##### LTAFILE : node51:/data2/gpuuser/gputest_08aug2013.lta.2 ##### Settings: Observe date 08 AUG 2013 OBJECT 3C147 LO1 1420 MHz RF 1350 MHz Band = flipped (GUI), meaning LO1 > RF (or USB) No. of Channels 2048 Channel-width 97.66 kHz ##### Note from Yash: Just wanted to check about the status of getting some plots and results out from the test data of Thursday, 8th, so that we can feed back into next week's plans. Note from Dharam/Sachin: Sachin has generated a few plots (Data acquired on 8th Aug, after 3pm, IST onwards from one 6-hour long lta file), which are attached along with. - Time averaged, self bandshapes (file names Self_scan*) for scans 0, 5 and 10 for each antenna as a function of channels. Each file show 8 pages, corresponding to eight antennas used. Here in the top X-axis, the adjacent two ticks are 200 MHz apart. Self bandshapes for several scans are presented to get a feel of variations in it, if any as a function of time. - Phase and amplitude for channels 200, 500, 1000 with C04 as the reference antenna are also attached. Settings: RF = 1350 MHz LO1 = 1420 MHz BW = 200 MHz In this lta file, data was acquired for ~6 hours. There are still issues, e.g., with Date (17/Nov/1858) being displayed by ltahdr (in node51 and astro0 machines). Also, it is hard to understand RF and LO1 used using ltahdr, these are creating confusion. Sanjay pl. fix these... Reply note from Yash: Really good to see these plots, with stable amplitude and phase (on at least central square baselines) over 5-6 hrs duration : shows that the basic systems are working fine to first order -- very nice ! Upon a more detailed perusal, the following items come to mind : 1. It would be much more illustrative if the cross plots could be made "with lines" instead of "with points" as the finer variations are more easily seen with lines e.g. the small oscillations in the self plots. Request that these be remade with lines. 2. Also, is it possible to make one version of the self plots with all the antennas overplotted on the same page, again for better illustration and comparison. 3. Amongst the self plots, those for S02 and S04 look completely unexpected (wrong) in position for scan 0, intermediate for scan 5 (only S02 is off) and better for scan 10 (S02 is still a bit off) -- was there some adjustment of the LO settings done between the scans? Other than these, the self plots look reasonable, except for the level of oscillations in the band. 4. Like the selfs, is it possible to make some plots of cross amplitude, to show how the cross-correlation behaves across the band? 5. Similarly a couple of plots showing how the cross phase behaves across the band would be useful -- to check the amount of phase wrap etc. 6. The time plots of ampl and phase for 3 channels across the band are interesting; however, I think couple of them are incorrect plots : amp_all_chan200_ref_C04.jpeg is actually a phase plot; phase plots for chan 200 and 500 are identical plots. These corrections can be made when replotting these with line option. 7. All the time plots show frequent, quasi-periodic breaks in time -- I presume these are gaps between scans ? 8. The cross-correlation coefficients show large variations for different antennas (and in some cases with time) : values range from as low as 0.004 to 0.03. How do these compare wrt the expected values? Dharam : could you look at the first order comparison with expected values. 9. S04 seems to be the only antenna not working at all -- a cross-check on the settings for this antenna would be useful. 10. The issue of values reported in the ltahdr not matching with the correct numbers is disturbing -- this should have been resolved and corrected a long time ago. Sanjay : you need to look at this URGENTLY. Is it possible to cross check what is seen when the file is converted to FITS and read into AIPS? Overall, I think this is good progress. We are over the first hurdle of getting reasonable data from the wideband system. We now need to move forward to sort out the lower level issues and problems. ##### Note from S. Kudale: > There are still issues, e.g., with Date (17/Nov/1858) being displayed by ltahdr (in node51 and astro0 machines). Also, it is hard to > understand RF and LO1 used using ltahdr, these are creating confusion. Sanjay pl. fix these... > 'ltahdr' in node51 ( released version ) should work, I'll fix it if it does not. Also, I'll give you whole set of lta offline package installed in some NIS m/c. Existing observatory package shall work for new gpu corr lta data. Also, reg. the freqs RF/LO, we need to decide what LO to come in ltahdr, but most relavant information is RF freq. Pl. ignore other freq. There is conflict in understanding RF/LO values given by ONLINE vs GPU GUI TPA to GPU correlator. We may have live in this situation for some time. But, you could suggest your model to implement, I'll implement it.. Also, this is related to sideband flag going into LTA data also. So we'll add these both features when you are ready. ##### Note from Sachin/Dharam: additional plots incorporating your suggestions in comments 1-6 are attached along with. - Plots (phase_vs_allchan_ref_C04_*.jpeg) show phases over all channel with C04 as a reference for 0th, 5th and 10th scans. - Plots (amp_vs_allchan_ref_C04_*.jpeg) show amplitudes over all channel with C04 as a reference for 0th, 5th and 10th scans. - Plots (amp_vs_allchan_self_*.jpeg) show self bandshapes over all channel for 0th, 5th and 10th scans. - Plots (phase_allant_*_ref_C04.jpeg) show cross-phases for 200, 500, 1000 channels with C04 as a reference antennas for all scans. - Plots (amp_allant_*_ref_C04.jpeg) show cross-amplitudes for 200, 500, 1000 channels with C04 as a reference antennas for all scans. - Time averaged, self bandshapes (file names Self_scan*) for scans 0, 5 and 10 for each antenna as a function of channels (all antennas over-ploatted, file self_*_allant_overplot.ps). Two concerns: - now on node51 the disk space is not sufficient, a quick solution is requested. - node51, does show correct DATE, but we are still worried of LO1 value in it. For example, in this observation, RF = 1350 MHz, which corresponds to channel zero and we expect the frequency to decerease as channel no. increases, i.e., channel 2047 = 1150 MHz. But the presence of LO1 = 1420 MHz is creating confusion. And to S. Kudale - node51, yes! released ver. os ltahdr does show correct DATE, but we are still worried of LO1 value in it. For example, in this observation, RF = 1350 MHz, which corresponds to channel zero and we expect the frequency to decerease as channel no. increases, i.e., channel 2047 = 1150 MHz. But the presence of LO1 = 1420 MHz is creating confusion. re. your suggestion, sure the users should tell the inputs needed in the header, etc. May be, Yash may have better suggestion.