ChangeLog 1999
Jan 99
Tried to implement 2 bcard DSP working with link ports. However,
ran into hardware problems with make this mode run slower that
the 1 DSP mode. Gave up on this mode & removed the code.
Working on understanding the animatic motor in hilo.
Define Movie stream Mode for bigdog - will free run clock DSP, and
bufferDSP/host will acquire and save images as best it can.
Feb 99
Returned animatic motor to irlab
DV improvments - Added 1%scaling and option for reference for origin.
Spex UltraSparc arrived - buildup solaris 2.6 system for spex.
Did some restructing of DSP code in preparation of movie mode.
Did some restructing of dv code to support 24 bit color displays.
Many changes needed to support IC on both SPARC564 and USPIIi.
Need new sharc device driver for USPIIi and workaround for buggie vme driver.
Completed bigdog conversion to USPIIi. - Bigdog now running on USPIIi.
Installed Bigdog XUI/IC/DV Version 99.0225 on lab system (USPIIi).
March 99
Upgraded the SPARC564 from Solaris 2.5.1. to 2.6.0. This required
a rewrite of the sharc device drivers.
Added 24/32bits support to DV. All graphics modes work in TrueColor.
Have established that MPR are working on guidedog. Need to install P2 hardware in bigdog.
Waiting for 2nd electronic set & development hardware needed for littledog
applications developement.
Removed CBMODE_MUX as a clocking /buffer mode option.
Starting adding structure for FastGuide & GuideMovie in DSP/IC/XUI.
clocking pattern changes:
ConvertTicks from 3 to 2 to increase speed on GuideDog.
Increased FatTicks for D16 (3 to 4), and S32 (16 to 17).
Installed P2 MPR hardware modules in bigdog.
Reconfigured DSP boards so each IXZ1 board has 1 Mword in ixr1 and ixr2.
Fixed minor bugs:
image_number didn't update during cycles
XUI Macro Function Buttons didn't work.
Installed version 99.03 on 3/29.
April 99
Able to get GuideMovie clocking half to work, but ran into MPR problem
when working the the buffer card code. After some testing it seem our
MPR signal & states are not working on our DSP boards. GuideDog's DSP is not
working at all, and bigdogm is flakey. Trying to work with ixthos to solve this
problem, but it is taking lots of time. Progress is slow...MPRs are needed
for Movie and GuideModes.
I would like to change our GUI libraries from Motif to GTK+. Install GTK on
duke and started learning it. Would like to rewrite all X apps to use GTK.
LittleDog work: Installed the newest software developed when the motor
test wheel was in hilo. Will need external hardware (motors/sensors) for
further development.
Decided S32 bbmode will only be supported in basic mode.
while wait to resolve ixthos/MPR problem, we move the 'good' dsp
board to guidedog to allow development of movie mode to continue.
Development of GuideDog Movie - single buffer mode continues...
04/25:
On the advice of ixthos Technical support (Tom Phelps), we reburn the
Xilinx (40300301.hex) on bigdog ixz4 clock board. I also reburn the
root controller firmwware (rc4h4200.ldr). After this signal
started to work. States are still not functioning. So far we just
need signal for movie mode.
04/27:
Able to get move mode working on guidedog in double buffer mode.
Performance numbers:
DMA DSP->SPARC 4MB in ~0.210 sec or avg of 19MB/s.
SCSI Sparc->disk (with formatting overhead) 4MB in 1.4sec or 2.5 to 3 MB/s.
So, each movie frame can be saved in about 2 seconds or so.
Transistion between frames:
A gap can occur between frame. Are the number of dropped frames acceptable?.
Some test number on little dog:
96x96 at .100 sec: 0 missed after 4 .movie files (113 frame/movie).
96x96 at .050 sec: 2 missed after 7 .movie files (113 frame/movie).
Takes about 2 seconds for sparc to vme/save data. So in general array
size & itime should be limited to:
(w*h) / (1024*1024) * itime <= 2 or (w*h)/524288 <= itime.
Some itimes for example array sizes:
512x512 0.5
256x256 0.125
128x128 0.03125
96x 96 0.017578
64x 64 0.007813
32x 32 0.001953
May 99
05/03 - installed spex99.04 as current version.
