from:	 Young, Ken 
       to:	 Mark Gurwell ,
                 Jonathan Weintroub ,
                 Rurik Primiani ,
                 Chunhua Qi ,
                 Jun-Hui Zhao 
     date:	 Mon, Nov 4, 2013 at 10:52 AM
  subject:	 Notes on last weeks SWARM test
mailed-by:	 cfa.harvard.edu

Dear SWARM Data Enthusiasts,

   Last week we got some data from SWARM that could be considered the first, very very preliminary science verification data.   There are many caveats:

1) Only the lower sideband was recovered, and only the
first of the two SWARM chunks was produced.   We stored the same data as both the USB and LSB of chunks s49 and s50.    These chunks have all 16384 channels.

2) The software assumed SWARM was running at the
design-goal speed, but in fact SWARM was running at 1/2 speed.   So the channel widths and total bandwidth
for the SWARM chunks is recorded as a factor of 2 larger than it really was.

3) The calibration of the ADCs was not done, which led to strong spikes every 1024 channels.   One of the things we'll want to look out for in future data sets is how well calibration of the ADCs suppresses these spikes.   In a way it's nice to have this uncalibrated data set to serve as a basis for comparison.

4) The scan times are all messed up.   It's not clear what caused that.   Rurik and I may not be agreeing on time units.

5) The pointing was probably bad during most or all of the data collection.   The legacy correlator was running in parallel with SWARM, so we should be able to at least do relative sensitivity calculations.

6) The atmospheric phase was very unstable.   Phase-only self calibration for the planet data helps a *lot*.

7) The meaning of the amplitude scale of the SWARM data is unknown.   It certainly isn't either correlation coefficient or Jy.

8) The delays were being adjusted as the data was being taken.

9) I've attached a timeline with notes that should be
helpful in deciding which scans to look at if you wish to play with this data.

10) We hope to get much more data, with calibrated ADCs next week!

Timeline attachment:

Time (UT)	Dataset			Scan		Comments
		131031_12:50:16		0-2		This file has nothing useful in it
15:52:40	131031_12:52:45		0-4		Restarted dataCatcher interactively, so I could
							see the debugging output statements.   The
							correlator had not been restarted in wideband
							mode.   There is no SWARM data in this dataset
13:11:38	131031_13:09:04				Correlator restarted in wideband mode, which
							should tell dataCatcher to expext SWARM data
							The last scan in this data file *appears* to
							have SWARM data, but MIR won't plot it.   Things
							may be messed up because the correlator changed
							configurations during the life of this file.
							The one SWARM scan was fake data (sine waves
							anyway).
		131031_14:13:57		0-2		This file contains a few scans of SWARM data from
							the noise source.   The channels are out-order.
							The file is small because autocorrelations were being
							sent to dataCatcher, annd that caused it to
							die with a segmentation fault.
15:04:01                                                Integration time changed to 4 seconds
							Source is 3C84
		131031_15:22:21		0-188		There is probably no good data in this file.
15:39:00						Walsh demodulation turned on.   False digital
							fringes go away.
16:14:48	131031_15:57:35				Moved to Mars
		131031_15:57:35		0-66		It looks like none of the data in this file is good.
		131031_17:12:01         107		This is the first scan with a real astronomical
							detection.   Only baseline 2-5 shows a strong detection,
							and there is a significant delay error.
		131031_17:12:01		126		The delay error on baseline 2-5 is gone on this scan,
							and all later scans.   Scan 126 might have mixed delay
							data, so maybe strating with 127 would be safest.
18:47:59	131031_17:12:01				Ambient load inserted, to check that fringes go away.
18:49:32	131031_17:12:01				Ambient load removed.
19:07:00	131031_17:12:01		132		Slew to Jupiter
		131031_17:12:01		133		The other baselines show fringes now, but with big
							delay errors
19:49:30	131031_17:12:01		188		The delays on antennas 2, 5 and 6 look OK at this
							point.   Only 7-* baselines have delay errors.
19:57:49	131031_17:12:01				Antenna 7 delay fixed.   All delays are OK now.
20:08:00	131031_17:12:01				Slewed to 3C273.   The 3C273 data is probably all good,
							but se tried to ipoint and that didn't work.   It
							didn't work, because ipoint cannot yet control the
							integration time of the SWARM correlator.   Much
							of the 3C273 data will be off-source, do to the
							ipointing.
20:19:40	131031_17:12:01				Slewed to Can A.   All the Cen A data should be good,
							but the pointing will be poor (the data is from long after
							sunrise, and we didn't point after sunrise) and the
							atmospherinc phase stability was poor.   Nonetheless
							Cen A was very clearly detected by the SWARM correlator.