NAOMI Troubleshooter
- Problems which used to require NaomiRestart
- GlasRestart
- Spots on LGS WFS confined to one corner of cells, even though
LGS well-centred in 8-arcsec aperture.
- Either (1) misalignment, or (2) range-gate incorrectly set
(Sep 2008).
- All entries on glas mechs gui turn red.
- In a sextans xterm, type startglascomms
- Repeated GP error messages from any application.
- Try a GlasRestart (advice from Jure).
- TopGui very slow to respond.
- Kill duplicate processes, restart topgui
(type: ps -ef | grep TopGui; kill XXX; TopGuiGlas &).
Try running oasisgui on lpss94 instead of sextans.
- Beam dump can't be taken out while parked at zenith.
- There could be a discrepancy with the TCS (if telescope in ENG mode?),
meaning that the laser safety system won't permit
the beam dump to be moved out.???
- Acqtool can't move the pickoff probe.
- Ensure that the NAOMIHOST environmental variable on taurus has
been set to 'sextans.ing.iac.es'.
Check that the EPMDataServer is running on sextans.
- TopGui won't come up in LGS mode
- GlasRestart
- When BLT tracking switched on, it takes several minutes to initialise.
Then, once the TCS is tracking and the LGS loop is closed,
the status of the BLT remains as TCS tracking
not LSM Offloading.
In addition the LSM plots show static DAC values, while the plots
from TopGui indicate that the LSM DAC values were changing (from DR 18195).
- Probably EPM needs restarting; try a GlasRestart (and report the fault).
- Serious GLAS RTC failure
-
Rene's notes 26/8/07:
- kill TopGuiGlas process
- GlasRestart
- Look out for error messages of scripts that fail to load. Currently three
fail always (this will be fixed; is easy).
- To load the three failing scripts, enter the Python script names in
sextans xterms:
Sequencer2.py (this should be run from a separate terminal)
SubscribeBridge.py (this and PublishBridge.py can be run from the same
xterm)
PublishBridge.py
- WFS SetGlasDefaults to reload defaults
- TopGuiGlas to run up TopGui
- on CPU page, close the strain gauge loop
- on CPU page, start LGS WFS
- If there is a problem with the framing it may
be necessary to re-init the
LGS WFS on the SDSU page. However do not do this by default.
- on mirror control page, LOAD last flat
- on real time page, LOAD NAOMI reconstructor
- on WFS page, load offsets
- on SDSU page, set integration time
- on an Xterm type: CameraSwitch.py and in the box that appears click
'slave'.
Check that slaveSynch is set as the camera.
Then exit that box.
- on TopGuiGlas activate continuous readout and adjust display control if
necessary.
See Nigel's list of restart procedures in green NAOMI folder.
- TopGui freezes or crashes (common problem)
- Double-click on xkill icon, move cursor to TopGui window.
Wait a few sec until the cursor symbol is replaced by a square,
then left click. The frozen TopGui window will vanish.
- TopGui won't restart, or components of it won't
respond in the usual way
- Type NaomiRestart (in an lpss42 xterm). Sometimes it's necessary to issue
it twice for all faults to be cleared.
Don't include an '&' after NaomiRestart!
- TopGui freezes while guiding.
- Don't panic, it's probably still correcting (but the autoguide
signal is not being sent to the TCS).
First type c40CommunicationsRestart.
TopGui will disappear. Type TopGui & to recover it, and the loop
should still be closed. Open the loop immediately, stop the
autoguider (ctrl/c to get rid of window if necessary),
start autoguider, close the loop.
- Error in getting status from Binky.
- Ignore it.
- 'Alignment' and 'Observation' flagged red at bottom left of top gui.
- Click on 'show details', but usually it's something long-fixed, but
which can't be unflagged.
- Loss of c40 communications
- Type ?CommunicationsRestart? as naomi@aocontrol1
- Sequencer dies
- Type: Sequencer2.py
- Suspect communications
- ping scorpio, aocontrol1, fisba, naomisdsu, naomiomc, then
c40Ping 1, c40Ping 2 etc. c40Ping 0 checks contact with the
communications process running on aocontrol1.
WFS Status is useful for checking the status of all the C40s.
Check network connections (usually ports 5 6 7 8 10 in wall
behind control cabinet) by plugging in a portable PC.
fisba is only pingable when plugged into its own port
(reason unknown).
naomiomc and naomisdsu are VMEs (no disks) so are not pingable
unless aocontrol1 is.
