GB . Software . RegressionTest_6_1

Regression Test of M&C 6.1 : February 10, 2006 (08:00 - 20:30)

Jim Braatz

Overview:

These notes describe tests of the M&C 6.1 software on 02/10/06. No major problems were found during the testing, but there are some issues that should be addressed prior to release. Most urgent: we should investigate the Hanning window issue in sdfits/gbtmsfiller.

Due to a UPS problem at the GBT (fixed by Joe and Ray) the tests were a little late starting. Actual testing ran for six hours, a bit shorter than usual. I conducted tests at L-band and K-band, including standard peak and focus observations, spectroscopy, cross-pol observations, and tracking using an ephemeris. Since the GFM spectral line plugin was not identified as ready for release, I did not test it heavily but it was up and running in a window separate from astrid.

The new AIPS++ release was not yet tested. We can do this offline.

Data: /home/gbtdata/TREG_d060210

Items to be addressed by Software IPT

Critical

sdfits Hanning windowing. A single hanning applied to gaussian noise should produce < ~35% noise reduction. The measured value was up to ~70%, indicating the ripple really does need to be removed. Details on this issue will be provided separately.

Important

gfm cannot access data in the archive data area via the "open" menu option. Also, the /home/gbtdata area. Fixed by adding a directory exploration dialog. AmyShelton
Still getting the "no new corrections detected" error
gbtstatus displaying incorrect freq for ACS J8 (?) It should have been identical to J7 for the config I was using, but was not. J7 = 19.2477 J8 = 6.2334
After a peak/focus on Mars, the next scan failed because of a bad focus value. I didn't notice whether the focus was accepted after the bad attempt at scan 63. It should not have been accepted. We should check gfm on scan 63 and see that the heuristics were acceptable. The focus offset calculated by GFM was applied to the antenna. The fit found by GFM v6.1 is the same as GFM v5.7, which at least means that this is not new behavior. AmyShelton
"missing or malformed fits file" error on scan 64 This seems to only have happened at run time. It processes fine after the fact in offline mode. AmyShelton
after switching from a config that uses cross-pol to one that doesn't, it is necessary to explicitly enter spect.crosspol='n' in the config. May cause confusion since most config scripts don't contain this line.
Auto* procedures do not check to see if source is above the horizon, or too near the zenith
When a SB contains only "AutoPeakFocus(location="3C123")" (i.e. no catalog defined) need a better error message.
The entry for "Obs Freq" in the gbtstatus window is confusing to me. It is not the observed frequency. It read 8100 MHz when I was observing at K-band. Also "Center Freq" are really "IF Freq".

Cosmetic

New GFM fonts are attractive, but the labels are often crowded and the plot window does not start large enough to show all 4 peak plots without manually resizing.
AutoPeakFocus will not run as the first SB unless a valid configuration including the active RX has been executed first
gbtstatus scroll area was much larger than the space filled with information. Also sometimes the scroll bar was hidden under the panel below it.
astrid was slow/difficult responding to window resize commands
new astrid window size requirements may present a problem for remote observers with smaller screen space
L-band Tsys = 14.6K seems too small. Tcal problem?
When I choose Cygnus A from fluxcal catalog I get some useful info but it doesn't have units, and some info is not identified. e.g. what is F1, F2? Units for size and flux?

[18:47:11] source : CygnusA
              F1 : 4.7
              F2 : 10.6
        location : J2000 @ (19:59:28.35, 40:44:02.10)
           S1400 : 2730.0
         angsize : 170.0

in gbtstatus, instantaneous wind velocity is not as useful as max wind velocity over some time span, say 1 minute or 10 minutes.
DCR Tsys values sometimes unreliable.

Scan Log for project TREG_d051114

Scan	Source	Description	Comments
			12:00 Joe and Ray here to do version change
			UPS problem at GBT
			L-band
			13:30 Getting started
35-39	auto	AutoPeakFocus	no arguments given for intial test - data fine but gfm requires window resizing to fit all plots
40-43	Cyg A	peak	fluxcaltest ok
44-47	DR21	peak	DR21 on fluxcal list, but doesn't appear to be a v good flux calibrator
48-51	3C48	peak	ok
52	3C48	focus	ok
			started a tip - abort - not a good az. abort was fine - caused no problems.
53	tip	tip	ok
54-58	auto	AutoPeakFocus	testing selection of source near a location - ok
			Amy fixed font size in astrid edit window - great!
59-63	mars	AutoPeakFocus on mars	problem - no source - ephemeris ok?
			fail to AutoPeak on mars - abort - b/c bad focus offset
64-67	mars	AutoPeak
			error processing scan 64: Error processing scan 64 : Missing or malformed Antenna FITS file for scan , 64
69-71	w3oh	FS, 2 IFs, HI & OH	ok - but bad high-freq ripple from lack of hanning
72	w3oh	crosspol 1
73	w3oh	crosspol 2	bad rest freq
74	w3oh	crosspol 2	fixed rest freqs from 1420 to 1420.405
75	w3oh	crosspol 3
76-	now check if going to a NON-crosspol mode correctly configures ACS. NO.
			need to explicitly turn off cross pol
			ACS would not stop properly - needs to be restarted

			run a config with spect.crosspol='n'
78	w3oh	2 IF L-band	turned crosspol off in the config
			K-band
79	mars	AutoPeak	abort
80-83	mars	AutoPeak	good - ephemeris is fine
84-85	n1052	AutoPeak	"no corrections detected" error
86-90	n1052	AutoPeakFocus	ok
91 - 92	n1052	nod ACS, 2 IF, 200 MHz	ok
93		tip	abort detected in middle of tip scan
94		tip	try again
			abort detexted again -- Eric says that scans were being written to wrong directory
95		tip	abort - 3rd time
			Mark & Joe ID problem -- crosstalk between DCR simulator and telescope. not software related.
97	tip		ok
98-102	auto	AutoPeakFocus	using location option - great!
103-104	n1386	SP scans	note that the SP cannot handle dual beam observations
105-116	n1386		ringing problem due to no hanning
117	dummy		ephem test ok
118-122	auto	AutoPeakFocus	test case where source is not found w/in 10d
			It says "may take a few minutes". should take milliseconds! and it seems to take milliseconds.
123	0641-0320		focus on a source by specifying source name
124-8	auto	AutoPeakFocus	autopf on source with location & flux specified
129-	u3193	nod ACS 2 IF	ok

-- JimBraatz - 10 Feb 2006

Topic RegressionTest_6_1_First . { Edit | Attach | Ref-By | Printable | Diffs | r1.3 | > | r1.2 | > | r1.1 | More }

Revision r1.3 - 13 Feb 2006 - 18:16 GMT - AmyShelton Parents: SoftwareReportCentral_IntegrationRegressionTestingReports	Content copyright © 1999-2007 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.


NRAO Home > Green Bank \| Wiki Topic: GB > Software > SoftwareReportCentral_IntegrationRegressionTestingReports > RegressionTest_6_1_First

Changes \| Index \| Contents \| Search \| Statistics \| Go