PolA ripple 2011-2012
The 430 dome receiver has had trouble with
polA during 2011 and 2012. There have been:
- Ripples in the bandpass that changed with time. On,off
position would then not cancel the bandpass shape.
- Power dropouts of 5db or more. This was correlated with
turning the load switch from antenna->load-> antenna. This
was thought to be a problem with the Jennings relay that does
- ripple has been seen in polA and not in polB for much of 2011
- On 18oct12 it was found that the jennings relay for polB was
not letting much signal through from the sky.
- So prior to 18oct12 polB saying polB had no ripple did not
necessarily mean the polB was ok.
- The typical period of ripple was 2-3 Mhz. A period closer to
1.5 Mhz has also been seen. Later there was some higher freq
ripple < 1Mhz.
- the rf system contains a number of components that could
cause this problem. Starting at the omt.:
- jennings relay (load,antenna).
- After 18oct12 polB jennings relay was removed.
- radar, radio astronomy switch.
- cavity filter before the dewar.
- The polA input connector was damaged long ago. It
has a C clamp holding it.. (a bit suspicious)
- The filter was replaced by a spare filter, but the ripples
continued to come and go.
- Various pieces of connecting wave guide.
- Over time various pieces of the above were thought to be the
- We may actually be looking at various things failing
- Some measurements that may tell us something:
- radar, radio astronomy switch:
- The ripple did not go away when the radar, radio astronomy
switch was bypassed (11aug11).
- Taking data while switching between radio astronomy
and radar mode showed ripple in both positions of the
- Jennings relay: load/sky
- The 2.5 Mhz ripple was also present when the receiver
(jennings relay) was on load (26mar12).
- there was a 5db drop in power after a load,sky,load
switch. This only occurred once. (26mar12).
- when terminations were placed at different positions in the
chain (30mar12) we saw large oscillations in the signal when
the termination was removed.
- This eventually went away during the test...
- The ripple is not always present.
- This makes it harder to test.
- Data taken after 18oct12 (when the damaged polB
jennings relay was removed) showed that polB did not have any
History and Measurements
in reverse order: (top)
19nov12: take data while switching radio
-> radar mode switch
On 19nov12 spectra were taken while the radar,
radioastronomy switched was moved from radio astronomy -> radar
mode -> radio astronomy.
The plots show the spectra
while the switch was moved (.ps) (.pdf):
- Page 1: total power vs time
- Top 460 secs of data
- black is polA, red is polB
- the jump in polA around 120 secs is when carlos was
tapping on the polA output cable of the radar,radioastronomy
- bottom: blowup when the switch was turned
- 290 secs. moving radio astronomy to radar
- 300 secs, moving radar mode back to radio astronomy.l
- When the switch was between contacts, the power dropped to
60% of Tsys.
- When the switch was in radar mode, Tsys increased by 15%.
This is because the monoplexor is in the chain when in
radar mode position.
- Page 2: spectra during switch change. These are the spectra
from the bottom plot of page 1.
- Top: over plot spectra
- Black in in radio astronomy mode ( at beginning and at end
- Red: moving radio astronomy -> radar
- green: in radar mode
- blue: moving radar mode to radio astronomy mode.
- You can see the ripples in both positions of the
radar,radio astronomy switch.
- Bottom: over plot spectra with offsets.
- We started in radio astronomy mode at the bottom of the
- The ripples appear in polA in both positions of the
radar,radio astronomy switch.
- There are two sets of switches: one to switch the input, and
a second set to switch the outputs.
- If the switch is bad, it is not the output contacts (unless
both are messed up).
- Previously we saw ripples with the switched bypassed so it's
probably not the switch.
- When the switch is between positions, the ripples go away.
- the Tsys only drops to 60% so there is still appreciable
power in tsys
- this power is probably coming from the lna (with the lousy
match when the path is open).
- This might be saying that the ripple is no in the lna or
downstream from the lna?
- Switching between radio astronomy and radar mode increases
Tsys by about 15%.
nov12-dec12:A2704 430 continuum mapping.
Project a2704 is doing continuum mapping at
430 Mhz using the dome receiver. Their setup is:
Images showing spectral averages for each strip
- 430 dome receiver. 12.5 Mhz centered at 430 Mhz.
- interim correlator 2048 channels across 12.5 Mhz. 1 second
dumps, 9 level sampling
- ra,dec strips ranging from 60 to 300 secs
To get an idea of the amplitude of the ripples, the spectra for each day were
averaged and then over plotted (.ps) (.pdf):
- the spectra for each strip in a map was averaged.