05/04 - Changed data save in IC so it uses BITPIX=16 if max-min < 65636.
Previously, it only used shorts if ranage within (-32768, 32767).
Also needed to change to DV to support BSCALE & BZERO keyworks.
05/06 - Added WAIT_UNTIL_READY macro in asm bcard code.
Replaced bit tst & if/jump with single if/jump instruction.
We should measure & compare to see if there is any speed improvement.
GuideMovie is not performing well when testing with actually electronic.
I was fooled during development since without electronics the convert
never wait. I need to modify MovieMode so bcard DSP doesn't rely on the
host to start acquiring the next movie by using shared memory flags to
coordinate ixr accesses.
05/10 - modified /dev/sharc to support SHARC_DECINTCNT_CMD. Needed for movie mode.
Added FITS informaition variables/command for Institute, telescope, instrument.
05/13&14 Little motor work (2 day at irlab)
Tested animatics motor /Hall-effect sensor combination.
Tested stepper motor / HeSenors-Home combo.
- works well with 12500 step/rev at vl=10000 for worm/detent wheel (no spring).
Completed 1st release of the ldog programs testsm & testStepper.
These allows me to develop/test Motor related code.
05/17-28 - on national guard active duty.
99/05/28 - Doug summarized the info on the motor see Doug's email message.
Note: this resulted it alot of email questioning some of the text, but no official
replay/correct was made. (So we just we what we want?).
June 99
06/04 - working with the test motor setup. Developed much of the pc58_help.c
and sm_help.c functions and tested with testsm and testStepper.
Ran actually motor with rot_init & rot_pos.
Ran testStepper's FindDet command. Data is in
990604.testStepper.finddet.txt
06/14 - Put stepper code into with osf_init & osf_pos & ran test stepper.
06/16 - Tested osf_init/osf_pos with real device. The osf worm was changed
so I re-ran testStepper's FindDet command. Data is in
details/990616.testStepper.finddet.txt.
990616.testStepper.finddet.txt
Used these numbers for the Osf2pos[] array in spex_def.h
06/17 - Rec'd FocusStage data from Seth (DW's student help). There are some
detailed notes:
Notes.txt.
990617.FocusStage.txt - Original data.
990617.FocusStage2.txt - Modified data.
990617.FocusStageGraph1.jpg - Volts vs. Steps.
990617.FocusStageGraph2.jpg - Steps vs. Volts.
06/21 - Trip to IRLAB 6/21 to 6/22.
Installed a new sbus care with 100MB Ethernet & FW SCSI on guidedog.
Attached 9GB FW-SCSI disk and moved system to new disk. Removed
old disk (from SCSI-2 interface).
Had problems with VME harddisk mounts. Peter will handle mounting
new disk to BigdDog.
Started Web based documents in src/spex/docs to be mirrored on IRTF site.
Compiled and install new version of bigdog & guidedog's IC/XUI (v99.06).
Update /dev/sharc drivers on bigdog & guidedog to insure they are the most current.
Did backup of bigdog & guidedog to tape.
Saw the slit mechnicaism with its spring loaded worm, I don't think it will
work according to Doug's email
message summarizing the motor issues.
06/24 - Tried BigMovie mode (with electronics turned off). The computers seem to work
need to test next time I am in manoa. Hilo link is too slow.
GTK note:
Completed GTK port of many of the tools (ad16, tc330, pc58, sm3410).
Needs a little documentation and visual improvements.
Started port of dv to the GTK toolkit. Building of the GUI interface
is in progress.
- Much work on the GTK version of DV. About 40% of the port is completed.
- Implemented Web based computer documentation under the spex->computers
link. Some documents are posted.
July 99
Development work on porting DV to gtk version occupied most July. As of
7/15 almost all the of functionality of motif DV has been duplicated with
gtk. This port also added these new feature to dv:
- Improved the user interface design (we hope).
- tcs offset dialog window added.
- Option of fix the size of the stats wid and hgt
- Added DV to spex subarray commands.
7/12-13: working in irlab manoa
a. A guidedogm freeze problem has appeared. My theory is that the
new Fast-Wide disk goes to sleep and the cpu can't wake it.
