History:

• Tsys for polA:
  
• dec11junped up
  
• jan12 jumped back down.
• apr12 jumped back up:
• (see Tsys polA-polB for cband)
• 2011,2012:A2623/a2471 (monitoring brown dwarfs.. alex wolszczan) would occasionally see gan variations  in the total power.
• 2011-2012   stage 2 pola A drain current has tended to have a larger rms than polB
  
• 08may12: X102  20:00 saw large oscillations in polA amplitude (3*Tsys) with a period equal to the cross head (1.2hz).
• 08-12may12: stage 2 biases adjusted.
• stage 2 polA drain current rms increases
• stage 2 then switched to be driven by stage 3. drain current rms gets even larger.
  
• 09may12: crosshead oscillation caused 1% gain change in rf output of system.
• 12may12: The receiver was brought down to the lab around 12may12.
  
• some loose cables were found and repaired
• It was reinstalled 31may12.
• This seemed to fix  oscillations at the crosshead  frequency
• 31may12: look at rf  stability after reinstalled in the dome
• Don't see cross head frequency in the data any more.
  
• 31jul12: plot dewar bias monitoring for jul12.
  
• Spikes seen  in stage 2 drain current polB,
•  rms increase in drain current polA stage2.
  
• 20aug12:total power data taken during rms increase of polA stage 2 drain current showed that rf  signal is affected.
  
• 21aug12: polA drain current driven by stage 3 (,ps) (.pdf)
  
• stage 2 polA drain current rms increases
  
• 27aug12: cband now using bias box from sbh.
• polA stage 2 drain current instabilities remain.
• 29aug12: replaced the bias cable from bias chassis to dewar.
• Note polA amp:
• Old Amp; #19
• new Amp #18 installed 25sep-18oct12
  
• 25sep12: new amp  replaces  polA oldAmp. tsys jumps to 40k (more info)
• 18oct12: cb warmed up, opened in the dome. new amp replaced with old amp. found bad elbow in dewar
• oct12    : old amp has larger rms in drain current  ( new amp doesn't, but tsys increases).
  
  

Summary:

• crosshead oscillations:
• PolA: The cross head oscillation caused a gain change of about 1% in tsys.
• Wacking on the dewar caused large spikes in the rf output of the dewar.
• This was fixed when the dewar was brought down to the lab and worked on (may12)
  
• PolA stage 2 drain current:
• PolA stage2 drain current readings have had rms increases during 2011 and much of 2012.
• Adjusting the biases caused stage  2 drain current rms to increase.
• driving the stage 2 polA by the unused stage 3 caused the rms to increase on the drain currents.
• RF total power variations occur when the stage2 drain current rms increases.
• replacing polA amp 18 with amp19 got rid of the rms increases in the drain current (although tsys went up).
• Putting amp 18 back caused the jumps in polA drain current to reappear.
  
• PolB stage 2 drain current spikes.
• narrow spikes started to appear in polB stage 2 drain current around 6jun12.
• vlbi reported dropouts 14jul12 on polB. there were a few polB stage 2 drain current spikes during this time.
• It's not obvious whether the polB spikes are real or a monitoring artifact.
  

26oct12:polA drain current jumps around with amp#18

    On 18oct12 cband was warmed up and opened (in the dome). The new amp #18 was replaced with the older amp#19. A bad elbow was found in the dewar.
The plots show the drain current (polA,B) using amp#18 (25sep-18oct12) and amp#19 (pre 25sep and after 18oct12) (.ps)  (.pdf)

• top,2nd frames: drain current stage 1 polA and polB
• day number 290 was 16oct12.
• 3rd,bottom frames: drain current stage 2 polA and polB.
• The left of the break is amp 18 before the 18oct12 warmup
• the right of the break is amp 19 after the 18oct12 warmup.
• You can see that polA stage 2 drain current started jumping around after amp 19 was put back in.
• It was not jumping around when amp 18 was installed.

29aug12: replaced cable from bias chassis do dewar

• cband drain currents.
• Dates:
• 240.4 : (27aug12)bias box was switched
• 242.4: (29aug12) bias cable replaced
• 248.3: (04sep12) bias voltages turned off for start of 430  dual beam run.
  
• Top Stage1 polA
• 2nd: Stage1 polB
• 3rd: Stage 2 PolA
• 4th: Stage 2 polB
• 
  
• Looks like stage2 polA still has more glitches than the other measurements even after the bias cable switch.
  

27aug12: cband switched to use sbh bias box

    To debug the polA instabilities, the bias box for cband was replaced with the bias box from sbh on 27aug12.

The plot shows the cband bias drain currents for 27,28aug12 (.ps) (.pdf)

• The switch occurred around 10AM on 27aug12. (marked with a vertical dashed blue line).
• Black is stage 1, red is stage 2, green is stage 2 powered by stage 3.
  
