[evlatests] Demise of the 'Heat-Pipe' Theory?

[Rob Selina]

Hi Rick, 

Just a quick note:

ALMA Memo 466 has a table of temperature coefficients, and you can see they are all over the map (look at table 2) 

http://library.nrao.edu/public/memos/alma/memo466.pdf

This sample of L-band amps vary between 3% and 0.21% gain change per degree K. Coefficient has a lot to do with manufacturing process, so I'm not surprised by a clustering by band frequency. 

A quick look at the temperature sensors in the RF/IF boxes (where the post amps reside) and the temperature sensors in the vertex room over the last two days all show peaks in temp around 5pm and low points around 7am, consistent with your plots (no lag). 

Temperature changes for U/K/A/Q RF/IF boxes are all around 10 C diurnal, L/S/C/X vary between 5-7 C, with the larger receivers being more stable. 

So I'd conclude that the post amp temperature is changing on the cycle consistent with the gain change you see, and we have overall gain coefficients of order 1%/K for the low frequency post-amps and closer to 1.5%/K for the high frequency post amps. 

Wes - if you had the real data from the manufacturers for the temperature coefficients it would be interesting to see if they cluster as expected. 

Cheers,

Rob


-----Original Message-----
From: evlatests [mailto:evlatests-bounces at listmgr.nrao.edu] On Behalf Of Rick Perley
Sent: Tuesday, March 14, 2017 2:36 PM
To: evlatests at aoc.nrao.edu
Subject: [evlatests] Demise of the 'Heat-Pipe' Theory?

     We long ago noted the very large diurnal gain changes seen with the EVLA receivers -- particularly at the four high frequency bands.  
Typical power gain changes of 15% are noted, and in some cases greater than 20%.  These are clearly driven by external temperature -- and are expected if the actual temperature of the post-amplifiers change by 
significant -- but plausible -- values.   It was also noted that the 
low-frequency bands have similar, but considerably smaller (~5%), diurnal variations.

     To explain the effect, Bob Hayward posited the 'heat-pipe' 
mechanism:  That the horns, being exposed to the cold night air, were strongly thermally coupling the external temperature changes to the post-amplifiers -- which are mounted with good thermal coupling to the horns, and which are known to have a strong temperature coefficient.

     My recent analysis of data taken on 10/11 February casts some significant doubts (at least in my mind) on this theory.  I've attached eight plots, showing the change in PDif (hence, the change in system
gain) at the eight cassegrain bands over an 8-hour period, starting at 3AM in the morning, and finishing at about 11AM, for ea13.  This is not a 'specially chosen' antenna (except that it is one of only a few for which all switched power data are really clean at all bands).

     Some points stand out:

     1) The curves are all essentially identical.  Most importantly, the all maximize at the same time (7AM, local time). There is *no lag* between the time of maximum gain and the time of minimum temperature at any band.

     2) The four high frequency bands have exactly (to 1%) the same gain change -- 15%.

     3) The four low frequency bands also have the same gain change -- but at 5%.

     If the 'heat-pipe' explanation were the dominant cause, I would expect the S and (especially) L band gain maxima to be significantly later than that seen at the high frequencies.  This is not the case.  
Further, we might expect the X-band and Ku-band gain curves to be at least fairly close in amplitude -- since the horns are not very different in size.  This is also not the case.

     Furthermore, the gain changes seen at Q and Ka bands are exactly the same as at Ku and K bands.  Yet the horns for Q band (and I think also for Ka band) do *not* extend outside -- they are underneath weather windows.  Hence, externally-driven temperature changes should be much less at these two highest frequency bands than they are for K and Ku 
bands -- who horns do indeed extend well out into the night air.   But 
in fact, they are all the same.

     So how is the external temperature changes so strongly correlated with the gain changes?  And why are the four high frequency bands so similar?  (And the four low frequency bands also similar, but different than the high?) One thing connects all four high frequency bands -- the U/X converter.  Is it possible that this unit -- which is bolted to the side of the receiver cabin -- is the largest contributor to the observed diurnal effect?

     Rick




-------------- next part --------------
A non-text attachment was scrubbed...
Name: RF-IF-box-Temp.pdf
Type: application/pdf
Size: 65768 bytes
Desc: RF-IF-box-Temp.pdf
URL: <http://listmgr.nrao.edu/pipermail/evlatests/attachments/20170316/d55b733f/attachment-0001.pdf>