[evlatests] [Fwd: ph-slope]

[Vivek Dhawan]

		         Phase jumps and Phase slopes / 2007 Jan 4.

A one-hour test file was run at L-band, for reasons below.
First some minor notes:

A. No jumps of the global sort were seen on the EVLA, but
   EA23 alone did jump by 170+_2 deg. The jump scales with
   frequency, comparing IF1 and IF2.

B. Many VLA antennas (7,8,10,11,12,19,22,25,27,28) had small
   jumps of ~10deg, at mostly unrelated times. Only a
   small fraction of data was bad - the jumps were mostly
   very short. This seems unusual to me, but unless it
   persists I'll skip it for now.

C. The main purpose of the test was to poke at an un-
   explained feature that persists: The two IFs on the
   EVLA have, SOMETIMES, a phase slope with respect to
   one another (when referenced to a VLA antenna).
   This is a 'global' phenomenon, i.e., the phase of
   (IF1-IF2), on any VLA to EVLA baseline, has the same
   slope. It could be on either array, but my money is
   on the EVLA.

   I have mentioned this before, but Jim Ulvestad's
   imaging result prompted me to write it up in a bit
   more detail. I don't think it explains what he found,
   i.e., the EVLA added in makes a 4 sigma VLA detection
   20 microJy, to degrade to 2-3 sigma (I forget the exact
   number).

	The facts so far:

o  IF1 phase drifts at 28deg per hour compared to IF2.

   The exact number may be 25-30 deg. It is close to
   2 turns per day, or 23 microHz. Half that value has
   been seen, just once.

o  At L band, it is always present at the default settings
   of 1465 and 1385 MHz.

   It reverses sign when the IFs are interchanged in
   frequency.

   It disappears (unmeasurable, < 1deg/hr) when the IFs
   are close together (1421.46 & 1420.28 MHz).

   It is unchanged (still ~28deg/hr) when the IF's are
   set wide apart 1341 and 1666 MHz.

o  At C band it was
   ~0 (i.e. <1deg/hr) on June 10th.
   ~25 deg/hr on July 13th.
   ~12 deg/hr on Oct 4th.
   ~0  deg/hr in recent December data.
   All at the standard settings 4885 and 4835MHz.

o  At X-band it is nearly always zero at the standard
   settings 8435 and 8385 MHz. Except on Oct 4th,
   when it was 12 deg/hr like C band on the same day.


   I found the slopes while looking at the round-trip
   fiber delay change, which causes phase slopes of the
   same order of magnitude on individual IF channels, but
   which should scale with frequency difference. The fiber
   slope varies with temperature, and changes sign every
   day, whereas the IF differential slope is much more
   linear, at any time of day.

   I do not know if the IF differential reveals a roundoff
   or truncation in a frequency calculation. In normal use
   it is removed if the IF's are calibrated separately; even
   if the IF are combined, it is a 15 deg maximum error (for
   calibration every 30 minutes, typical at L-band).

Vivek.