• Top frame: PolA drain current.
• Till 10am it was driven by stage 3 (green). After 10am it was driven by stage 2 of the new bias box
• Bottom frame: polB drain currents

• switching to the new bias box did not stop the rms increases for stage 2 polA drain current
• The polB values drain current values differ from the old values by a few 10's of milliamps
• The next place to look for the stage 2 polA drain current problem is:
• cable bias box to dewar
• In the dewar.

20aug12: polA total power changes with stage2 drain current rms increase.  (top)

• telescope sitting at: az=285,za=11
• data was recorded 9:30 to 12pm
• The receiver was set to 5000MHz.
• The interim correlator recorded the following:
• 25 MHz bw, 1024 channels, 1 second sampling
• this was duplicated in 4 separate interim correlator boards.
• The total power was then computed over the 25 MHz band.
• The dewar monitoring was recording the bias voltages. Cband's values got recorded every 30 seconds.

• Page 1: the full 2.5 hour stretch of data.
• Top: total power vs time ( 1 second sampling).
• Black in polA, red is polB.
• The data has been normalized to the median value and then 1 has been subtracted. This makes the vertical scale units Tsys. There are 4 copies of pol
  A and polB from the 4 interim correlator boards. An offset was added for display purposes..
  
• You can see the rms of the total power increase around 10 and 11 am (after a drop in value).
• Middle: drain voltages
• the different amp stages and pols are color coded. It looked pretty stable.
• bottom: Drain currents
• The stages, pols are color coded.
• PolA stage 2  starts jumping around when the rf power rms increases. This happens in both instances.
• Page 2,3: blowup of the rf power rms increase.
• Page 2: first event.
  
• Only polA is plotted
• Page 3: 2nd event
• again only polA is plotted.
  

Summary:

• The rms of polA  rf power increased.  This occurred after a drop in the value (don't know whether it is gain or Tsys).
• The drain current for stage 2 starts to jump around during this occurrence.
• It happened twice in two hours. It lasted for 15 to 20 minuted.
• The dewar bias monitoring for july was plotted on 31jul12. It showed polB jumping.
  
• If you look closely you will also see the rms on stage 2 drain current was also increasing and decreasing.
  

31jul12: plot dewar bias values for jul12. (top)

    Jumps in polB were reported on 14jul12 during a vlbi run. The bias monitoring for jul12 was examined to see if it was stable.

The voltage and current monitoring of the dewar  biases from  jul12 is plotted (.pdf):

• Black is polA, red is polB.
• When all stages jump up, the sband or 430 tx is usually on.
  
• Top: voltages for stage 1
• 2nd: voltages for stage 2
• 3rd : currents for stage 1
• bottom : currents for stage 2
• PolB( red) has narrow spikes in it while the other values do not show any jumps.. This  started around 6jun12.
• there were 2 jumps during the 14jul12 run that chris reported.
• This may be real, or a problem with monitoring
• You can also see intermittent increases in the rms for polA drain current stage 2 during the month.
  

summary:

• stage 2 polB drain current measurements show narrow spikes while the other measurements are clean.
•  This could be a monitoring problem, or it could be the problem with polB.
• stage 2 polA drain current show an intermittent rms increase in values.
•  This has been going on for much of  2012.
  

31may12: check cband after reinstalled in dome.  (top)

    the cband receiver was brought to the lab on 13may12, worked on , and then reinstalled 30may12. On 31may12 blank sky was tracked with the cband receiver to check the stability. The setups was:

• The cband receiver was centered at 5000 Mhz.
• The mock spectrometer took a 172 Mhz band with 8K channels. The data was sampled at 10 hz.
• Blank sky was tracked: 10:30:00,24:00:00. The azimuth went from 240-222 deg, the za fromj 12.3 -> 7.8 deg. The data was taken around 17:00 hours ast.
• About 25 minutes of data was taken.
  

Processing the data:    

• The rms by channel was computed using the 14000 .1 second spectra.
  
• A robust linear fit to the rms was used to exclude frequency channels that had rfi.
• the total power was then computed over the remain channels

• Page 1: total power vs time:
• The vertical scale is +/- 1% of Tsys.
  
• The raw data (black)  is samples at 10Hz.
• the colored lines (red,green) have been smoothed to 5 seconds.
• Top: polA pwr vs time. You can see a jump around 1145 seconds.
• 2nd: polB pwr vs time. polB does not jump when polA jumps.
• bottom: over plot the smoothed polA ,polB total power vs time.
• Page 2: total power vs time: blowup around polA jump
• This shows the jump with .1 sec resolution.
• the pwer jumped .003Tsys in about 5 seconds.
• Page 3: spectral magnitude of total power.
• the transform of the total power was computed and then the magnitude was plotted vs frequency.
• Top: 5 Hz
• Bottom: 1.5 Hz blowup.
• You know longer see the 1.2 hz cross head oscillation. It was strong before the dewar was brought to the lab (see below.
• Page 4: cband bias values during measurement.
• I plotted the cband bias values (currents stage 1,2 and voltages stage 1,2) during the measurement.
  
