# 1 Mhz comb in 327 receiver (coming from AC units in dome).

14oct05
Links to plots:
the single birdie characteristics (.ps) (.pdf):
Dynamic spectra shows the birdie drift rate (.gif):
Looking at all of the 1 Mhz comb birdies in 310-340 Mhz (.ps) (.pdf)
Birdie changes when we turn off the Old AC (.gif):
Birdie changes when we turn off the New AC (.gif):

#### Initial problem

The experiment a2057 was doing long integrations at 327.38 and 310.4 Mhz. After getting rid of the 16Khz comb (from the alfa motor controller) a large narrow birdie showed about in both of these bands. It wandered by many KHz.

#### Characterizing the birdies: (top)

• Fig 1  one second spectra: Black is polA, red is polB. The bandwidth is 195 Khz with a 95 hz resolution (no hanning smoothing). The top two plots are the full bandwidth 327.4 then 310.4. The bottom two plots are blowups around the birdies. The birdie is stronger in polA. At 95 hz rbw the birdie has not been resolved.
• Fig 2 variability over 1 hour: 1 second dumps were taken for  3600 seconds on 07oct05. This setup had 760 hz resolution (The amplitude is less than Fig 1 because of the wider channel width).
• Top: this shows the amplitude (polA)  vs time for the 327.4 Mhz birdie (black)  and the 310.4 Mhz birdie (red)
• 2nd, 3rd plot: The frequency of the peak versus time is plotted (black is 327.4 birdie, red is 310.4 Mhz birdie).  Both birdies are wandering by about 30 Khz over the 60 minutes.
• Bottom plot: The variation of the 1 Mhz fundamental with time. The black plot is 327.4/327, the read plot is 310.4/310. They lie on top of each other so they are coming from the same 1 Mhz fundamental. The variation of the 1 Mhz clock is 1 part in 10000 over 1 hour (not a great oscillator).  When looking at wider bandwidths, a stronger harmonic at 327.48 was also seen. Some of the harmonics were not present.
• Fig 3 azimuth dependence: On 11oct05 two az spins were done to check the azimuth dependence of the birdie. No azimuth dependence would bolster the case for the birdie coming from inside the dome.  The az was spun from 270 to 630 degrees at .4 degs/sec and then back to az=270. This took about 900 seconds for each direction. The channel width was 190 Hz. The amplitude of the peaks was plotted versus azimuth position. Black in is the clockwise spin while red is the counter clockwise spin.
• Top: The strength versus azimuth for the 327.38 Mhz birdie. The upper two lines are the birdie peak. The bottom two lines are the Tsys variation.
• Bottom: The azimuth variation of the 310.4 Mhz birdie.
There is no strong  azimuth dependence of the signal so it is probably coming from within the dome. The jumps up and down do not repeat in azimuth so they are probably a time variation.
• Dynamic spectra (600 seconds) from 07oct05 shows the birdie drift rate:
• Top Image:  The channel width is  3 khz. There is  a birdie at 327.38 Mhz and the next one at 328.48.
• Bottom image: The 327.38 birdie is blown up with 760 Hz channel width. The birdie is moving through 20 Khz in 400 seconds (50 hz/second).

•
• Looking at all of the 1 Mhz comb birdies in 310-340 Mhz (.ps) (.pdf)

•     On 12oct05 the 327 receiver band was scanned from 310 Mhz to 340 Mhz. 2 minute integrations were done with a 760 hz resolution and then the band was stepped up in frequency. It took 25 steps (50 minutes) to cover the entire band.  The band passes were removed and then the data was interpolated to a uniformly spaced grid (the individual sbc were overlapped). Two 1 Mhz combs were found in the data (one starting at 310.235 and the other one at 310.278). 40 Khz about each comb frequency was extracted and then plotted (with offsets for plotting).
• Fig 1: The comb elements from the first comb ( 310.235  Mhz start freq.). Each strip is 40 Khz centered on the 1 Mhz multiple. The top plot is polA while the bottom plot is polB. You can see the comb drifting over the 50 minutes. The vertical scale is in units of Tsys. The comb is many times Tsys in some of the comb elements.
• Fig 2: This is the second comb which started at 310.278. It is weaker than the first comb (10-20% of tsys). It also has a different drift rate.
Summary of the birdie characteristics: (top)
• The birdie drifts in frequency. In one hour it drifted by 30 Khz. The drift rate is not constant.
• The birdie is narrow (< 95 Hz ) in a 1 second integration. On set of base band sampled data showed it to be < 8 hz wide. The birdie width will depend on the drift rate/integration time.
• The birdie is strong. In 95 Hz it can get up to 8 times Tsys (at 327 Mhz).
• The birdie belongs to a 1 Mhz comb. It is not locked to the station clock (since it drifts so much). The oscillator stability is about 1 part 10^4.
• There are two combs present. One is close to 310.235  (and 1 mhz multiples). The second is near 310.278 Mhz (and 1 Mhz multiples). Because of the drift, these frequencies will  change with time. The two combs drift at different rates.
• Azimuth spins showed no azimuth dependence so the birdie is probably coming from inside the dome.
• Putting the receiver on load made the birdie go away. The initial strength was large enough so that it should have been see with the higher load temperature. So the birdie is coming in through the horn.

#### The birdies are coming from  the  air conditioning unit. (top)

On 12oct05 we looked in the turret room for the birdie. The setup was:
• In the dome we used a Tektronix portable spectrum analyzer with 100 hz rbw, 700K Tsys, and a small loop antenna.
• The correlator was simultaneously taking data through the 327 Mhz receiver (760 hz rbw).
We used the birdie frequency from the 327 receiver to track the birdie in the tektronix spectrum analyzer. The first pass through the room showed nothing. When the probe (which is not very sensitive) was raised up in the air close to the ac unit above the focal point (the old Ac unit) the birdie appeared.  The plots show the birdie changes when we stopped the AC  unit:
• Birdie changes when we turn off the Old AC (.gif): These are dynamic spectra.
• The top two plots show the 327.3 Mhz birdie and the 310.3 Mhz birdie. At second 120 we used the remote ac control to turn off the old ac unit.  At second 232 we turned the ac unit back on. The birdie reversed it's drift direction when the unit was turned off, and picked back up the original drift direction when it was turned back on.
• The bottom two plots are  when we shut off the breaker to the Old Ac unit. At second 118 we shut off the breaker to the old ac unit. At second 130 we turned it back on. When the breaker was off,  the birdie went away (or at least it was well off of this plot..).
• Birdie changes when we turn off the New AC (.gif):

•     On 14oct05 the old ac unit was shut off using the breaker (no power to the digital control unit). Data was taken with the correlator set to 390 hz resolution for 300 seconds.  At second 120 the NEW AC unit was shut off using the remote control unit. The breaker was not shut off so the digital control continued to run. You cans see that the birdie disappeared for a few seconds and then came back with a new drift rate. This shows that the NEW AC units also has a set of birdies. b
Shutting off the Ac unit caused the birdie to change drift directions but it did not go away. When the ac unit is turned off with the portable control, the compressor is shut off but the  digital control board in the AC unit remains on (so that it can tell when you want to turn it back on). When we shut off the breaker to the Old AC unit, the AC digital control unit is shut off. This caused the birdies to go away. So the birdies are coming from the digital control portion of the AC.

We need to change the AC unit from a digital control to an analog version (like we did in the vertex shelter). This should get rid of the birdies.

processing:  x101/051012/accomb/327acall.pro, 327acimages.pro, 327step.pro