It may not be wise to make this the book disk. Asked Peter to mount
the old SCSI-2 disk. Maybe keeping the SCSI 2 the boot disk and
just use the SCSI-3 as the data disk will make this go away.
Talk to Peter - he will try to get a mount made for 2nd disk.
b. Installed GLIB/GTK 1.2.3 on Bigdog and Guidedog.
This required installation of pearl on the systems.
GLIB/GTK installed _without_ -g option (which is the default).
c. Replace all the little dog GUI tool with their GTK version.
These are named: ad16_tool, pc58_tool, sm3410_tool, tc330_tool.
Put some instruction on the Spex Computing Page.
Added sample to file to the ad16_tool.
d. Added 2 account to LittleDog for rebooting & shutdown the OS.
Use these account as they don't require password.
reboot - will reboot littledog's OS.
shutdown - will shutdown (halt) the OS.
These account _do_not_ require passwords.
e. Did some work on motor code. But will wait until notified
by manoa team to continue testing on motor hardware.
f. Thing to do:
Need to make better instruction on the web for little dog tools.
I can delete the code for motif version of the tools.
look at timer function in tools, as ad16_tool has interval bug.
7/15 - Tony&Miranda has written documentation on the following spex applications:
ad16_tool, pc58_tool, sm3410_tool, and tc330_tool.
They are avaliable off the spex home page.
7/20 - Finishing off DV release 1. Miranda stated user's guide and will
post them off the Spex Computer&Software page.
Started fixing up mako to be extra development host.
Did some work on the littledog motor programs...program ready for testing/use
are dit, slit, gflt, osf.
Sensing of detent io is now being done by host (littledog), needed rewiring
of Home/Dio/Limit for detented axies for better performance.
7/23 - When the dewar was still code, we took some position&voltage data for the
Array Focus Mechanism. Here is a summary of the data & fitted equations to
be used in the afoc control programs.
Visit 9907.arrayFocusData/index.html for details.
7/26 - 7/27 - Tony's Trip to the irlab, Manoa.
- Finished littledog's afoc_init and afoc_pos programs.
- Took some voltage/step data when the dewar was warm to compare it to
the cold data. Results were appended to
Visit 9907.arrayFocusData/index.html
August 99
2nd-6th:
Miranda has stated the port of guidedog XUI
to gtk. This port should occupy most the the first 2 week of August.
1st release of the DV documentation is available on a newly created
DV Home page.
We stopped heavy developemnt on gtk DV as it can replace the motif
version in its current form. But then tony decided to add true color
support.
Tony has ported the ldog mechanism program to run in simulation mode
on venom (linux), and has ran the gxui (in simulation mode) under linux.
A littledog mechanism Tech Document was started to illustrate the motor
hardware and documentation spex's little dog applications. Text was written
we just need to make the figure and drawings.
9th-13th:
Miranda completed port of guidedogxui and bigdogxui to the gtk+ toolkit.
M and T spend much time testing/fixing/tweeking the software in preparation
of the Aug 19 installation. Misc fixed to all programs: xui, ic, dv &
littledog code.
Worked spex documentation: Added figures to the little dog mechanism.doc.
Had the aug 13th, conference call between us (Doug, Darryl, and T) and manoa.
John to issue email to summarize discussion.
- pc58 replacement has arrived. Need to install & test.
- Everyone to prepare for the Sept 10 cold test. Full up test except:
multiplexers used, not real arrays.
no prism; no gratings.
Work slightly interrupted by darrly's move to hilo. Need to give up the computer lab
to doug T for his office.
16th-19th:
Short week due to TechGroup Meeting(17th), Holiday (20th).
This week we out last finishing touches on the XUI & DV. M&T traveled to the irlab
on the 19 to install the version 99.8 on. Summary of what changed:
Replaced motif with GTK+ on DV and XUI.
Numerous visual changes to DV and the XUI (too much to mention/keep track of).
Misc Documentation updates includes: mechanism.doc, testsm, testStepper, DV home page.
LittleDog code has be tweaked to better work with the GUI.
Very little IC work but some changes are included: tcscom, tcshostname commands added;
After the install, here are some is fixes needed for the 99.08 version:
xui path in guidedog (maybe init file?)
blank lines should be treated as comments for macros (xui).
solve dialog to front problem.
guidemovie has some bad frames in movie.
make rotator dialog better.
dv: should the open dialog box reset the default path?
dv: no push command.
dv: solve virtual colour map problem.
