2.2 Responsivity
2.2.5 Resonant Circuit
nonthermal quasi-particles.
Equations (2.59) and (2.60) give the desired relationship between the measurable quantities — the fre- quency and loss of the resonator — and the quasi-particle density of the superconducting thin film. The final step is to examine how the frequency and loss of the resonator are probed by the readout electronics.
DAC ADC
Figure 2.4: Diagram illustrating the basic principle behind the MUSIC readout electronics. Two digital-to- analog converters (DACs) output the real (in-phase or I) and imaginary (quadrature-phase or Q) components of a pre-programmed waveform that consists of a superposition of sinusoids (carriers or carrier tones) at base- band frequencies (−225 to 225 MHz). The absolute magnitude of the FFT of an example waveform is shown in the upper left panel; it has the appearance of a “frequency comb”. The FFT of this waveform is also shown in the complex plane in the lower left panel. Note that the phases of the carriers are randomized in order to prevent clipping and utilize as much of the dynamic range of the DACs as possible. The I and Q components are up-mixed with a local oscillator (LO) to microwave (RF) frequencies and sent through the cryostat on coaxial cable. The baseband frequencies are chosen so that each carrier is centered on the resonant frequency of an MKID after mixing. The MKIDs modulate both the amplitude and phase of the carriers. The carriers are then amplified by a cryogenic high-electron-mobility transistor (HEMT) and down-mixed with the same LO back to baseband frequencies. The I and Q components are digitized by two analog-to-digital converters (ADCs). A field-programmable gate array (FPGA) is used to FFT the raw ADC timestreams, select the bins corresponding to the carrier tones, and low-pass filter and decimate them to the desired sampling rate of 100 Hz. The final output is the I and Q components of all of the frequency bins that contain carrier tones, sampled at 100 Hz. An example of the output at a single time sample is shown in the right column. The difference between the left and right columns provides a measurement of the overall gain and phase delay through the system.
Figure 2.5:Top:Schematic of the MUSIC readout electronics board.Bottom: Picture of one of the MUSIC readout electronics board. Each board output 144 carrier tones and measures the complex amplitude of these tones at the output of the system at a rate of 100 Hz. Two boards are used to probe each detector array.
resonant frequency,S21res(fres) =1−Q/Qc=Q/Qiand it is maximally attenuated.
The squared magnitude of the transmission near resonance is given by
|S21res(f)|2=1− 1−(Q/Qi)2 1+4Q2
f−fres fres
2 . (2.64)
It is a Lorentzian dip from unity with depth 1−Q2/Q2i and full width at half maximum FWHM=2∆fres= fres/Q. We denote the half width at half maximum as∆fresand also refer to it as the resonator bandwidth. The magnitude|S21res(f)|and argumentθ=arg(S21res(f))are shown in the left panel of Figure 2.6.
It is useful to introduce the variables
x= f−fres fres
, y=Qx= f−fres
2∆fres
, (2.65)
wherexis the fractional detuning of the carrier tone from the resonant frequency andyis the normalized
Figure 2.6: The magnitude (upper left), phase (lower left), and complex transmission (right) near resonance.
Stars denote the location of the resonant frequency. The cross symbols in the right panel are separated by
∆f =10 kHz with increasing f corresponding to clockwise motion along the circle. The resonance curve changes from the solid black line to the dashed gray line when the optical loading is increased. Note that we measure the phase with respect to the complex origin, whereas many others measure the phase with respect to the center of the resonance circle. All curves were calculated using Equation (2.62) with parameters typical of a MUSIC resonator under sky loading: fres=3.2 GHz,Q=40,000, andQ/Qc=0.75.
