radio telescope receivers Flashcards
by a total power telescope, we mean
something that turns the received radio power from a source into a proportional voltage that can be easily measured and recorded
components - antenna
turns the EM wave into a proportional voltage
this noise voltage has an approximately white spectrum with a power per Hz of KbTa
components - Low-noise preamplifier
boosts voltage by around a factor of 1000 so the microvolt signal from the antenna is now millivolts and strong enough not to be degraded or lost in further processing
components - filter
restricts the range in frequencies, defining the bandwidth
also helps to cut out interfering signals
components - mixer
shifts the band to a lower frequency by mixing (multiplying)the signal with a sinusoidal local oscillator
the signal can be shifted to baseband, around 0Hz
components - square law detector
we are interested in the power of the signal, not the wave voltage
power prop to v^2 so need to sqaure v
a suitable diode can do this or it can be done digitally
components - integrator (low-pass filter)
this averages the fluctuating ouput of the detector to determine its mean level accurately and hence improve the SNR
this can be done in analog electronics, with a low-pass filter or digitally
in practice what else is involved in total power radio telescope?
several filtering and mixing stages
the LNA is often cooled to reduce noise
drift scan
letting a source drift in and out of the beam of a fixed antenna
in radio astronomy, bandwidths are usually
narrow
delta v /v around 0.01
the output of the filter is
band-limited nosie
all the adjacent frequency components in the band beat together to give
a quasi-sinusoidal underlying waveform
an envelope fluctuating randomly
if we mix to baseband, the output of the detector is
the fluctuating envelope squared
self noise
uncertainty in the mean level due to the naturally fluctuating nature of the signal
what is the solution to self-noise
average the signal over time with the integrator
what is needed for averaging the signal over time with the integrator
samples must be statistically independent, so there is no advantage in a sampling period shorter than 1/ delta v
if the observation lasts for t, the number of independent measurements is
N= t/coherence time
=approx delta v t
the uncertainty in a measurement drops as
sqrt N
SNR=sqrt N = sqrt (delta v t) is only correct if
there is no other source of noise
in a perfect system, with only source noise present, the mean output is proportional to
the antenna temp from the source alone
in practical systems, additional factors contribute to the noise. EG
the LNA adds extra noise, making antenna temp appear as T_lna when no source is present
T_A from the antenna may itself contain many components
Ta= Tsource + Tbackground + Tatmosphere +Tground
we can distinguish Tsource form the rest by
chopping (beam-switching) on and off the source
if nothing else changes while switching then
Tsource+Ton - Toff