23th-27th:
Continue with Spex XUI/DV work (fixes to the 99.08 release).
- dv:added push command.
- dv:added path entry in setup frame.
- xui: more work done on XUI's rotator dialog.
- xui: fixed blank line in macro files bug.
- gtk: able to raise dialog boxes with gtk_window_raise().
Installed these fixes in version 99.08 on 8/25.
September 99
Aug30th - Sept 3th:
New mechanism.doc revision posted (8/31).
Started working on implementing slowguide mode for actual hardware. Miranda
will be porting ldisplay to gtk+. After discussion and sugguestion with
John R & Doug T, more work is need on the slowguide mode. Made the following
changes:
- Added GuideAngle, GuideGain X & Y commands.
- Change guidemode to internal variable control by new commands:
Guide.takesky, Guide.clearsky, Guide.fullimage.
- allow realtime changes to sleep, angle, gain, method while guiding.
- DV changes summary:
+new command 'SPeXCom' communicates a command to the IC if running
on the same host.
+Setting of subarray from DV to the IC is enable via the stats window.
This command support mirroring for bigdog & auto auto adjusts the value
around the readout limitations.
+Active canvas is highlighted on the dpytitle area & clicking on it sets
the active dpy.
7th-10th
More changes:
- Added NOT_BUSY error to all motor command in IC.
- the gtk+ port of ldisplay is ready. But gtk+ doesn't have support
for event loop notification from within a signal. Maybe support in
glib 1.3 - will put on hold until later.
- worked on more documentation stuff: Spex User Guide - started
subarray and motor position sections.
(slight interrupt of spex work on the 9th & 10th. Needed to help make a casspc
replacement disk).
13th-17th
Worked at irlab on 13th/14th to test slowguide mode:
- Basicly it works. Ran using star-on-a-string mechanism provide by
Peter/Doug. There changes were made:
ez_stop, a cycle level stop implemented for basic & slow guider modes.
This is a temperary fix, until a DSP level stop can be implemented.
Changes continue...
DV - added support for custom math routines - implemented smooth command.
IC - added filters to Guide method : peak(smooth) & centrold (filter1 & filter1)
Different algorithms for different conditions.
ldisplay to start/terminate with the IC. Use -t to not run ldisplay.
XUI- misc GUI tweeks.
General: Fixed up makefile on bigdog/guidedog.
20th-24th
Installed new version of software (Version 99.09) for bigdog,
guidedog, and littledog on 9/22.
Make misc update to spex computer documentations.
27th-30th
Installed new version of GTK+ (1.2.5) on duke version 1.2.5.
Reimplement sytles in GTK code (needed 1.2.5 style bug fixes ).
DV changes:
Fat DV option enable by using -F.
Increase number of buffer to 8.
October 99
1st - 8th:
DV work continues:
Asked Miranda to come up with a better File Open/Save dialog box.
Tony fixed the virtual colormap problem and is using the newer
gdk_colormap_alloc_colors() function found in gtk+ 1.2.5.
On the 7th, I worked in thr irlab to test motor software while the dewar
was warm. Tested stuff, and note some items:
- grating home switch was misaligned. Lars fixed it.
- grating speed and (maybe) size incorrect. Need to remeasure.
- Change rotator in to 10000 (from 40000).
- Took note on all kinds of change. Will implement for cold test.
- Next software testing (warm dewar) will be done on oct 20th.
8th - 19th:
Working towards version 99.10 release. Release Date to coincide with the 99.10
cold test of dewar/mux/mechanisms:
Spex software changes(highlights):
- user feedback - added percent done on motor initialization and calibrations.
- continuously sample ad16 voltages and display on XUI's motor details status.
- added engineering password on xui.
- allows device.pos command to work from XUI.
- dv improvements:
- fully implement stats display type.
- new commands: BoxCopy, BoxPeak, BoxCentroid, filter1, filter2.
- dv comes in 2 sizes: small(default), medium, & large.
- Update DV documentation and Home page.
- updated mechanism.doc.