detuning in terms of the full width at half maximum of the resonance. Equation (2.62) then becomes
S21res(f) =1− Q/Qc
1+2jy . (2.66)
In the previous section we showed that a change in the quasi-particle density results in a change in both the frequency and dissipation of the resonator. We now calculate how these perturbationsδffres
res andδQ1
i affect the transmission of the carrier tone. This is given by
δS21res= ∂S21res
∂Q1
i
! δ 1
Qi
+ ∂S21res
∂x
δx
= ∂S21res
∂Q1
i
! δ 1
Qi
− ∂S21res
∂x δfres
fres
, (2.67)
where in the last line we have used the fact thatδx≈ −δffres
res . The partial derivatives are evaluated using
Equation (2.62) to be
∂S21res
∂Q1
i
= Q2/Qc
(1+2jy)2 , ∂S21res
∂x = 2jQ2/Qc
(1+2jy)2 (2.68)
and therefore
δS21res= Q2/Qc (1+2jy)2
δ 1
Qi
−2jδfres
fres
. (2.69)
If the carrier is centered directly on the resonant frequency, theny=0 and the response is at a maximum. In this case perturbations to the resonator frequency (dissipation) will result in purely imaginary (real) changes in the transmission. The factor(1+2jy)−2encodes the effect of detuning on the small signal response. It is helpful to write it in the form(1+2jy)−2=χyejφywith
χy= 1
1+4y2 , φy=−2 arctan(2y). (2.70) If the carrier is mis-centered, then the response will be degraded by a factorχy. The frequency and dissipation response will still be orthogonal, and will still be oriented tangential and normal to the resonance curve.
However, this “frequency and dissipation basis” will be rotated with respect to the real and imaginary basis by an angleφy. It is traditional to define a coupling efficiency
χc= 4QQi
c
1+QQi
c
2 , (2.71)
which takes a maximum value of 1 when the coupling quality factor is matched to the internal quality factor, orQc=Qi. Equation (2.69) can then be written as
δS21res=1 4Qiχcχy
δ 1
Qi−2jδfres
fres
ejφy . (2.72)
This equation suggest three steps should be taken in order to maximize response to a fixed change in frequency and dissipation. The sources of loss intrinsic to the resonator — including the loss sourced by optically generated quasi-particles due to the background loading — should be minimized so that Qi is large. The coupling to the feedline should then be designed so that under typical loading conditionsQc≈Qi ensuring thatχc≈1. Finally the carrier tone should be centered on the resonant frequency so thatχy≈1. The coupling efficiencyχcand tuning efficiencyχyare plotted as a function of their respective arguments in Figure 2.7.
Equation (2.72) holds for slow perturbations to the resonator frequency and dissipation. For fast pertur- bations we must account for the resonator ring-down response. This effect can be described in the Fourier
Figure 2.7: The coupling efficiency (left) and tuning efficiency (right). TheFWHMof the resonator is defined as fres/Q.
domain as [105]
δSe21res(ν) =1
4QiχcχyHres(ν,f) δQei−1(ν)−2jδefres
fres
(ν)
!
ejφy. (2.73)
whereHres(ν,f)is the resonator transfer function given by
Hres(ν,f) =1−S21res(f+ν)
1−S21res(f) . (2.74)
If the carrier tone is centered on resonance this reduces to Hres(ν,fres) = 1
1+j∆fν
res
(2.75)
and the resonator acts as a low-pass filter with bandwidth∆fres=2Qfres. We are interested in signals at temporal frequenciesν≤50 Hz (the readout electronics sample at a rateνs=100 Hz). Since∆fres&20 kHz for the MUSIC resonators, the transfer function can safely be ignored for our purposes.
We now consider perturbations to the frequency and dissipation that are sourced by perturbations in the quasi-particle densityδnqp. In this case Equation (2.72) becomes
δS21res=1
4Qiχcχyα[κ1(T,ω,∆0) +jκ2(T,ω,∆0)]ejφyδnqp
=1
4Qiχcχyα|κ(T,ω,∆0)|ej Ψ(T,ω,∆0)+φyδnqp. (2.76) This can easily be converted into changes in the amplitude of the carrier tone using Equation (2.61). This is the quantity that is digitized by the readout electronics, which means that we have followed the propagation
of signal through the entire instrument. We are now in a position to derive an expression for the response to an astronomical source. Before doing that, we should address several complications to the model just presented.