Optics Calibration

TIP Last Update: JeffMangum - 06 Oct 2006


  • Feed Setting
    • Vertical: 280 \mu m
    • Lateral: 3200 \mu m
  • Subreflector Setting
    • Vertical: 28 \mu m
    • Lateral: 140 \mu m
    • Rotational: 1.7 arcmin


Optics Calibration Memos (ALMA, EVLA, etc.)

Residual Atmospheric Phase Errors and Holography

_From MarkHoldaway _

I've done some calculations of the residual phase errors for holography,
and at this point I need to have some guidance concerning what phase errors
are needed.  It seems that the phase errors are larger than what my gut 
instinct indicates we want, so I suggest that we do holography with several 
passes through the map at shorter integrations -- this helps two fold, in that the residual phase errors are smaller AND the statistical nature of the RESIDUAL phase errors will be nearly random, so we'll gain something like sqrt(N) when we average down
(this will help with the pointing errors as well).

From JeffMangum : This is, in fact, the "preferred" way to do any OTF imaging.

I've looked at doing a 64x64 scan with 0.8s on each point, considering NO
slewing overhead between points on each raster's row, but 3s of overhead
in getting from bore-site to the raster's start, and from raster's end 
back to bore-site -- this is critical  to do, as it provides the phase calibration.

From JeffMangum : Right. Always done this myself. Even used it to track pointing changes and update satellite (LES8) ephemerides.

Atmospheric conditions:  I considered April - August, midnight to 8am
local time, and broke the residual phase errors up by quartile within
the restricted times.  I also transferred teh phase errors to an elevation
of 60 degrees (site test interferometer observes at 36 deg elevation).

For the 64 points per row at 0.8 s (54s cycle time), we have rms
residual phase errors at 90 GHz of:

1st Quartile Phase Errors:  5.8 deg
2nd Quartile Phase Errors:  10.4 deg
3rd Quartile Phase Errors:  21.2 deg

If we shorten the integrations to 0.4 s, (29s cycle time), we have rms 

At 90 GHz:
1st Quartile Phase Errors:  4.1 deg
2nd Quartile Phase Errors:  7.2 deg
3rd Quartile Phase Errors:  14.7 deg

OR - if we take two scans through at 0.4s integrations and average them

At 90 GHz:
1st Quartile Phase Errors:  2.9 deg
2nd Quartile Ehase Errors:  5.1 deg
3rd Quartile Phase Errors: 10.4 deg

I think this is closer to what we want, though we may still be limited
by residual phase errors rather than by thermal noise.  Unless WVR just
plain works exceedingly well (and it may for an observation like this, where
we have a "calibrator" at the same elevation angle as our "target 
observations"), we will probably want to do several very short scans of the raster and 

From JeffMangum : Agreed. At a wavelength of 3mm I think that 2.9 degrees corresponds to about 25 micron in surface RMS. So, it looks to me like indeed we will be limited by the residual phase errors.

If we needed to do better, we'd do 0.2s integrations, 16s cycle
time (for 64 point rasters), run four maps, and average down to get
residual phases of:

1st Quartile     1.5
2nd Quartile    2.5
3rd Quartile    5.1

If there were a compelling reason to get better data, by combining
the best conditions with better technique, you would get the data you 

From JeffMangum : Agreed. This sets the limits for when interferometric holography measurements should be made.

Calibration Plan Section

Masao Saito's comments 2004-08-12...

Jeff and Bryan,

I send some comments on optics calibration. I could not finnish all of

9 Optics Calibration

It may be good to give general and introductory description of optics
issues before subsections start. In the subsections, it is good what
strategy we should take and what sources we should use for the optical
calibration. If you say, focus curve, the measurement actually requires
x,y,and z shifts since the feeds are off-axis. This requires
simultaneous three axis motion should be accurate as good as 5 micron or
so. I'm afraid the antenna technical specification covers only one axis
at one time.

For example,

The ALMA employs offset cassegrain optics in which receivers are not
on-axis. Misalignment of the optical components, primary surface,
subreflector, mirrors, and feeds causes gain degradation, beam
distortion, add noise, or change polarization character. Among them, the
surface error should be minimized at certain elevation where most of
astronomical observations are performed. Celestial holography will be
employed to confirm the surface accuracy followed by primary beam
measurements or efficiency measurements. 

Further, unstable optical components bring the unwanted path length
change. The polarization character of the ALMA antennas also depends on
the optical configuration. Therefore, it is essential to calibrate the
optical system, namely to reduce the amount of misalignment.