- compile/install gtk 1.2.6 on bigdog & guidedog (and duke & venom).
21st-22nd
- Worked in manoa on Thur/Fri (21/22).
- Installed version 99.10 of spex on bigdog, guidedog, littledog.
- Many fixed duing the 2 days of testing at Manoa. Changes included:
flipping colormap in dv; more header info; tuned motor software to
machine mechanisms (dit), renamed menus entries, and update step
positions.
- installed gtk & dv on rayleigh.
- attached guidedog & bigdog to the fast ethernet
25th-29th
- after the trip work (To be folded into version 99.11 or later).
- change DSP code so bigdogmovie mode works on Quad2 (Upper-Right).
change r1 = fext r0 by 16:16
to r1 = fdep r0 by 0:16
files changed are movie_*.asm in dsp/common.
- Some GTK++ work on the BrowseForPath and Open/Save File Dialog
windows. Adding these feature cause many subtle change in the
code affecting the option menu, dirlist, and update_file_list
code and others. Affected DV and the XUI's.
November 99
1st - 5th:
Spex Cool test was performed. Read John's email on it's status.
Over the weekend I left tc330_tool running on guidedog's
temper controller. This is the raw data 991029.cooldown.log, and
991029.cooldown.pdf is a PDF graph of the data.
Work on the DV open/save dialogs continues....
Tony's Trip to irlab (11/4 & 5th):
1. bigdog and guidedog was fitted into 1 rack by Peter.
2. Install version 99.11 of spex, changes include:
+ Dit menu name fixes.
+ added Calibration feedback on detent wheel.
+ tested dit (couldn't reproduce position failure).
3. Test found the followng problems:
- need TC infor in headers & should state of TC.sim flag.
- check initialization of path entry stuff.
- fix pc58 device driver. Clean IO buffer on shutdown.
- fix dv zoom (only works in X axis).
include new open/save as dialog.
- Rotator - default should be 180 deg. (range 90 to 270).
- Found out that movie mode corrupts the FIFO.
- Test GPS IO with dsp board.
- Update mechanism.doc figures & DV documentation.
8th - 12th:
No Spex progress.
Worked on cshell XUI fixes.
Thur(holiday), Fri(vacation).
I saved John's post cold test email entitled " SpeX progress 11/11/99".
15th - 19th:
Miranda finished the open/save dialog code. her stuff was merged into the dv sources.
Ran the ixthos mprtest program on board 0 of guidedog and bigdog. These
boards passed the mprtest. See details in 991116.mprtest.txt.
Modifying movie mode....
Went over to manoa on the 18/19th to work on spex's movie mode. Improve things a
bits by avoiding reading/writting to the BCARD DSP while the movie GO is
in progress.
Also tested both TPRO GPS board. The DSPs can read the dates from both boards.
This email reports the times for 2 consecutive reads: tpro_email.txt .
Reading the GPS time from the DSP take 2 vme read command (about 77 microseconds
for a A32D32 word read ). Thus a GPS timestamps take 153 mocorsecond in the DSPs.
Called in sick on friday 19th - caught the flu in oahu. MovieMode and GPS support
work is unfinished.
22nd - 26th:
No Spex progress or work:
Tony is on vacation this week;
Miranda is starting the FMPro database project.
John's email about flexure problems and completed mods to cryostat: 991128.john_email.txt.
29th - 30th:
Worked in manoa's IRLAB this week to finish up spex work.
Continued work on movie mode. Changed things so that bcard can control
the clock DSP's using the MPR signals. We need to sample every frame
we clock out to prevent FIFO corruptions. This fix seems to be working:
Ran movies with no dropped frames. However, a stop may require an automatic
DSP reboot if the bcard DSP doesn't exit cleanly, with happen about
50% of the time. This happens because the stop works by writing to
the bcard DSP memory - a bcard NO-NO while data acquistion is going.
Had to remove EVERY read/write to the Bcard DSP to get movie mode to run
without errors (freezes). It is very important to NOT disturb the buffer
DSP while running. Needed to remove the ixr1 & irx2 available DSP flag. The
movie setup should allow for enough processing time for the data DMA and
saving. 2 seconds should be enough.
Fixed some bugs in DV related to the BoxZoom and Image ScrollBars.
