in the WBCORR data are seriously reducing the measured SNR. The pDFB shows no such artificats and the SNR seems to increase as expected with time.

For the 1-hour observation simultaneouly observed with the WBCORR and the pDFB, I formed a standard profile from the entire 1-hour observation and then used that standard to obtain TOAs from individual (and combined) sub-ints. The results are shown in the Figure below. The green line (lowest) indicates the residuals obtained using the pDFB (rms residuals from 400ns to ~100ns) for different data lengths. The red (central) line gives the same for the WBCORR data and the blue (line) is for the "corrected" WBCORR data after removing the 2-bin artifact. I also attempted to obtain TOAs using Russell's method for removing the WBCORR artifacts in 'pat', but obtained almost identical results as the uncorrected WBCORR data (red line) - maybe I used Russell's program incorrectly? Again, the figure shows that the pDFB is providing better results than the WBCORR although there is some issue that the rms residual is not decreasing below 100ns. To explore this in more depth, I'll try and make some longer observations of 0437 tomorrow with the same setup.

2048 channel x2048 bin mode

Prior to this observing session, the 2048x2048 mode did not work on the pDFB. Warwick fixed the bug and we managed to obtain observations using this mode. Note, the cycle time needed to be changed to 10s to 20s for this mode (and changed back afterwards!) because the writing of the cfits data takes up to 13s. We observed the "newclass" of pulsars timing list with both the WBCORR and the pDFB (at 2048x2048). PSR B1713-40 was detected using both systems (pDFB result below):

Simultaneous observations of other pulsars (same bandwidth/different nbin/nchan)

0024-7204J

Single observation made (left-panel = pDFB, right-panel = WBCORR):

0613-0200 (note: the profile should be in the centre!)

0711-6830

1022+1001 (should be in the centre!)

1024-0719 (should be in the centre!)

1045-4509

1435-6100

1600-3053

1603-7202

1623-2631

1643-1224

1713+0747

1721-2457

1730-2304

1744-1134

1757-5322

1824-2452

1857+0943

1909-3744

1939+2134

2124-3358

2145-0750

2317+1439

Tempo2 developments

Have turned grouped parameters into arrays, e.g. F0,F1 ... F10, DM, DM1 ..., glitch parameters, BTJ jumps ... For instance, it is now possible to have as many glitches as memory will allow.
Have implemented Shapiro delay parameters into the ELL1 model
Have installed tempo2 at UBC for Ingrid
S. Ord developed new plugins: m2sini and resplot
Currently removing the need to link with AIPS++ in order to carry out global fitting

Observing

John S., John R. and I suggesting using part of the "shut-down" time tomorrow to observe J0437-4715 from rise to set using the correlator, DFB and wide-band correlator simultaneously at 10/50 to study systematic effects.
Summary webpage is working again and John S. is zapping channels etc. However, we need to organise the RAID arrays better and the scripts at Parkes and Epping are not currently checking the DFB directories
DFBCORR (equiv. of PSRCORR) does not automatically archive the DFB data. This needs to be fixed.
J1022+1001 is currently behind the Sun (directly behind on 29/08 !) and will be affected by DM variations. Suggest in the next run to observe regularly at multiple frequencies.