Pointing model 19, oct-Nov17
Model 19 was installed 30nov17 17:00
Links to plots:
Taking the data:
swings using the tilt sensors
Daily plots checking the
data used to compute model 19 (.ps) (.pdf)
Fitting the model:
19 fit with residuals (.ps) (.pdf)
model by removing a source at a time and recomputing the model
azimuth encoder table results (.ps) (.pdf)
Measuring the other receiver offsets:
checking the data quality
pointing errors vs azimuth (.ps) (.pdf)
pointing errors vs zenith angle (.ps) (.pdf)
Links to sections:
Introduction/ sequence of events
Taking the input
data for the model
Daily checks of the input
Data used to compute the
Fitting the model.
the validity of the model.
Measuring the constant offsets for
the "other" receivers.
Introduction/ sequence of events (top)
- Hurricane maria hit AO on 20sep17 with high winds and > 34
inches of water (in 1 day).
- The linefeed fell into the dish, damaging panels and
changing the weight distribution on the azimuth arm.
- on 27oct17 1.4 kips was added to the lead ballast at the
end of the azimuth arm to compensate for the missing
- calibration scans done immediately after maria (without
tiedown control) showed problems with the pointing.
- The last model update was feb 2012 after the platform beam
repair, so a new model was overdue.
- The platform tilt was checked using the tilt sensors
and azimuth swings.
- Calibration data was taken, but not used since the tiedown
tracking was not yet working
- Calibration data used to make the model was taken
- sbn tracked 6 sources with the new model (model19)
- the other receivers (327,430,lbw,sbw,sbh,cb,xb,alfa)
tracked 1 or more of the 7 sources using model19.
- This was used to measure the constant offset between sbn and
the other receivers.
- 01dec17 installed model
- 04,05,06,07 added B1005+077. tracked with sbn,
- 09dec17 - updated 430,cb,xb to offsets to include 04-07
- but they where added incorrectly. 4-7dec was taken with
model19, not the model19verify.
- these had different offsets.. mainly affected 430.
- 11dec17: added 05-07 offsets correctly and
- Note.. some of the text does not include the sources
04-07dec17..never got around to updating it.
input data for the model:(top)
- Before taking the data for the model, the platform tilt was
- If one or more tiedown cable lengths changed, then a tilt
sensor azimuth swing would have a residual 1 azimuth term.
- The measured 1 azimuth term amplitude was <= .0021
degrees. This is a small value.
- The tiedown reference positions were left unchanged.
- The data for model19 data was taken with model18A
- Each day the pattern data points were examined (see daily plots)
- A pattern was excluded if any of the tiedowns lost tension
during the pattern.
- There was a problem with this test 01-14Novnov17:
- tiedown 8 had 1 working load cell, and it had an offset of
- When the loadcell read 0kips, there was still 8 kips on
- Instead of throwing out all td8 points with 0 tension, i
kept those points when the tension 1st read 0. I also looked
at the pointing errors throwing out any that showed large
- This may have included some points with no tension.
- We ended up with 435 usable spider scan patterns.
Daily checks of the data quality. (top)
Bad data could be caused by no tension in a
tiedown , or the platform out of position. For each day, plots
were made to find any points that needed to be excluded(not
tiedown tension, or platform way out of position).
These plots included:
- Page 1: az,za errors for entire night
- top: za error vs pattern number (there is a separate pattern
number for each frequency band of a cross).
- each source is plotted with a different color/symbol
- red + are plotted where the td tension was less than 1
- The vertical dashed lines show the patterns that were
- 2nd frame: az error vs pattern number
- 3rd frame: tiedown tension vs patten number. black: td12,
red: td4, green: td8.
- The purple * occur where one of the td tensions was less
than 1 kip. The dashed vertical lines were the patterns
- bottom: platform vertical offset (in cm) from the nominal
- dropouts are where the distomats failed to read
- Page 2-N. az,za errors vs za are plotted for each
- + (black) are the za errors, * (blue) are the az errors.
- The red + are the points that were excluded.
processing: x101/model/oct17/procmm/doyymmdd.pro, procfiles.pro
The data used to compute model 19. (top)
The data used
to compute model 19 (.ps) (.pdf)
was taken using model18A (the previous model). Figures 1-5
show the pointing errors relative to the previous model18.. Figures
6 and 7 remove the model 18 correction and show the raw telescope
pointing error. All errors are great circle arc seconds.