C40s 0-8 are WFS, 9-16 are SG
If naomiomc and naomisdsu won't ping from navis,
after aocontrol1 reboot, try powering them off and on.
- aocontrol1 not responding
- May happen if not been properly shut down.
To check state, plug VT220 into RS232 (need to check parity),
look at messages on screen.
If looks bad, try pressing the reset button on the aocontrol1 card.
To power cycle, rsh -l root aocontrol 1 (password in the back of the
NAOMI logbook), type shutdown now (i.e. 2 words).
Wait a few sec for shutdown to complete. Power off aocontrol1 using switch
top right on front of aocontrol1 cabinet. aocontrol1 is the left-hand
card in top crate.
Power on again. Ideally, have a VT220 plugged into the socket on the
back of the aocontrol1 box while this happens, to watch the messages
as it reboots. To simply reboot, type reboot on the VT220 keyboard.
3 NaomiRestarts (on navis) are needed after an aocontrol1 reboot
(c40ResetSystem in an aocontrol1 session may also work).
After a power cycle, the C40s may not correctly boot (NAD email 15/7/02),
in which case you need to c40Run with an old configuration, by typing,
at aocontrol1?:
c40Run NaomiInterleaveZeiss
which should do a naomirestart as normal (still needed after
cold start 11/02).
If this doesn't work, worth checking all cable connections (especially
after a NAOMI move).
(When aocontrol1 taken down, no longer necessary to
rlogin aocontrol1 then c40Run NaomiDev.)
Alternative to using vt220 plugged in back of navis:
tip session then use the TIP name as a command-line option to start a tip
session to access the rs232 engineering ports for NAOMI on Portserver 11
(PCM email 9/7/02).
- NAVIS dead
- E.g. after entering password into screen lock, screen goes blank and
won't respond to keyboard or mouse.
May be disk full of fisba logs in /var/adm/wtmpx, call CFG.
-
Alternatively, the AOcontrol1 network connection switch (at the back of
the top crate) may not be switched to BNSC.
In general, if communications problems are suspected, first ping aocontrol1
from the navis prompt to see if navis can talk to aocontrol1.
- NAVIS very slow on startup
- If it takes ~ 10 mins to login and then another ~ 10 mins to start
TopGui, which then freezes, a reboot may be necessary.
Hit the stop/a keys, typoe boot at the go prompt.
If there's a disk problem, navis asks for root password.
To bypass this (and fix from another machine), ctrl/d.
- Can't telnet to navis from SLO
- Telnet not permitted, try rlogin instead
- TopGui comes up as empty square
- Check that logged in as naomi, not root.
- TopGui crashes or freezes or no centroids
-
After TopGui freezes, may still be correcting closed-loop, try:
c40CommunicationsRestart at the aocontrol1 prompt,
TopGui window disappears, run it up again, open the loop,
stop the autoguider, ctrl/c autoguider window if still open,
start autoguider again (IKS 9/03).
Or...
ps -ef | grep TopGui, then kill nnnn, then at navis prompt, TopGui &.
To restart TopGui without touching C40s,
navis promt,
NaomiRestart -slave -noc40Boot.
The -slave parameter ensures that only the machine logged into is affected.
See Nigel Dipper's
list of restart procedures.
- No communication with OMC
- Press reset button on 167 card.
- C40s problem (no response to c40ping 0)
- At navis prompt, WFS Status to see if still framing (frame numbers
will be incrementing).
If C40 comms lost, try c40CommunicationsRestart.
If this doesn't work, NaomiRestart. Don't add '&'!
If this doesn't work, try c40Run NaomiInterleaveZeiss
(NAD email 15/7/02).
- TopGui, IngridAlign no longer recognised
- E.g. on typing TopGui, error message is:
`libspinbox.so.1 open failed, no such file',
or on typing IngridAlign, error message is:
`Python not configured for Tk'.
LD_LIBRARY_PATH should contain /usr/local/lib (for e.g.
libSpinbox.so.1) so change to path
suspected.
Last edit to .postcshrc removed (the simplextex line), logged out,
restarted, but not clear whether change to file, or restart, was
critical factor.
