hardware 3: phaselocks

Transcription

1 hardware 3: phaselocks 1. individual telescopes: focus incoming signals onto receivers 2. receivers: amplify signals, convert them to lower freq 3. correlator: detector and spectrometer

2 local oscillator system in order to convert the incoming signals to a lower frequency we must supply a local oscillator reference on each receiver the first LO is produced by a Gunn oscillator: basically a negacve resistor in a tunable cavity; mechanically tune cavity length to change freq the local oscillators on all antennas must be synchronized oscillacng at cycles per second, drii by less than 1 cycle over periods of hours how is this possible?

3 mm phaselock mechanical tuner Gunn osc GHz f = f 0 + α(δv) to SIS mixer coupler V 0 loop gain lowpass correction voltage ΔV = A sinφ phaselock mixer generates harmonics of YIG freq e.g , , MHz reference (from lab) mixer 50 MHz phaselock IF amp mixer GHz YIG osc cos(ωt) sin(ωt + φ)

4 mm reference frequency comes from YIG (Xband) phaselock tuning range GHz YIG osc GHz f = f 0 + α(δv) to mm phaselock coupler V 0 loop gain lowpass correction voltage ΔV harmonic generator produces harmonics of ref freq e.g , , MHz reference (from lab) mixer 10 MHz phaselock IF amp mixer harm gen GHz ref from freq synthesizer in lab

5 summary of phaselock system start with 10 MHz reference (from rubidium oscillator or H-maser or GPS receiver/time standard) send 3 signals to each telescope via underground fiberoptic cables: 10 MHz 50 MHz (actually, not exactly 50, for reasons to be discussed in a moment), ~10 MHz from DDS + 4 x 10 MHz ν synth ( GHz) generated by a frequency synthesizer phaselocked to 10 MHz on each telescope: ν YIG = n ν synth + 10 MHz ( GHz) ν Gunn = m ν YIG + 50 MHz ( GHz for 3mm, GHz for 1mm) for 1mm observations, send GHz to a frequency tripler (another nonlinear device) to generate GHz

6

7

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

9 fiberopcc transmission laser 1310 nm modulator photodiode Vout single mode optical fiber Vin input voltage modulates laser power photodiode output voltage is proportional to laser power, hence reproduces the input waveform very low loss through the fiber (80% transmission for 2 miles of fiber)

10 complication #1: temperature variations of the fiberoptic cables temperature coefficient of delay is ~ 10-5 /C most fiber is buried underground (thermally stable) but a ~65-foot length (100 nsec delay) runs up each antenna if this length changes temperature by 1 C: -> delay changes by 1 picosecond (10-12 sec) -> phase of ~1 GHz reference signal changes by > LO phase at 230 GHz changes by 80 how do we deal with this? observe phase calibrator every ~30 min to track instrumental phase linelength system continuously monitors the delay through the fiber; stored in data header; during data reduction, use linecal, then uvcat, to apply to your data

11 fiberopcc linelength system measures roundtrip phase at the reference frequency synth laser TRX fiber 1 LOref cpl RX phslck ref ϕ RX fiber 2 echo no electronics at the antenna, just a fiber coupler fiber1 and fiber2 are in the same gel filled tube; their temperatures track very closely

12 FiberopCc linelength system delay uncertainty ~ ±10 femtoseconds = 1 degree of phase at 230 GHz = ΔT = 1 C for 8 inches of fiber (or 0.1 millikelvin for 7000 I of fiber!)

13 linephase example shows effect of air conditioner cycling in rcvr cabin or base

14 linephase correccon example

15 1mm phase wraps in A array (due to Doppler tracking through very different fiber lengths)

16 complicacon #2: lobe rotacon suppose B= 10 meters B v velocity of ant 2 relacve to ant 1 v = 2πB/day LO tau corr tau LO = 2π(10m)/86400 sec = 0.72 mm/sec Doppler shift of signals at ant 2 relative to ant 1 due to Earth s rotation =.072 cm /sec cm /sec Hz = 0.24 Hz -> fringes wash out in a 4 second integration

17 lobe rotacon (= fringe tracking) wait! how does a 10 m single dish work if signals from the 2 edges arrive at the receiver with different frequencies? answer: the dish moves to track the source, cancelling the Doppler shiis same here: our telescopes are part of a large synthecc telescope; adjuscng local oscillator frequencies is part of tracking the source

18 offseing the LO frequencies in our example, we must operate the LOs at: ant 1: 100,000,000, Hz ant 2: 100,000,000, Hz to do this, send a slightly different phaselock reference signal to each antenna: ant 1: 50 MHz exactly ant 2: 50 MHz Hz the phaselock reference signal for each antenna is generated by DDS (direct digital synthesis) in the lobe rotator chassis (in the fiber entry room), then is sent to the receiver via optical fiber

19 lobe rotacon in the correlator Doppler correction LSB LO1 USB frequency each frequency within the passband has a slightly different Doppler shift; tracking LO1 only corrects the average phase the remaining phase variations change more slowly, can be removed in the correlator pipeline software

20 complicacon #3: sideband separacon RF LO IF RF IF shiiing LO frequency shiis USB,LSB IF frequencies in opposite direccons LO

21 sideband separacon for an interferometer we do not need to change the freq, just the phase 90 degree phase switch on LO1 allows automacc separacon of USB and LSB signals at the correlator note: we cannot separate USB, LSB noise; only signals that are common to a pair of antennas can be separated send out phase switch panern on 50 MHz phaselock reference

22