- Fig. 1 is the azimuth/zenith angle coverage.
- most models have strips spaced every 2 degrees in dec. We
did not have complete coverage for model19.
- Fig. 2 az,za errors vs azimuth
- This is relative to model 18.
- The left half of each plot is the northern portion of the
dish (southern sources with declination < 18.2 degrees).
- The right half of each plot is the southern portion of the
dish (northern sources).
- You can see a large offset in the za error vs azimuth. Most
likely a 1 azimuth term.
- Figure 3 az,za errors vs za.
- A linear ramp in za error vs za was fit to the data :
zaErr(za)= 7.2 + .063*za [asecs](-1.6 + .28*za).
- This change could be from a change in the weight of the
dome... but my guess is it is a 1 az term (see later plots)
- Figure 4 az,za errors by source
- The sources are color coded.
- This is used to check for source position offsets.
- Fig. 5 arrow plot of total errors versus azimuth and za.
- The length of the arrow is the magnitude of the pointing
- The angle of the arrow is the direction (radial errors are
za only errors)
- 1 tick mark on the plot is 5 arc seconds.
- At the bottom is a table of the average magnitude and
rms for the entire dish and computed for every 5 degrees in
- the northwest quadrant (upper left) is the only one with
- Fig. 6 has the raw az, za errors plotted versus azimuth.
- The model 18 correction has been removed. Model 19 will be
fit to this data set.
- Fits to 1az, 2az, and 3az have been over plotted with
the amplitude and phase angle of the maximum.
- The table below compares the values measured for model18
(feb12) and model19 (Nov17).
- Notes on the values:
- with the model removed, these values should be measuring
the underlying errors in the telescope (either structural
or tiedown imbalance).
- the azimuth arm is suspended on the ring girder by the
- the wheels are spaced up to +/- 30 degrees from
the central azimuth direction.
- so tilts in az or za are spread over an area of about
- model 18. the az,za term have similar amplitudes and
they differ by 90 degrees.
- model 19. The 1 az amplitude for za increased
by 10 asecs. The az amplitude decreased.
- The tiedown offset positions were not modified between
model 18 and 19 (of course the tiedown encoder
position could have changed by a small amount).
- a tiedown change of 1.7 inches (1 inch at platform)
would give a pointing error of about 12 asecs,
- model 18: the amplitude is small. 6-9 asecs
- model 19: the az error amplitude is 25 asecs
- there isn't an obvious 2az symmetry in the
telescope to cause this error,.
- model 18: about 14 asecs with phase angle of 67
degrees for both az and za.
- model 19: the azerr amplitude increased to 47.4 asecs
- Fig. 7 shows the same raw errors plotted versus za.
Fitting the model. (top)
The model is fit to the raw errors.
- I excluded points below 2deg za (they had very large errors)
- I also excluded 4 bad points (error in az or za > 35 asecs)
An encoder table is computed using the pointing
error residuals after the model is applied:
- The encoder table is spaced every .5 degrees in za
- At each point a separate value is computed for the azimuth and
za residual error.
- The encoder table is then removed from the model residuals.
- The final residuals are great circle errors.
- To point the telescope to the source, the model
correction/encoder table is added to the computed position. .
The model 19
fit with residuals (.ps) (.pdf)
|mod 19 with Enc Table
- Fig. 1:the error residuals versus za for the azimuth and za
- Raw errors - model is plotted (the encoder table has not
- The computed encoder table is over plotted in red.
- Fig. 2: the error residuals (with encoder table) vs za for
the az and za errors.
- raw Errors - ( model + encoderTable) ) is plotted
- Fig. 3: the error residuals (with encoder table) plotted vs
- There is more scatter in the azimuth residuals than
the za. This could come from:
- The tilt
sensor measurements showed a 6az term over part of
- The encoder rack gear for the azimuth also has some
runout. This will cause azimuth scatter (with a za
dependence since these are great circle errors).
- Fig. 4: the error residuals (with encoder table) by source.
- B2337+227 (light blue) has larger az residuals.
- Fig. 5: the error residuals (with encoder table) by source
processing: x101/model/oct17/domodel.pro, pltmodel.pro
- Fig. 6: arrow plot of residuals (with encoder table) vs
- The length of the arrow is the magnitude of the residual
error ( 1 tick mark is 5 arc seconds).