- Top Gui won't come up
- Check that the following processes have all started up on navis
by NaomiRestart
(list provided by Nigel Dipper 6/02):
> ----------------------------------------------------------------
> UID PID PPID C STIME TTY TIME CMD
> eladmin 22934 1 0 04:03:33 ? 0:00 BillDoors -v -l /software/Electra/save/log/BillDoors.log
> eladmin 22945 1 0 04:03:34 ? 0:01 python /software/Electra/bin/Sequencer2.py
> eladmin 22949 1 0 04:03:34 ? 0:00 /bin/sh /software/Electra/bin/TemperatureCompensation.scheduler
> eladmin 22880 1 0 04:02:31 ? 0:00 NameServer -addr navis:2000 -p /software/Electra/save/log/ports.log -V -r /soft
> eladmin 22942 1 0 04:03:34 ? 0:03 python /software/Electra/bin/EPM2.py
> eladmin 22953 1 1 04:03:34 ? 0:32 python /software/Electra/bin/EPMDataServer
NaomiRestart also starts the following processes on aocontrol1:
/software/Electra/bin/BillDoors
/software/Electra/bin/c40Comms
(/software/Electra/bin/BillDoors.py - Different to above. Not much used)
Note that all of the services started by NaomiRestart belong to a user
called 'eladmin'.
- Reconstructor apparently not running
- The button on the front page isn't activated until you click on
the realtime page. It should say 'running', when it is.
The start and stop buttons on the realtime page should work.
WFS GetReconstructorMatrix returns the matrix. If it contains only
zeros, the reconstructor is not running.
-
Information regarding Ao instruments (OASIS, INGRID etc) remain in
red in AO mimic, mainly after ahutdownobssys/startobssys procedure in
taurus.
-
Solution from Sergio 12/12/08:
Restart codeso NBUPDATE server in lpss94 as follows:
1-Log into lpss94 as whtobs
2-type obssys
3- find and kill the previous NBUDPTE instance running
TO@lpss94>ps -ef | grep -i nbupdate
whtobs 20204... java
-DServerRunner=1 -DName=NBUPDATE -classpath
/wht/release/lib/iParameterNot
whtobs 24606 14267 0:00 grep -i nbupdate
TO@lpss94>kill 20204(of course you will obtain a different
number here)
4- start again NBUPDATE component
TO@lpss94>startserver taurus NBUPDATE
- Pupil-view mode on WFS
- Check pupil-view on WFS.
Set camera stage to 'pupil' or 'near_pupil' (usually 'focussed').
Should give superimposed images of lenslets and DM, parallel light
impinging on each lenslet.
- DM not aligned with secondary mirror.
- Check table is not floated.
- WFS not aligned with DM
- 10/02 back end of WFS needed swinging round ~ 4 mm to realign,
to fix half-cell shifts of WFS spots.
- Not enough light on WFS from pinhole
- Check lamp on, beamsplitter in,
attenuator set to minimum, NCU mask out (especially
after power-cycle), correct pinhole in, covers off, filter and
dichroic have sufficient bandwidth.
- Light on WFS but not on dectector
- Check above, plus INGRID filter and pupil wheels datumed,
instrument focus, sensitivity to ambient light.
- Not enough light from star
- Check above, plus clouds, dome position, target coordinates and mag
in relevant band,
telescope focus,
mirror petals open, Nas flat positioned, table down, pinhole
out, beamsplitter out.
If any light at all from star, check out-of-focus image (e.g. if
half of pupil missing, dome is probably in wrong position).
- Too noisy
- Check ambient light (moonlight, dome lights,
GHRIL room lights, monitors, fisba, LEDs),
detector temperature (WFS Peltier and chiller both have to be on),
etector read-speed.
Units on epics guis are mm.
INIT does not move a mechanism, INDEX does (search for datum switch).
- All mechanisms have status unknown
- The fix is to go into the old
eng mechs gui, open e.g. the WFS pickoff page, then close it.
The mechs status on the TopGui pages updates.
- Mechanisms don't move on startup, or stop moving
- Jure advises 5/12/05 that if the WFS mechanisms fail,
check the status of the EPICS control by with the following:
- Open Naomi mechanism control GUI
- Press Naomi systems and WFS components
- Right click on Device displays (right side of Assembly control
GUI) and choose "CCD stage device"
- On the window that appears press "more Status".
- On the new window check for error messages, especially the
"Action state" status.
- Report the findings.
The camera stage now has the encoder enabled but the velocity is reduced
to 5.
Recipe for recovery:
INIT then INDEX from top gui (needed after power-cycling
mechs crate). If this fails, go to
low-level epics gui via the pickoff.dl gui, INIT, START,
INDEX, START.
