[evlatests] Results from 20-db LSC converter attenuator

[Jim Jackson]

The level setting algorithm in the T304 may take considerably 
different times depending on the starting and stopping points and 
which direction the change is in.  I suspect much of the strange 
timing that was seen in this test is related to this.  As long as it 
ended up at the right level within a certain amount of time (I think 
20-30 seconds is not unexpected), it is working as designed.

Any Antenna/IF that did not end up at the right level after the T304 
ALC routine ended needs to be looked at - though the known RF 
switching issues are probably the most likely culprit.

Cheers,
Jim

At 04:11 PM 4/14/2011, Rick Perley wrote:
>     We ran the 'solar attenuator' test script on April 4.  Stephen White
>extracted the essential results, which I am only now able to review.
>There are some 'interesting curiosities'...
>
>     To review:  The test is quite simple:
>
>     Observe 3C84 and the sun at standard L, S, and C band frequencies.
>For each, spend a couple minutes observing in normal ('cold sky') mode,
>then turn on the 20-db solar attenuators, and watch the system react.
>In principle (and in an ideal world), there should be a dramatic
>factor-of-100 drop in power delivered to the samplers, which is restored
>within a few seconds as the T304's attenuators react.  The fringe
>visibilities should be about the same in either state ( a couple of dB
>different will be fine, as the switched power monitor should correct the
>difference), as should be the calculated system temperature.  (In fact,
>because we have not any solar cals yet, the calculated Tsys, while on
>the sun, will be only noise, as the regular cal will be swamped by the
>sun's emission.)
>
>     What we saw:
>
>     A)  L-Band
>
>     The total power (PSum) change while on 3C84 matches the
>expectations:  A dramatic drop, lasting three seconds, followed by a
>rapid (1 or 2 seconds) rise to about the same level as before.
>     The behavior upon transition while on the sun is entirely
>different!  There is the drop as before, but instead of a rapid rise,
>ther is (usually) a slow rise, lasting 20 seconds or more, culminating
>in a power which is as much as six times the original value!!  After
>this, a dramatic drop to (about) the correct level is seen.   For some
>antennas, an entirely different behavior is seen, with a large up *up*
>(rather than down), followed by a slow stepwise decline to the correct
>level.
>     Why is the behavior while on the sun different?  The actual power
>level is the same (the solar power having been attenuated by the T304's
>own attenuators).  So how does the T304 'know' that the power it is
>seeing is from the sun, rather than cold sky, and why is it reacting
>differently?
>     The other curiosity is in the before/after values of Tsys, when on
>3C84.  Since the T304 and T302 should not add any appreciable noise, the
>Tsys values should be mostly independent of the actual attenuator
>setting.  This is generally true -- but there are notable exceptions:
>(All are in LCP, as these are the plots Stephen left for me):
>     ea15, Tsys doubled after transition (Tsys is higher with the 20 dB
>T302 attenuator in).
>     For ea08 and ea24, there was also a rise, but by about 10%.
>     For all other antennas, the change was not perceptible.
>
>     B) S-Band
>
>     We would hope the behavior would be similar to that for L-band, but
>we would be wrong in thinking it so ...
>     On 3C84, the power levels drop as expected, but rather than stay
>there for 3 seconds (as at L-band), they typically  stay there for 5 to
>7 seconds, then jump back up to the more-or-less correct level.
>     The behavior on the transition while on the sun is generally similar
>to L-band, except antennas which do not have the drop plus slow
>rise/overshoot at L-band, do while at S-band, and vice versa.
>     The change in Tsys while on 3C84 is similar to L-band -- most
>antennas return, some (like ea08) do not.
>
>     C)  C-Band
>
>     The power transition while on 3C84 is like L-band:  3 or 4 seconds
>down, then rapid return to expected levels.
>     The power transition while on the sun is generally like L and S
>bands, but appears to be faster -- the 'slow rise' is terminated
>earlier, with less overshoot.
>     The rise in system temperature upon transition on 3C84 is more
>notable at C-band:  Antennas with large (25% or more) changes include:
>ea08, ea14, ea17, ea18, ea22, ea26.  (Notable is that for ea15, which
>doubled its Tsys at L-band, there is no perceptible change upon
>transition at C-band).
>
>     Many thanks to Stephen, for his diligence in getting these plots out.
>
>
>
>_______________________________________________
>evlatests mailing list
>evlatests at listmgr.cv.nrao.edu
>http://listmgr.cv.nrao.edu/mailman/listinfo/evlatests