TGBT14A_912_62 - Testing Bank to Bank variations with noise source
17 and 21 January 2015
17 January 2015
- We cannot use Astrid, so simply use VEGAS and the Archivist
- Use the noise source with an eight-way splitter, and connect up both pols of banks E,F,G,H to one noise source output
- The previous observer was using KFPA, so VEGAS will have been configured for that.
- But, Ray releases a new version of the Manager, so it will loose knowledge of where it was.
- Use VEGAS quick-set to set tp_nocal
- Configure Archivist DAPs manually.
- Set Archivist to above project, source name "KFPA", and 60 sec scan time
- Got scan name wrong - became scan 103. Start time 15:36:37 - but went into _61 directory, so that is ok
- Scan 1: 15:40:14
- This all works fine; samplers are all logged into the /home/gbtdata directory.
- Write Archivist procedure to configure and run Archivist:
[rprestag@titania noiseTests]$ source /home/gbt/gbt.csh
[rprestag@titania noiseTests]$ source /home/gbt/sparrow/sparrow.tcsh
[rprestag@titania noiseTests]$ ipython -i control.py
- Scan 2: started through control.py
- Ray restarts VEGAS - should come back to mode 22, since Coordinator was not restarted.
- Scan 3: junk test of procedures
- Scan 4: start both Archivist and Vegas; vegasdm is reporting ~ -24dB in all channels
- Need to set lbw_gains manually, since I don't have a vegas balance option
- Need to deselect banks A through D, and set up tp_nocal
- Note - all subband frequencies are set differently; need to make these the same
- Scan 5: looks like everything is working correctly.
- Write setSubFreqs() to set all sub-band frequency values to 750000000
- Jason changes 6dB pads to be 3dB pads. measpwr levels are now around -21dB +/- 0.4dB
- Scan 6: Note that sbNsnap values are around -40dB, even though lbw_gain is set to 1024. Why is this?
- VEGAS files seem to have the wrong data in them; Archivist data is fine - what is going on? Try again
- Scan 10: still wrong. Need to set next scan number in coordinator? - Yes
- Scan 10: works correctly now. Now want to plot the data.
- For reasons I don't understand, the sbNsnaps are really low when lbw_gain = 1024. Looks like we need 16384. Set all subbands to this.
- up until now, I had only been configuring banks E and F (typo in control.py script)
- Scan 11: Should now all be set up correctly for all four banks, and correct values of lbw_gain.
- Archivist file is: 2015_01_17_18:39:48.fits VEGAS is: 2015_01_17_18:39:44H.fits
- So, I think I am all configured now for Mode 22. Take a scan with no attentuation, then change the attenuation
- Scan 12. Zero input attenuation. Arch: 2015_01_17_19:23:56.fits VEGAS: 2015_01_17_19:23:53E.fits
- Scan 13. 1dB of attenuation. Changes input power level by almost exactly 1dB. Output snap blocks still look sensible
- Scan 12 0dB
- Scan 13 1dB
- Scan 14 2dB
- Scan 15 5dB
- Switch to Mode 21 (Dana's broken mode). Just change mode number in CLEO screen - Can't! CLEO is broken. Use Dev Explorer
- VEGAS CLEO screen does not reflect this, but VEGASDM does. sb1snap is now -43dB. lbw_gain has gone back to 1024. Set to 16384
- All looks good.
- Scan 16 0dB
- Scan 17 1dB
- Scan 18 2dB
- Scan 19 5dB
- Switch to Mode 11 (A supposed good mode. Shouldn't need to change anything else, but we'll see. Seems ok.
- Scan 20 0dB - note, sb1snap values have become asymmetric: pol1 has moved to ~ 25, and pol2 to ~ -25
- shift is slightly different from bank to bank. Effect is not there in input samplers, and I don't recall seeing it before.
- Scan 21 1dB - bank-to-bank changes in sb1snap ADC peak is much more pronounced.
- Scan 22 2dB - bank-to-bank changes in sb1snap ADC peak is extremely pronounced.
- Scan 23 5dB
- Look at Mode 4 - can I set this up correctly? setMode() seems to have done it - hopefully Valon freqs, etc, are ok.
- filtersnap values are again way too low, ~ 40dB. Set lbw gains to 2**30 = 1073741824. Only comes up to -30dB.
- probably need to set some other value. But I don't know what else to set. Stop here, and concentrate on data analysis.
- With astronomical signals going in, need lbw_gains of 1024. Must be that the noise source cuts off at the frequency of the subband I chose?
- 6pm: Jason configures VEGAS back up as normal. play around with lbw_gains; looks normal again.
21 January 2015
- 9:00am - Jason configures noise source again
- Use Astrid to configure for L-band; Mode 7 - bank_gain_mode7_lband
- Remove A,B,C,D - need A to Quickset phase table
- switch period and integration time set to 1 secs.
- Only one subband in use - set to 400 MHz. Leave it there
BANK lbw_gain
E 874577333
F 834781943
G 842144254
H 848663467
- All input powers around -20dBm, and all filtersnap values about the same.
- scan 30 - mode 7, 0dB input attenuation; all looks correct
- scan 31 - mode 7, 1dB
- scan 32 - mode 7, 2dB
- scan 33 - mode 7, 5dB - Bank E has a 10dB "dropout" to -35dB - run again
- scan 34 - mode 7, 5dB - looks ok.
- 2pm. return to work. Configure for mode 4 through Astrid
- All subbands are set to 750MHz - leave them like that. Input values and balancing looks good.
- Note: G & H pols agree perfectly; E and F differ by ~0.4dB
BANK lbw_gain
E 3695684124
F 3798716948
G 3636967430
H 3784571147
- Integration times below were incorrectly set to 4secs, change to 1 sec
- scan 35 - mode 4, 0dB
- scan 36 - mode 4, 1dB
- 1 sec integrations
- scan 37 - mode 4, 0dB
- scan duration is somehow way too short - aborting due to detection of invalid data. Cycle Off/On
- Now complaining about integration time; set blanking time to 0? can't. Did this affect other scans
- scan 38 - mode 4, 0dB - now seems ok
- scan 39 - mode 4, 1dB
- scan 40 - mode 4, 2dB
- scan 41 - mode 4, 5dB
- Change to mode 3. run bank_gain_mode3_lband - leaves VEGAS in mode 4!! Suddenlt CLEO flipped to Mode 3 - ok?
- In this mode, there are no output samplers, just the input sampler.
- scan 42 - mode 3, 0dB - all seems ok
- scan 43 - mode 3, 1dB
- scan 44 - mode 3, 2dB
- scan 45 - mode 3, 5dB
- all done for today!
--
RichardPrestage - 2015-01-17