If INIT, START fails, try INIT, STOP first (NAD 9/02).
For the pickoff x,y and WFS focus, which are combined
under one page, it may be necessary to go another page down to e.g.
move the pickoff mirror in x without the camera stage also being
asked to move.
WFS foreoptics stage has stopped moving in past, because of dirt
on the track.
NB the devices index from each other.
Check NAOMI PSU (below mechs crate) is powered on at back.
If this doesn't work, telnet to naomiomc (no login) and type
reboot 1 , or
power-cycle the crate and initialise the mechanisms.
The WFS carriage is particularly troublesome, summer 2003.
If there are problems, record probe and carriage positions,
from epics screens, and commands/scripts being used, file a
fault report.
- Problems with USP mechs.
- Go into lower-level epics pages with start_epics from SYS prompt.
- Runaway pickoff mirror.
- Stop top gui, restart.
- Lamp too faint or too bright
- From *old* MECHS gui, NCU, calibration lamp, devices, set value
between 0 (faint) and 230 (brightest?).
See plot in AJL's binder for calibration of intensity vs setting.
- Bench not floating properly
- Pump up basketballs, but *not* so hard that they prevent
table setting onto the mounts when it is lowered.
- Most elements can't be moved from gui
- Check `secret switch' at bottom RH corner of back of strain-gauge
power-amplifier rack, which controls power (e.g. left off after move 12/02).
If this doesn't work, try SG ChanPrint
with a few values and check last line (strain-gauge ADC sample).
If these are floating to max (0xffff) or min (0x0) values,
there is a problem with the SG electronics (RM 4/1/03).
- Mirror element oscillating (especially no. 18),
appears as streak instead of spot
- Open and close the SG loop.
- DM appearance doesn't change when load flat or mid-range, but does
change on opening or closing loop.
- naomirestart (synonym for NaomiRestart).
- DM appearance peculiar after naomi restart.
- At navis prompt, TemperatureCompensation.
- Sudden mirror degradation.
- TemperatureCompenstion needed if full NaomiRestart not executed.
- Use AutoRectifyFlat
- DM almost mid-ranges on using AddZernike
- Seen 13/6/03 - DM moved successively closer to mid-range with
each AddZernike. Reloaded flat, then OK.
- FISBA fringes fuzzy or won't stabilise, mirror won't flatten
- Check gain ~ 0.8, not e.g. 0.000001.
There must be almost no fringe motion(< 1 fringe per 5 sec)
for the laser-flattening
to work. If there is more than ~ 1 fringe wander per 5 sec,
laser-flattening will probably make things dramatically worse.
Unstable fringes can be due to:
- Table not up, one of legs making contact with table,
telescope oil pumps on, INGRID compressor on in GHRIL
control room, people working on telescope e.g. filling
cryostats.
- Convection
(in the test focal station, check the plastic sheet is protecting
the bench from air currents). If at GHRIL, try rotating the
WHT in azimuth to reduce prevailing draughts.
- Slow oscillation of the table
(the beach balls need pumping up).
- Laser still warming up (intensity rises and falls), usually
takes about < 30 minutes, but can take > 6 hours.
At the end of 2001, the fringes drifted continuously for days
before finally settling down. Seems OK now (6/02).
If laser-flattening still giving poor results, try reducing
the x,y,z gains. Or put in pinhole, bright lamps, zero WFS offsets,
AutoCentre, close WFS loop, open loop WITHOUT reloading flat to
leave last closed-loop demand, SG GrabMirror.
- FISBA ftp dies, or display won't come up from eng gui
- Click on the AIPC9 icon on aodisplay screen 1, to bring up the
PC Anywhere screen. Click on file, then exit, to shut down the
fisba interface. When the windows screen comes up, click on the
microshape icon to restart the fisba interface.
(Old way, go up to GRACE to restart it.)
The click on the button at the top left of the interferometer display,
and slect zoom +3.
If it won't reoot satisfactorily, delete all windows,
remove sessions on bar at the bottom of the display,
double click on microshape icon.
If the above fails, power-cycle the cpu.
- FISBA image degraded
-
The FISBA image may be degraded by dirt in the optics e.g.
on the beamsplitter or CCD (or, 6/01, by a lens falling out and
leaving a lump of glue close to the focal plane).
The FISBA optics can be removed from the back of the lens unit
and blown clean.
(NB hatching on the displayed image indicates saturation,
turn down the intensity.)