The errors in the optics are time dependent or not, are systematic or
random, elevation dependent or not, and frequency dependent or not.
Systematic error components should be reduced as much as  possible to
avoid degradation in the interferometric images, particularly mosaicing

Here attention will be mainly paid to surface error, misalignment, and
impact on other calibration issues. 

Most of these are common for both the 12m and the 7m antennas although
some are quantitatively different. For example, positioning the
subreflector of the 7m antennas will be severer because of its short
effective focal distance.

9.1 Main Reflector Surface
The main reflector is measured and set at the OSF using a holography
tower at an elevation of less than 10 degree. The good surface should be
achieved at elevations where astronomical observations are carried out.
The FEM analysis provides a correction table for surface setting between
the holography elevation and reasonable elevation, say 60 degree or so.
In order to verify this approach, we will perform a celestial holography
to measure the surface accuracy at certain elevation.

S/N calculation (time, spatial resolution)
notes:planet unusable (not a uniform disk), polarization of quasars is an issue
      elevation change during measurements

Primary beam measurements with different focus positions may provide
some ideas on global deformation of the primary surface (e.g. Richard
Hills group memo on JCMT).

9.2 Positioning Optical components
Positioning will be done by scanning strong point sources for each band
at different elevation angle and will provide a calibration table as a
function of frequency and elevation. Thermal or wind effect also should
be taken into account when reducing data.

 9.2.1 Subreflector
 9.2.2 Feed
   => ALMA Memo 479
   It's less serious compared to the subreflector in terms of gain. 
 9.2.3 ALMA Memo 395?

9.3. Impact on other issues
 9.3.1 Primary Beam
   => probably covered by Mark Holdaway
 9.3.2 Scattering
   => my concern is that antenna noise changes with elevation.
 9.3.3 Polarization
   => so far, I have not made any progress
 9.3.4 On-Off beam shape in beam Switching and OTF beam
   => single dish issues are also

9.4 Misc
 Since 64 antennas and ACA system are not always simultaneously
operated. These calibration measurements strategy may be different. For
example, ACA celestial holography will need more time given signal to
noise ratio.

Finally, my colleague, Prof. Momose (Ibaragi Univ.) points out the following
thing on the polarization calibration.

p29 - 33 of the calibration plan:

As stated in the calibration plan, the determination of "D" terms 
should be crucial to the polarization imaging.  The phase difference 
between the receiver pair is especially important. The stability of 
"D" term should be tested when the prototype receiver system is 

If the "D" term is fairly stable and essentially shows no time 
variability, one can use a calibrator with unknown polarization 
characteristics (if the gain of each receiver is correctly 
calibrated). In this case it will be a good idea to keep a database 
of the instrumental polarization as a function of frequency, 
elevation angle and so on, as stated in P 32. 

If the "D" term has considerable "irregular" time variability, on the 
other hand, this should be calibrated with a fairly short interval. 
Artificial transmitters or comb generators with tunable polarization 
characteristics may be required, as stated in the calibration plan. 
Another strategy is to maintain database of calibrators whose 
polarization parameters are precisely determined in a timely manner. 
ACA or its 12 m single dishes could be used to maintain such database 
(, although the calibration strategy for the ACA polarization 
observations is also lacking).  

-- JeffMangum - 12 Aug 2004

Topic attachments
I Attachment Action Size Date Who Comment
214e201e254ff0560534c42e2a24bf73.pngpng 214e201e254ff0560534c42e2a24bf73.png manage 361 bytes 2005-04-08 - 08:34 UnknownUser  
4246fb992e4cd2aff1a840c3fb32e68d.pngpng 4246fb992e4cd2aff1a840c3fb32e68d.png manage 413 bytes 2005-04-08 - 08:34 UnknownUser  
746fa280b4bf4f24dee20baf4994764f.pngpng 746fa280b4bf4f24dee20baf4994764f.png manage 368 bytes 2005-04-08 - 08:34 UnknownUser  
ca4008728fc097c08354ad1c1044fdab.pngpng ca4008728fc097c08354ad1c1044fdab.png manage 395 bytes 2005-04-08 - 08:34 UnknownUser  
This topic: ALMA > WebHome > AlmaSci > AlmaCal > CalOpt
Topic revision: 2006-10-06, JeffMangum
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding NRAO Public Wiki? Send feedback