• Top,2nd bias currents
• 3rd,bottom: bias voltages
• The blue dashed line is when polA jumped. The bias values remained stable.

SUMMARY:

• The cband dewar no longer has a gain variation at the cross head frequency (as it did prior to being brought to the lab).
• There was a small jump in polA (.003Tsys) during the 25 minute measurement.
• It is a bit suspicious that polA was the one that had the problem prior to being worked on.
• There were slow variations in  both pols of about .005 Tsys during the measurements. This could be sidelobes (sun?) or temperature.

09may12:  cross head oscillation causes a gain change. (top)

    Cband was seeing an rf oscillation of about 1% synchronized to the cross head  cycle (1.2hz). High time resolution data was taken with the 25 hz cal running to see if this was an increase of Tsys or a gain change.

 The setup was:

• 25 hz hardware winking cal with high correlated cal (20 milliseconds on  , 20 milliseconds off)
• mock spectrometers 7 x 170 Mhz bands covering 5 to 6 Ghz.
• 1024 channels, 1 millisecond dump time.
• Take 60 seconds of data.
  

• For each 172 Mhz band compute the total power for polA, B for each 1 millisecond. We ended up with a 1 ghz total power single at 1 millisecond time sampling.
  
• data set 1: Scale the data to the average cal deflection for the 60 seconds. This assumes the gain is constant.
• data set 2: Scale each 40 millisecond cal to its own cal deflection.
• Remove the cal by subtracting 1/2 the cal deflection from the cal on, and adding 1/2 the cal deflection for the cal off.
• For dataSet 1 use the average cal deflection for the 60 seconds
• for dataSet 2 use the average cal deflection for each 40 millisecond calon,cal off.

• Page 1 total power vs time:
• Top: Total power vs time polA for each of the 7 172 Mhz bands.
• The bands have been offset for display. The oscillation seen is the 25 hz cal.
• Bottom: Average the 7 172 Mhz total power time series.
• Black is polA, red is polB.
• Page 2: Tsys vs time:
• use the averaged (over frequency) time series
• Black is polA, red is polB
  
• Top: Use the average cal deflection for the 60 seconds to scale the data.
• This assumes the gain is constant.
  
• remove 1/2 the cal deflection from the cal on, add 1/2 the cal deflection to the cal off.
• Smooth the data to 41 milliseconds.
  
• PolA (black) shows the cross head oscillation of 1.2 hz (.83333 seconds).
• The amplitude is  about 1% Tsys peak to peak.
  
• PolB (red ) does not show this  oscillation
• Bottom: compute Tsys by scaling each 40 millisecond conon,off separately. 
• this assumes the gain is   constant within the 40 millisecond calon,off.
• remove 1/2 the cal deflection from the cal on, add 1/2 the cal deflection to the cal off.
• Do this separately for each 40 millisecond calon,off.
• The 1.2 Hz oscillation is no longer evident.
• Page 3. Fold the 60 second dataset to 3 cycles of the cross head.
• Top: Use data set 1 scaled to the average cal deflection.
• You see the .833 second cycle
• Bottom: Use data set 2 scaling each 40 millisecond calon,off separately.
• You no longer see the cross head period.
• There is a glitch the cross head start.
• This is probably caused when a 40 millisecond cal cycle crosses the cal cycle.

Conclusions.

• The 1.2 Hz cross head cycle is seen in the rf output.
• The peak to peak variation is about 1% Tsys.
• The cross head cycle is not seen in the Tsys signal when each 40 millisecond cal deflection is scaled separately.
• This implies that the change is a gain change, and not a Tsys increase.

08-12 may12: Dewar monitoring of Drain voltages and Drain currents. (top)

    A plot was made of the drain current and drain voltage for 08may12-12may12 (.ps) (.pdf):

• Data is recorded every 30 seconds,.
• Black is polA, Green is polA stage 2 after switch to stage 3 circuitry, red is polB, dashed purple line is where the biases were adjusted.
  
• daynumbers:
• 129 - 08may12
• 130 - 09may12
• Dewar biases adjusted near 130.48
• sband transmitter on  130.8-130.9
  
• 131 - 10may12
• bias chassis brought down to the lab. stage 2 monitoring was switched to use the stage 3 circuitry in the bias chassis.
  
• 132 -  11may12
• 133 -  12may12
  
• Top: dewar drain voltage stage 1
• 2nd: dewar drain voltages stage 2
• 3rd : dewar drain current stage 1
• bottom dewar drain current stage2

Summary:

• Adjusting the bias currents on 09may12 and then returning them to their original values caused the current in stage 2 polA to start jumping around
• It continued for the next day.
• When the stage 2 monitoring was changed to port 3 on the  bias card, the polA jumping around continued.
  
• This may be saying that the cband problem may be in the bias cards rather than the dewar.
• I looked at the bias voltages and currents during the large signal jumps of the x102 run on 08may12 (18:50-19:30). Nothing unusual occurred.
  

cband page
home_~phil