- FISBA fails with memory error
- Check autoguider not running.
- FISBA misaligned
- The fring-analysis software can cope with x,y shifts and changes
of scale as long as the red crosses are realigned (set display
to contiguous), but probably not with rotations of more than
~ 2 deg (which might be a problem > 11/05, after the installation
of 2 fold flats in the light path).
Procedure for power-cycling the SDSUs and camera controllers is:
switch camera controllers off (bottom of rack),
VME off (above camera controllers), wait a few sec, VME on, hit
reset button, camera controllers on (reinit on the SDSU page may also
be needed).
If the power to the VME has been off for a while, the above may
not work, see
DR 16822 (and Sergio recommends
this document).
To grab a WFS image, try WFS SavePixelsToFits (the topgui button which
foremerly effected this no longer works, 4/10).
- General WFS problems
- At the navis prompt, type WFS Stop, WFS Start, WFS Status.
Always hit the WFS Start button on TopGui as well (Nigel's email
17/7/01).
- TT server not running, results in NGS loop status differing on
top gui and the TT gui.
- In a sextans window, type ttServer in a sextans window in which
obssys has previously been executed.
To check if ttServer is working take a look at the AO
mimic. You should see the value of image background changing about once per
second. Normally ttServer is executed from the GlasRestart script.
(This advice is from Jure.)
- Vignetting, e.g. spots towards top right are fainter than the rest.
- Check that the filter wheel isn't in an error state, by attempting
to change to another filter. If top gui displays an error state,
initialise the filter wheel (bring up the mechs epics gui, then
from that select (right click) the filter-wheel gui).
- Bright stripes (1000s counts) left and right,
high background several 10s counts, bright pixels
- Hydraulic cooling has probably failed, check the flow meter.
May simply need switching off and on again. 5/06 - looks as though
it can't cool both the CCD39 and L3 heads?
- WFS display comes up in bizarre colours.
- Reset the simplex display from rainbow to grey,
and/or kill any ds9 (and maybe netscape) sessions.
- Top spot (starlight) in each column on WFS display vignetted.
- Unfloat the table!
- WFS spots elongated
- ADC may be in wrong position, try moving one ADC to see if
value of the other changes. May need init and index.
- Spots missing along one radial direction.
- Malfunctioning mirror petal.
- Spots for segments 70, 75 are invisible on WFS display, even
though centred on fisba display, but can be made to appear just below
their nominal positions if offset in y.
- CCD39/L3 switch not correctly switched to CCD39 position.
It may feel in the right position, but moving it out and in again
may cure the problem.
- WFS blank even though it is framing
- Hit stop then start on the WFS page.
- WFS quadrants going blank at random
- This happens especially with 1875-microsec
readout, if decimate is set too low (try 20).
With 1250 microsec, it always happens, i.e. this mode can't be used.
- WFS spots are too faint
- Is the bar sticking out of the NAOMI side of the NCU fully out?
- WFS illumination asymmetrical (e.g. left half bright, right half
faint, spots show large rms on average x, y
in WFS align
- Power cycle the SDSUs and camera controllers.
Camera controllers on first, then SDSU?
- WFS centroids jumping all over the place in some quadrants
- Could be background light in corner of quadrant (which affects
subtaction for that quadrant). Investigate by switching off
background subtraction with WFS Testmode??
But sometimes this is not the cause, SDSU reset or power cycle
probably needed.
- Few spots on the WFS, distributed at random
- Check that pickoff mirror is at sensible position, e.g. not 0,0.
- SDSU not responding e.g. won't update exposure time
- Hard reset by pressing reset on VME card. Wait for fault light to
clear. Press reset on BOTH fibre interface cards.
In software, either click on INIT on SDSU page of TopGui, OR
bring up CamVxWorks terminal, gives 'minimal' prompt,
type < startup_cam_no_epics, gives 'naomiCam' prompt,
type < readoutC40_cam1, check framing (with WFS Status),
close camera terminal.
Don't click on 'continuous' before the C40s are framing.
NaomiRestart does NOT start the SDSUs.
This advice is from Nigel Dipper's email 19/6/01.
- < cam1 hangs
- Sometimes takes ~ 2 mins
- There are patches on WFS where the spots are greatly attenuated.
- May be condensation, check the air hose is plugged into the back of
the WFS.