- The arrow direction is the direction of the error (radial
arrows are purely za errors).
- A table of the average error and the errors every 5 degrees
za is at the bottom of the plot.
- The model parameters and values are printed in the lower
Checking the validity
of the model. (top)
The raw errors (model18a removed) for the 33 sources
For each of the 33 sources:
source at a time from the data set and recomputing the model (.ps)
- The raw errors for the source are removed (leaving 32 sources)
- The model is recomputed using the 32 sources.
- This is repeated for each of the 33 sources.
Fig 1 has the model residuals removing one source at a time.
0 is the model without B2128+048, 1 is the model
without B2338+132.. to 25=J1042+120
Figure 2 plots the mean pointing error and its rms for each
source track that was not included in the model.
The black line is the total rms residuals while the red is
the azimuth and the green is the zenith angle.
The top plot does not include the encoder table while the
bottom plot includes it.
Removing the 2nd (count from 0) source B0116+082 makes
the largest improvement in the model. Looking at figure
6 from the previous section it looks like the last 6
points before setting for this source had large errors.
The model was evaluated without source i
then the mean and rms of the pointing model along the az,za
track for source i was computed.
An azimuth encoder table for azimuth residuals
was built by smoothing the great circle azimuth residuals in azimuth
and then removing this from the (model-zaEncTbl) azimuth residuals.
I first tried smoothing the little circle errors (azErr/sinza)
thinking that the azimuth encoder wrack gear was the largest culprit
and it should give a little circle error. The residuals didn't get
much better. The low za errors were messing up the averages. This
must mean that the azimuth residual errors are great circle and not
The table step has 1 degree steps in azimuth.
Different az smoothing was tried.
encoder table results (.ps) (.pdf)
are shown in the figure:
- Fig 1 top is the azimuth encoder table made by smoothing to 1
through 6 degrees azimuth (bottom to top)
- Fig 1 bottom plots the azimuth encoder residuals (black line)
for azimuth smoothing 1 through 19 degrees. The green line is
the azimuth residuals without the azimuth encoder table. The red
line is the total residuals (za plus az) for the various
- Using an azimuth encoder table smoothed to 10deg za
would decease the az error from 4.8asecs to 3.9 asecs.
- Fig 2 over plots the azimuth residuals and the az enctable
smoothed to 3 and 6 degrees azimuth.
- Fig3 is a Fourier transform of the azimuth encoder table
(built with 1 degree smoothing). The top plot is plotted versus
cycles and the bottom plot versus period (in degrees). The power
is at 4 ,8, and 12 cycles (90,45,and 30 degree spacing). 2,3,
and 6 cycle terms were removed by the model. (I think the az
encoder rack gear has 15 degree sections...see
az rack gear)
The azimuth encoder table has not been included in the final
output model ...
processing: x101/model/oct17/mkazenctbl.pro, pltazenctbl.pro
Measuring the constant
offsets between sbn and other receivers. (top)
The model includes constant terms (great
circle) in azimuth and zenith angle for each receiver. These terms
can differ receiver to receiver because of positioning error of
the horn on the rotary floor. The model is made with sband narrow.
After the model we need to compute what the constant offsets for
the other receivers are (ideally it would remain constant between
models if nothing is moved).
- On 19,21Nov17 6 sources were tracked using the sbn
receiver and model19 installed
- On 22,25Nov17 these same sources were retracked using the
other receivers (cbw was not used).
The table below shows the sources tracked on each day.
- The daily plots show:
- the pointing error all sources, by pattern number
- tiedown tensions
- and platform height (to decide if the points should be used
for the verify).
- plots of az,za error by source
sets used for verify, by date:
|rcvrs and sources
|sbn verify sources.
|sbn verify sources.
The tracking error for each receiver
is plotted below:
Other receiver pointing errors vs azimuth
Other receiver pointing errors vs zenith
angle (.ps) (.pdf)
(1 receiver per page.. 8 pages) with the sbn error for that source
over plotted in black.
- When computing the receiver offset, I excluded all
points below za=3 degrees (since the azimuth is moving pretty
fast and the pointing error can be large)
- The plots have 1 receiver per page.
- On each page the sbn errors for this source are over plotted
- The red,green, blue, purple colors are different frequencies
or bandwidths used for the observation
- the black plot is the error measured by sbn.
- For each band the offset is computed (median(sbnErr) -
median(RcvrErr)) and then printed.