- the average spectra was normalized to the median spectra
value between 422 and 441 Mhz.
- a 1st order linear and 3rd order harmonic was fit to the
normalized average spectra, and then subtracted.
- dynamic spectra were then made of the average spectra.
- The bottom of the image is the first spectra. The top of the
image is the last spectra.
- The 1st spectra of each run is labeled on the left of the
- PolA image
of strip averages (.gif)
- The large dark,light variation are the ripples
- The polA cavity filter was replaced on 03dec12.
- The ripples were seen on 04dec12, but the went away on
- PolB image of strip
- You don't see the ripples in polB.
- It is easier to see the rfi in polB
- there is a comb spaced by 1.5Mhz
- at each comb position there are about 6 spikes spaced by
- It seemed to go away after 07nov12.
- 431.7-431.85. This is a tone that is not resolved in
the 12khz (hanning smoothed channels).
- it wanders over this range during an evening.
- Top frame polA, bottom frame: polB
- Each day is color coded with the date (yymmdd) on the left)
- an offset has been added to each spectra for display
- Tsys is close 1.0.
- the polA ripples have been present for more than 1/2 of the
- The polA notched filter (with the bad input rf cable) was
replaced with a spare notched fitler (from the ch) on 03dec12.
- On 04dec12 the ripples returned, but they then disappeared
- So it looks like the ripples are not being caused by the bad
input connector on polA filter.
- I won't know if the ripples wreck the total power measurement
(but my guess is that they will).
- We need to find what was making the 1.5 Mhz comb
- the rfi a 431.7 is a problem since it wonders over frequency
(wiping out multiple channels).
x111 rfi monitoring data from oct,nov12 was
examined to see when the 430 polA ripple was present.
The plots show the the
average spectra for each 60 sec scan (.ps) (.pdf):
- the 25 MHz spectra for each 60 sec scan was averaged.
- the median spectral value between 422 and 441 MHz was used to
normalize the spectra
- so if Tsys got large, this would hide it
- a robust linear and 3rd order harmonic fit was done over the
- data outside of 422-441 was weighted to 0.
- The fit was then subtracted from the average spectra. This
was to remove some of the bandpass.
- the spectra are plotted with an offset for display purposes.
- The vertical scale in Tsys
- the top frame is polA, the bottom frame is polB
- The date and time (h: hour AST) for each measurement is
printed in the left column
- each measurement uses a different color.
- The bottom spectra of each frame is the earliest data for
the month. it corresponds to the date at the bottom of the
- Page 1: oct12. Top is polA, Bottom is polB
- there were a few days of ripple (04oct12,09oct12,10oct12
at the start of the month.
- The ripple went away, and then returned at the end of the
month (28oct,30oct,31oct.. but missing 29oct12).
- The period seems to be 1-2 MHz (at least for the end of
- 1-2am AST was around 3:30 lst. So we aren't looking at
- Note that spectra below 20oct12 (red) about midway in the
frame show very little rfi.
- On 20oct12 the polB jennings relay was finally removed so
the signal could then get in from the sky.
- Page 2: nov12 (01 thru
22nov12). data was processed before the end of the month.
- The spectra look ok until 07nov12 When the ripple starts
- the period is still the 1-2 MHz.
- The large ripple on 19nov12 at 12.5 hours should probably
- The sun could have been doing this.
- On 21,22 nov the ripple had disappeared.
- we exercised the radar radio astronomy switch many times
- But we saw ripple in our 19nov12 data after the test.
- no ripple.
- The rfi spikes are getting through (just like polA).
- The jennings relay is not present in the polB line.
- polA ripples were present in 6 of 31 spectra
- the period was 1-2 Mhz.
- the jennings relay was removed from polB after 19oct12.
- So polB measurements were now seeing the sky. No ripple
- polA ripples were present in 7 of 16 scans.
- The period looks like it might be less than 1 Mhz.
30mar12 - terminate input to dewar at
various spots. (top)
On 30mar12 we terminated the input to the
dewar at various locations and then took measurements looking for
The measurements consisited of:
The sections before the dewar are:
- Dewar input terminated with 50 ohm load.
- The interim correlator took 25 Mhz data centered at 430
Mhz. At least 400 1 second integrations were taken at each
- 4 different correlator boards were used to check that there
were no problems in the if/lo or correlator (none was seen).