Also the Spex MoveRead and MovieShow commands are now working.
Put GPS time infomation in FITS header for basic mode. The clock DSP
save this information just before the reset-clocking starts.
TimeStamped the movie frame using the GPS boards. The time stamp is
obtained by the bcard dsp after it has sampled the last pixel of the frame.
The time info is written in the last pixel location and is extracted by
the sun when saving the FIT movie. A separate .time text file is create
with this timestamp information. (The pixel value in the image is set to 0).
Misc XUI fixes: Added more setup widgets, clear macro text on edit, etc...
Dec 99
1st - 3th:
John's email about flexure mods/results and general schedule to next cooldown: 991201.john_email.txt.
Tested and coded up the CalMir linear motor/drive to little dog. Some specs
obtained from testing:
LED on JBOX: Green LED Red LED
Travel in steps: ~256866 <-----------> 0
Position of Mir: Out In
Motor programs using: accel=50,000, Vel=100,000
The Shaft has 6400 steps/revolutions
The General init procedure is to step the motor in the positive direction
until the limit is hit. The set the setup position to size (256866). Then
move it to the out position. The [Out/In] position as [256000,800].
I asked John R./Vern if the limits define the out/in position: Hopefully they
do not, as the out/in position should be inside the limits. John/Vern could not
answer my question.
Still no information on how we are going to control the lamps?
Changed the SmartMotor code to perform an E0 / E1000,PlastPos on every
move. This will de-power the motors while it is resting. Peter noticed noise from
the motor can be detected on the arrays as the motor are electrically connected to
the cryostat. This affects the Rotator and Grating Wheel.
Misc XUI fixes.
Installed new version (spex.9912) as current on 12/2/99.
Performed backup of BigDog and GuideDog on 12/3/99.
6th - 10th:
John's email about flexure problem's POAproblem's, cooldown postpone until fixes,
testing of the science array: 991206.john_email.txt.
Working on updating much of the documentation. Especially re-doing may of the
diagrams in mechanisms.doc, and the user's guide. Need to convert all of the clariswork
drawing to corelDRAW format, as clariswork can't import into word, scale badly, and
has problems with conversion to PDFs.
Install the libraries to support GTK+ Themes on duke. Ran GTK+ apps using various
themes. The Metal theme seem nice and fast. Thinking about installing this on the
guidedog and bigdog. Using the Metal theme, I'm changing some of the widget on the XUI and DV.
Used check_button for off/on (instead of toggle buttons), and fixing thing so it looks
nicer (more organized).
Installed all libraries to support themes on bigdog and guidedog.
GUI fixes on DV: replace toggle with check boxes, refresh on file dialogs.
SimMode didn't work on bigdog & Guidedog: Fixed.
13th - 17th:
Misc fixes: bfp dialog refresh button & resize policy. DV's default path
will be set when opening & saving FITS fits to match last directory used.
+ version 12.99 was recompiled today: 12/15/99.
Spex Meeting 12/15/99 in Manoa:
+ cal lamp will be controlled by ethernet base power switch. Lars
has a spare unit he will send down from the summit to us to use.
+ Told John R. I will need to obtain volts/step data for afocus during
next cold test.
+ Peter say SM may not power down if not shaft force is present. Will
monitor as it may work better when cooler is on.
Fixed per J.Rayner (testing during cryostat assembly).
12.16: Changed allowable error on SmartMotor from 10 to 15. Since the
grating is much stiffer, their seems to be greater error
(7-11 counts are common) in the grating's SmartMotor positioning.
My email showing debug output on the grating (shows amount of
error on warm initialization): 991216.tony_email.txt.
12.17: New equations for Array focus mechanism were calculated.
For details visit 9912.arrayFocusData/index.html.
This was done on a warm cryostat.
12.17: Recompiled & Reinstall all spex software. (still version 99.12).
20th - 31th:
Spend the xmas & new years week on oahu, mostly on vacation, but did drop in
the irlab for Spex support.
John's email on the lastest flexsure test and 'plan' for this week: 991219.john_email.txt.
Fixed up the Detent Error message. (Needs to be cleared on every move). IC's Pslow task
to monitor DE Bit for error messages.
Some minor html changes on the Facility/spex site.