- White-light flat giving rms ~ 1.e-13 instead of ~ 10s,
or fails with divide-by-zero error
- In the interferogram .peakH or .peakV
files, check for lines starting with
#. For these element pairs the interferometry failed. If most
start with #, lamp may simply not have been bright enough, OR,
was executed through a dichroic with insufficient spectral
diversity (use the transparent one).
Probably the wlflat has done nothing at all.
The last column in the .peak* files gives 1000/S:N. If this number
is > 200, data not useful.
Also, the wl flat procedure will yield rms = 0 if the SG loop is left
open.
- White-light flat yielding << 180 measurements.
- Check for vignetting on the WFS, see possible causes in section above.
- To plot white light flat, type mplot, and follow prompts for
interactive or non-interactive use.
- For the latter, select
the required option:
- DAC values a b c printed at each mirror-element position
- a-b, b-c
- a-b, b-c plotted as vector
- mean of a, b and c = piston
- colour representation of tilt, red = a-b, green = c-b
- colour representation of piston, red = +ve, blue = -ve
Set NORM = 1 in the parameter file to plot difference mirror(i)
- mirror(1). The names of the mirror files are given at the
end of the parameter file.
The file mplot.bad lists the elements on pass-through (rendered
hatched on the colour plots).
There are example mplot.par and mplot.bad files in ~/crb.
- To subtract one flat from another (or add), and display
- difflat = ~/bin/DiffFlat or addflat = AddFlat
(the ones in ...appsrc don't work)
- To plot wl flat in 3D
- cdflat, then plotflat3d filename = ~bin/FlatPlotter.py filename
- To plot wl flat such that each element is represented by a column
of asterisks
- plotflathist filename = ~/bin/PlotFlat filename
(NB master copy is in /software/Electra/bin).
- AddZernike apparently doing nothing
- Check correct flat is loaded.
- Plot interferometry from white-light flat
- E.g. if one suspects the white-light flat is not doing anything useful,
try:
wlplot filename.scanH (or .scanV) =
WhiteLightPlot filename.scanH (or .scanV) n1 n2
to look at the intensity vs separation plot for adjacent mirror
elements n1 and n2. If there's no clear peak, something's wrong.
If n1 and n2 are omitted, the procedure plots for all adjacent
element pairs.
On each plot the 7 curves correspond to individual pixels in the
WFS cell. The plot is made twice, first the data then the fit.
If the .residual and .solution files are empty, something's wrong.
If you need to recreate the output, try:
WhiteLightProcedure --existingScan
(RM 20/7/01)
*** wlplot is not working at 8/02 - library missing.
Substitute 5/04: ./scan2txt filename.scanH > junk (RM email 27/5/04).
Then run crb's scanpl programme (edit scanpl.par to give junk as
name of data file) to plot the fringes for all segment pairs.
The .scanH and .scanV files are now in directories e.g.
~naomi/WhiteLight/20050101
- Problems with udas detectors
- Try dasreset in the DAS window.
- INGRID grab stalls
- Often during a simplex, the grab stalls for ~ 1 min, then restarts
of its own accord. If it stalls for longer, try a dasreset
at the INGRID das window (RO 8/02).
- http fails
- dasreset INGRID at SYS prompt.
- Grab won't start - socket error
- Probably another IngridAlign running, kill it.
- INGRID grab fails (this needs rewriting for Craige's new version)
-
- ctrl/c out of navis window which has start-Drama command in it.
- stop-DramaServer.sh (ctrl/c after a few sec if it hasn't returned
the prompt).
- drama-cleanup
- Dserver.csh (return probably needed to get prompt back)
- start-DramaServer.sh (doesn't return a prompt)
- IngridAlign --camera SBIG crashes with dtm error or timeout.
- Check no other SBIG processes running.
Alternatively, the C40s may be down, see above.
- INGRID align gui crashes
- Happens if set zeros in start X or Y at top left of the gui.
Kill window (see below), restart gui.
- INGRID grab slowing down intermittently
- If the INGRID das has been up several days, restart it,
the memory may be getting fragmented.
If this doesn't solve problem, rec 9/02 is to power cycle the
INGRID controller (240V rocker switch off, wait 10 sec, then
switch on).
restart -slave 25/9/02 boosted speed from 1.2 to 3 Hz.
- Simplex slow and unreliable
- Sam and Jure found 7/09:
that this problem only seems to appear when all of the following are true on
INGRID:
* Width and height of the readout window are the same
* They are larger than 32
* They are odd.