- Waveguide to coax transition ->Jennings relay->
- cal coupler -> elbow->straight line->elbow ->
- filter cavity->elbow->elbow->cable ->dewar.
The sequence of terminations was:
The plots show the on/off-1 for
various termination locations (.ps) (.pdf):
- morning :
- Terminate at input to cal coupler
- Terminate at output of cal coupler
- hagen's timing box was probably running switching the
monoplexor on/off (but it was not in the chain). The blanking
switched were off.
- terminate at input to RadioAstronomy switch.
- termination -> line->elbow -> filter
- filter -> elbow -> termination -> elbow
- No ripple in polA. scan 74
- hagen's box was turned off toward the end of this scan.
- termination -> filter
- termination -> line-> elbow -> filter.
- no ripple polA. scan 76.
- This is the same termination configuration as scan 73
where ripple was seen.
- The only difference is that hagens box had been on for
- termination->line->elbow -> filter
- no ripple polA. scan 77
- turned on transmitter (no klystrons) and hagens box.
- this was no the same as scan 73 where ripple was seen.
- no ripple polA scan 78l
- turned transmitter off.
- tapped on elbow into filter for seconds 40-70 seconds.
- System reconnected, Jennings relay installed.
- System on load.
- scan 79- ripple present.
- We never turned the load switch antenna, load (it was
always on load) so the Jennings relay probably did not
send the switch to load after the installation.
- scan 80,81 - no ripple
- we switche load-> antenna->load before the start
of each of these scans. No ripple was seen.
Dynamic spectra were made for the
530 seconds of scan 73 (.gif):
- Page 1: on/off-1
- each frames shows a on/off-1 for a 200 second avg on, off.
- black is pola, red is polb.
- top: termination -> raSwitch
- 2nd: termination -> line->elbow -> filter
- 3rd: filter -> elbow->termination
- 4th: termination -> filter
- 5th: termination -> line->elbow-> filter
- bottom: termination->line->elbow->filter,
- Page 2: scan 73 termination -> line -> elbow-> filter
- Top. the spectra for average 1-200 seconds is over plotted
on the 201-400 second average
- black,blue is polA.You can see that the do not trace one
- red,purple: polB. These overlay one another.
- bottom: The total power (25 Mhz) vs time for this scan
- black is polA, red is polB. the vertical scale is +/- 1%
tsys (at 300k)
- there are gain variation in polA (.7%) and polB (.2%)
- The scan lasted for 530 seconds.
- Top polA, bottom polB
- The image was flattened using the median spectra for the 530
- The clean area to the left of 421 Mhz is where the filter has
cut off the power.
- You can see where the ripple is changing with time.
- Ripple was seen when a termination was placed:
- before the cal coupler, the raswitch, and termination ->
- Ripple went away after the termination was moved to in from of
- When returning the termination to the previous locations where
the ripple had been present, it was no longer there.
- So the presence of the ripple is not always repeatable.
- One interesting think we noticed was that:
- When ripple was present, removing the termination
would cause large (> 30db) oscilllations (probably in the
- When the ripple stopped, removing the termination to move it
to another spot did not cause these large oscillations.
26mar12- PolA ripple with 430 on
On 26mar12 data was
taken with the Gregorian dome on load to check out the polA
ripple. It had been noticed previously that the Jennings relay for
polA (that switches between load and sky) gave non-repeatable
results when switched from load to antenna and then back to load.
The setup was:
The first set of plots shows
the total power stability vs time (.ps) (.pdf):
- gregorian dome on load
- interim correlator using 1024 channels across 25 MHz, 9 level
sampling and 1 second integrations.
- The 25 MHz bands were duplicated in 4 different correlator
bands to look for any variation iin the correlator.
- data was taken for about 300 seconds
- Then the Jennings relay was switched load to antenna and
then back to load.
- the entire 25 MHz bandwidth was averaged for these plots. The
resulting 1 second values where then normalized the the median
value in each sbc (excluding the polA dropout).
The 2nd set of plots shows the
frequency baseline stability (.ps) (.pdf):
- Page 1: total power Pola (top) polB (bottom)
- the red dashed lines are where the data taking stopped and
then the antenna load switch was moved from load ->
antenna->back to load.
- the signal from the dome was repeated in 4 sbc of the
- You can see the dropout around 1100 seconds. It was about
a 5db drop in power.
- Looking at the spectra, there was no rfi, so the switch
did not remain on the sky.
- The switch could have been between sky and load.
- Page 2 : blowup the vertical scale.