- If 1 source tracked, then the az error is on the top
,and the za error is on the bottom.
- If multiple sources are tracked for the same receiver then
az Errors are on the left and za errors are on the left.]
- Page 1: 327 rcvr, 3 frequency/bw bands
- B1040+123. first 4 az error points have large pnt error
slope relative to sbn run
- Page 2: 430 rcvr. 3 frequency/bw bands
- B1040+123. only 2 points measured. The az error has same
large slope (relative to sbn) that was seen in the 327
measurement using this source.
- The offset measured is unreliable (but the beam width is
large... so maybe no too much of a problem).
- Page 3: lbw rcvr. 4 frequency/bw bands
- B2325+26. stable offset measured.
- Page 4: sbw rcvr. 4 frequency/bw bands
- B0758+143. offset not constant for part of
run. probably good to a few arcsecs.
- Page 5: sbh rcvr. 4 frequency/bw bands
- B0758+143. offset good to a few asecs.
- Page 6 : cb rcvr.. 4 frequency/bw bands
- Sources: B0134+329,B0316+162,B0518+165
- This did not give repeatable results.
- az error differences: -3 to +4asecs
- za error differences: -13.5 to -4 asecs.
- Page 7: xb rcvr. 4 frequency/bw bands
- Sources: B0134+329,B0316+162,B0518+165
- This did not give repeatable results. The xb za error went
from way above sbn to way below sbn rise to set.
- the xband beam is not very well formed...
- Page 8: alfa rcvr. 1 frequency band, beam 0.
- the offset is relatively stable.
- there is a 24 asec offset in za.
- page 9
- For each receiver:
- plot the median difference (median(sbnErr)
-median(rcvErr) for each band measured.
- If multiple sources tracked, over plot the errors in red
- + are the az errors
- * are the za errors.
Ideally the (sbn-rcv) value should be the same
for all sources and frequencies of a receiver.
The offsets for the individual receivers as
calculated from the above data is shown in the table below.
model 19 receiver offsets.
||za offsets asecs
The online model files are in /home/online/vw/etc/Pnt/
- The new model is model19A
- enctbl19A is the encoder table for model19A
- While the model is being made and verified, the subdirectories
in etc/Pnt used are:
- OldModel18 - this holds all of the old model18 models
- NewModel19 - this holds the final version of model19
- VerModel19 - this has the models used to measure the offset
between sbn and the other receivers.
- modelSB is the model19A model
- All the the other receiver models are the same as
- the offset for each receiver is computed using modelSB
and making the offset term difference between sbn and the
other receiver the same as model18
- Scripts to install the model:
- The pointing program uses the models in
- There are 3 scripts in etc/Pnt that copy models from a disc
location to the active location (vw/etc/Pnt)
- getModelsNew.sc - copies the final model19 to etc/Pnt from
- getModelsOld.sc - restores model18 to etc/Pnt
- getModelsVerify.sc - copies the models used for the verify
run to etc/Pnt
- Running any of these scripts in /home/online/vw/etc/Pnt will
execute their function.
- model19A was made 01Nov17 through 25Nov17,
- It was installed as the current model 30nov17 17:00
- 33 sources were tracked yielding 435 points (calibration
- We did not have a complete declination coverage for the
- The sbn pointing error residuals were:
|model 19 noEncTable
|model 19 with EncTable
- The other receiver constant offset measurement:
- The sbn run on 21Nov17 had trouble with some sources (lost
distomats, for awhile) (3C48)
- cb,xb used this source for 1 of the 3 sources.
- The sbn pointing errors for these sources was:
- So the errors were reasonable (these sources were used
in making the model).
- The offsets for 327,lbw,sbw,sbh and alfa were measured
- The za offset for alfa changed by 24 asecs.
- The error may have been present in alfa for awhile (we
hadn't done an alfa pointing run for over a year).
- The 430 offset only had 2 common points with sbn.
- The 430 za offset difference had a large change with
- The 327 receiver saw this same pattern (since it used the
- It may be that the sbn track of this source rising, had
- 3 sources were tracked: 3C48,CTA21,3C138
- The za error offsets varied -5 to -13 asecs.
- 3 sources were tracked: 3C48,CTA21,3C138
- the az offsets varied -4 to +5 asecs,
- the za offsets varied 2 to 9 asecs.
- the beam was not well defined. the 1st side lobe got up to
-10db or worse.