From bec@haystack.mit.edu Tue Jul 20 00:38 MES 1999 Received: from dopey.haystack.edu by kuestner.geod.uni-bonn.de with SMTP (1.37.109.6/16.2) id AA01094; Tue, 20 Jul 99 00:38:31 +0200 Return-Path: Received: (from bec@localhost) by dopey.haystack.edu (8.8.6 (sendmail_886_v2)/8.8.6) id SAA09690; Mon, 19 Jul 1999 18:39:18 -0400 (EDT) From: Brian Corey Message-Id: <199907192239.SAA09690@dopey.haystack.edu> Subject: Re: More about phase cal in MATERA To: petrov@kuestner.geod.uni-bonn.de (Leonid Petrov) Date: Mon, 19 Jul 1999 18:39:18 EDT Cc: bec@haystack.mit.edu, aen@dopey.haystack.edu, colucci@mlrohp1.atsc.allied.com, dbs@kuestner.geod.uni-bonn.de, dgg@aquila.gsfc.nasa.gov, mueskens@mpifr-bonn.mpg.de, nothnage@picasso.geod.uni-bonn.de, nrv@bootes.gsfc.nasa.gov, sorgente@hp138.mpifr-bonn.mpg.de, vlbi@asi.it, weh@ivscc.gsfc.nasa.gov In-Reply-To: <199907191835.OAA14947@hyperion.haystack.edu>; from "Leonid Petrov" at Jul 19, 99 8:31 pm X-Mailer: Elm [revision: 109.18] Status: R Dear Leonid, > Idea to resolve ambiguities in phase of calibration signal turned out > great. I've seen a clear harmonic signal instead of chaos of segments of > sinusoids. It is really exciting. What actually I've seen and continuation > of discussion you may find in Web: > > http://giub.geod.uni-bonn.de/vlbi/development/spur/matera/spur.html#sect4 Very nice discussion there and in your link at .../spur/spur.html. Your LSQ fitting program for spurs should prove to be very useful (unfortunately!) for other stations and experiments as well. A couple quibbles: 1) The harmonic number for case C, as defined in .../spur/spur.html (i.e., the dominant Matera spur), is actually 0.75, not 0.80. That is, one cycle is 480 degrees long, so in 360 degrees of phase there is 360/480 = 0.75 cycle. (I think that's how you're defining harmonic #, right?) 2) I wonder about the reality of the 1.00 and 1.25 harmonics you find in the Matera 8352.99 and 8512.99 MHz data. The 0.80, 1.00, and 1.25 cycle sinusoids are far from orthogonal, and there may well be pcal amplitude variations arising from non-spurious causes (e.g., weather-related time variations) that could mistakenly be well fit by a combination of those functions. Plus there are data gaps, and at times the unmodeled data scatter is larger than at other times. How large are the correlations between coefficients? I'm not saying the 1.00 and 1.25 harmonics aren't really there, but I am suspicious. What happens if you fit just a 0.80 harmonic to the data and then subtract it from the data? Do you see a large 1.00 or 1.25 harmonic signal in the residual? It might be possible to make use of pcal amplitude ratios between channels, which have much smaller scatter than the single channel amplitudes, in the fitting process, to avoid finding "spurious spurs"(!), but I haven't thought about it at all. Of course if you take the amplitude ratios of 2 channels that have identical spurs and plot the ratio against phase, you'll see a straight line and totally miss the spurs, so caution is advised! Another case you didn't mention is X-band contamination of S-band phase cal. That will result in harmonic numbers 3 (S normal) and 5 (S image). Several years ago we noticed a low-level contamination of the S-band phase cal at Gilcreek that was characterized by a sinusoid with 2.5-3 cycles per 360 degrees. I'm wondering whether X-band is leaking into Gilcreek's S-band phase cal. Recent data look better than what we saw back then, so either the problem's been fixed (accidentally or deliberately), or it's intermittent. I plan to follow up on this. > I hope you will also look at the next section > > http://giub.geod.uni-bonn.de/vlbi/development/spur/matera/spur.html#sect5 > > where spurious signals in phase cal at Westford are shown. I guess your office > is near the station and you may be able to fix these spurs and to tell us > how will you do it. The Westford spur at 8352.99 MHz comes and goes and has been giving me fits for years. The other 2 or 3 are a surprise. I'll check them out at Westford, but don't hold your breath waiting to hear the fix! Best regards, Brian