- The full y scale is now 1% of Tsys (tsys=about 300 K
- You can see the signal wandering in time in both polA and
polB. this is probably gain variation.
- polb had a jump of about .002 Tsys at 120 seconds.
- Both pola and polb drifted around at the end of scan 2
(around 650 seconds).
- Page 3: blowup up the plot around 650 seconds where polA and
polb moved around.
- The 5 300 second scans were used to measure the stability of
spectra density vs time.
- For each 300 sec scan:
- break the scan up into 75sec ON, 75 secOff, 75secOn,
- Compute on/off-1 for each pair of 75secon,offs and then plot
- Page 1: 75 second average spectrum
- black is polA, red is polB
- The falloff near 422 MHz is the 422 to 442 rf filter.
- Page 2: scan 1 on/off -1
- Black polA in the first 150 seconds has large ripple of
about 3 MHz.
- The large bump at 421 is the filter fallloff.
- Page3-6 on/off-1 for scans 2-6
- The 70 second scan 4 where polA jumped by 5db was excluded.
- Moving the load switch load->ant->load will occasionally
cause a power dropout of about 5db in pola.
- there are some jumps of up to .2% Tsys (@300K) that don't look
like they are related to moving the switch (they appear hundreds
of secs after the switch).
- There are gain variations of about .4% Tsys (@300K) over about
300 secs in both both PolA and PolB.
- PolA has larger ripple than polB
- the period (in frequency) for the ripple is
about 3 MHz (but it is not constant across the entire 25 MHz
- the amplitude is up to +/-.4% Tsys (@ 300K) ..about
- I also looked at some higher time resolution data (.1 sec
dumps) and did not see any 1.2 Hz periodicity in the total
power. So the cross heads don't look they are causing problems.
17mar12 - a2599 has bad ripple in
a2599 reported bad ripple in polA.
The plots show a 300 second
average spectra using the wapps (.ps) (.pdf)
- Black is pola, red is polB.
- The 1st two frames used a bandwidth of 25 MHz. the last 2 used
- The ripple is in all 4 wapps, so the problem is not in the
The 430 dome polA bandpass has shown variations in its
bandpass. They first appeared on 30mar11. The x111 rfi
monitoring data was used to check when it occurred. The x111 data
The plots show the 430 dome
x111 rfi monitoring data for mar thru 22jun 2011 (.ps) (.pdf):
- X111 data is taken whenever the schedule allows.
- 1 minute integrations of 25 MHz, 1024 channel are taken for
the 430 dome band.
- mar11 had 14 separate 1 minute integrations for 430 dome
spread throughout the month
- apr11 had 29 separate 1 minute integrations for 430 dome.
- may11 had 19 separate 1 minute integrations for 430 dome
- Each 1 minute integration was averaged and then normalized to
a mean value of 1 (Tsys).
- The Andrews connectors for polA were tightened 10may11
- Page1: mar11 with offsets and dates.. top is polA,
bottom is polB
- polA: 30mar11 (red trace) has the irregularities
- Offsets have been added for display purposes (page1,page2).
- Page2: apr11 with offsets and dates. Top polA, bottom polB
- polA: ripples occur more often. 08apr11 is an example.
- Page3: may11 with offsets and dates (thru 31may11).
- strong ripples at start of month.
- Page4: jun11 with offsets and dates. (thru 22jun11).
- Page5: Over plot band passes mar11,apr11 without display
- The top two frames are mar11 polA, polB. The bottom 2 frames
are apr11 polA,polB.
- Page6: Over plot may11
- 1st,2nd frames: polA , polB
- black is data before 10may11, red is data after 10may11.
- 3rd,4th frames with offsets for display
- black is before 10may11, red is after 10may11.
- For polA starts to have ripples 6 spectra from the top.
- Looking at page 3 this was 21may11.
- Page7: over plot jun11
- 1st,2nd frames: polA , polB no offsets
- black is data before 08jun11, red is data on and after
- 3rd,4th frames: polA, polB with offsets
- black is data before 08jun11, red is data on and after
- PolA has shown irregularities in the band pass:
- 30mar11 - 10may11processing x101/120326/chk430.pro.
- 10may11 Andrews connector tightened, ripples went away.
- 21may11 ripples in polA return
- 10jun11-06jun11no ripples
- 08jun11 - ripples started to get stronger. by 18jun11 pretty
- the largest deviation is about .5 Tsys.
- hopefully the repair
of the radio astronomy,radar switch for polA will fix this