Cause unknown.
- Simplex slows to 1 Hz
- Reboot navis. The IngridAlign.py interface also has a memory
leak, best to exit it after each simples, before starting another.
- Simplex yields 2 or 4 symmetrically-disposed spots instead of 1.
- Mirror may be split in piston, e.g. could happen after simplexing
through narrow-band IR filter. Make sure using broad-band filter,
or rerun simplex with more aggressive piston search.
- Simplex good in H or J but not the other
- If some piston errors have crept in, e.g. after several days
without a white-light flat, it's possible for the H simplex to
generate a diff-limited pinhole image, but with some mirror elements
lambda/2 out. On switching to J, the image is badly degraded.
The same will probably happen after simplexing in J and switching
to H. The cure is to white-light flatten, or to simplex
with an aggressive piston limit.
On one occasion (8/02) it appears that this problem may have
arisen
due to poor laser-flattening (because of slow fringe drift,
beach balls needed pumping up), followed by ineffective
white-light flats (gradients ~ 1.e-14 in the x direction)
for reasons unknown.
- Unable to grab.
- Swap power and camera-head cables of two black boxes on top of
right-hand electronics cabinet in GRACE.
See INGRID documentation. NB for pupil imaging, use focus -1200.
-
INGRID controller power cycled
- dasreset ingrid
- Pupil stop needs realigning
- login to crater as whtmimic, or as ingrid
- cd /bin
- start_ingrid, brings up gui
- select pupil stop wheel, CAD record
- action = move, move mode = absolute
- abspos = 197.4, start
- the wheel moves to a position marked invalid (which is OK)
- exit
- To get eng control, take out *left* ribbon cable, yields
'PSU drive on' message on the mimic.
- Response very slow, e.g. when doing hysteresis.
- Check WFS exposure time not set to 50000 (SDSU page).
- ds9 won't run up from inside iraf.
- Try /opt/astro/bin/ds9.
- Autoguider drives straight to limit, e.g. if accidentally close loop
with gains set = 0.
- restart = NaomiRestart.
Autoguider also does this if the AO loop is closed before the
TCS loop, usually on next occasion of opening and closing loop.
- Navis window won't close
- Background menu. right mouse button, select windows, select
destroy, position cursor on window, left mouse button.
- FSM not moving when loop closed
- Check that gain (REALTIME gui) is both set *and* applied.
- Peculiar image on the WFS
- Too faint? Check beamsplitter out, pinhole mask out,
WFS exposure time correct,
telescope not pulled off by autoguider, correct filter in WFS.
- Scale of spots has changed? Probably due to telescope focus error
(incorrect compensation for flexure with elevation).
- Spot array expanding and contracting irregularly?
Probably a strong focus term in the seeing.
-
- Image split on science camera
- may be a reconstructor problem, try turning it off.
For mirror-flat diagnostics (plotting), see above under 'white light
flattening'.
- DM temperature log
- PrintTemperatures 20030715 20030716,
where any pair of dates can be used. Log starts at 00:00 on first
date. Only works for 48 hours at the moment?
Alternatively, use SJG's /software/Electra/log/Temperature//TempLog
(see his email 2/12/03)
- Status of WFS
- At navis prompt, type WFS Status, for status of the 8 cpus.
- Strain-gauge pass-thru
- At the navis prompt, SG PrintPass to see which strain gauges are
on pass-thru mode. The result is labelled 'cpu11', but the table is
the same for all cpus.
- What's using the cpu?
- At a navis prompt, type top.
- Help on commands
- At the navis prompt, type WFS or SG to see lists of very-long-name
commands.
- Using JOSE as an imager
- Lenslet array can be unscrewed from reimaging lens. If removed,
provides reasonably in-focus images at 0.25 "/pixel.
- Inject tip-tilt noise into FSM
- Unit in GHRIL control room?
- Temperature logging
- SGlogData /software/Electra/save/TemperatureLog
- Power spectrum of FSM trace
- Note that this takes several sec to appear, and is not autoscaled,
you may need to change y scale.
ou need to login to aocontrol1 and issue the command
- DM temperature logging
- See RM's email 24/6/02 for location and format of DM temperature
logging.
- To grab WFS image
- WFStoFITS at command line. Directory name given at front of this
document.
- To GRAB WFS centroid data
- login to aocontrol1 and issue the command
diagDowload -dma
- To get true exposure time on WFS
- EstimateFrameInterval
|