2011 July 12. NIKA run 2 Data meeting at IRAM Grenoble, R. Zylka main results
Back to SL's NIKA page
R. Zylka found many oddities in the NIKA run 2 data (e.g. gradients, non-linearities, jumps, negative signals, cross-talk, beam / geometry distortion, correlation instabilities, tc.) txt file of comments on the images below,
or all these gathered in one pdf file
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
The analysis performed using mostly only 2mm, as a large part of the 1mm data we got is incorrect !
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
1. the major (if not the only) difference compared to other cont. receivers is the calculation of the signal:
(I,Q) -> R [mHz] R should be proportional to the total power/source signal
example of a "frequency sweep", i.e tuning of NIKA
fig.1a 3 freq. sweeps as function of time
clear gradient as function of the KID number visible, similar to the gradient of the flat
field in fig.12a; the signal is smaller in the 2nd & the 3rd sweeps;
the 2nd & the 3rd sweeps are nearly identical, i.e. this decrease is not a random effect
fig.1b same data in the (I,Q) plane; better visible that the second (in green) and the third
(in blue) sweeps gives smaller circles in (I,Q);
green is hardly visible as overplotted with blue, i.e. the two are almost identical
2. immediately visible problems in the linearised signal R
- instabilities
fig.2
- jumps, >11 during ~1300sec, marked with vertical grey lines
- jumps are not equal - KID dependent
- atmospherical signal is not well correlated (comp. red, blue ... black)
- negative signal
fig.3
KID87 shows negative signal
- cross talk
fig.4 in KIDs 4, 15, 18, 19, 20, 26, 53?
also important: KID71 does not show the signal of Uranus
clear distortion of most beams =>
- beam distortion (as no focussing), i.e. beam efficiency might be changed even by 50% =>
calib. factor by 2
fig.5
- pointing not corrected during the whole run =>
statistics on pointing corrections for NIKA not possible
the beam maps show offsets of up to 10asec
(these offsets affect even the calculation of the FoV geometry)
statistics of pointing corrections for EMIR, period of NIKA run #2
(NIKA used the pointing model of EMIR)
fig.6
3. further instabilities
- correlation (flat field) plots show many different problems, also in groups of KIDs =>
the "noise" in the final image is dominated by these instabilities
fig.7
- "50sec steps" (not shown)
- oscillations visible after removing of the correlated signals (~ sky noise)
with different zoomming factors different oscillation frequencies visible (as aliasing appears)
fig.8a e.g. KIDs 1, 11, 23, 32, 69, 70
fig.8b
fig.8c e.g. KIDs 1, 11, ... 69, 70, ... 99
4. relative calibration
- FoV geometry: distortion, position of KIDs 21 & 22 @2mm changed due to a cross talk
fig.9
fig.10
fig.11
- flat field: change >10 for 2mm, > 3 for 1mm
fig.12a clear gradient, similar to the gradient seen in fig.1a
fig.12b KID35 <<, KID42 >>
fig.12c very flat
- statistics of the flat field
fig.13
fig.14
5. comments on mapping
- efficiency
eff ~ (time of a subscan) / [(time of a subscan) + (time for an U-turn)]
time for an U-turn >~5sec
time of a subscan deltaAz/vAz
vAz = 10 to 20asec/sec, i.e. even 4*Nyquist samplig of the coords @1mm (slow loop with 1Hz !)
=> time of a subscan even just 6sec
=> efficiency down to ~50% !
- map sizes
the smallest map size in the scanning dir. is given by:
source size + array diameter + HPBW + "base range"
=> all maps too small
additional NCS problem: the subscan length (= map length in the scanning dir.) is always shorter
than commanded by vAz/freqSlowLoop, i.e. 10 to 20asec for maps with NIKA, run #2
fig.15 circles: the commanded coords, dots: coords interpolated to the time stamps of NIKA
