We present measurements on two high frequency S(uperconductor)I(nsulator) S(uperconductor) heterodyne receivers at 660 GHz and 810 GHz. Their noise temperatures are 163 K and 860 K, respectively. The contribution of the high frequency (RF) input loss to this number is measured with the commonly used method of intersecting lines. The result is compared to a more rigorous analysis using the full quantum theory of mixing. We will show that in this frequency range the method of intersecting lines in general leads to an overestimation of the RF input loss. Even at 660 GHz, with a well matched SIS junction, the input noise contribution is overestimated by about 25 percent. This is due to the summation of the contributions due to the vacuum fluctuations, which is the highest contribution, the mismatch at the intermediate frequency (IF), and the fact that the mixer output noise is not completely independent of local oscillator (LO) power. At 810 GHz, due to a much stronger loss-induced mismatch of the SIS junction, the mixer output noise varies considerably with LO-power. The large loss of approximately 8 dB directly in front of the mixer enhances this contribution to the measured RF input loss. The contribution of the IF-mismatch is likewise enhanced. Together with the contribution of the vacuum fluctuations, this can lead to an overestimation of the RF input noise by more than 100 percent, depending on the actual mismatch of the mixers. We therefore conclude that, for the severe mismatch that occurs in Nb based SIS mixers above 700 GHz, the method of intersecting lines gives incorrect results and the full analysis should be applied.
We present measurements on two high frequency S(uperconductor)I(nsulator) S(uperconductor) heterodyne receivers at 660 GHz and 810 GHz. Their noise temperatures are 163 K and 860 K, respectively. The contribution of the high frequency (RF) input loss to this number is measured with the commonly used method of intersecting lines. The result is compared to a more rigorous analysis using the full quantum theory of mixing. We will show that in this frequency range the method of intersecting lines in general leads to an overestimation of the RF input loss. Even at 660 GHz, with a well matched SIS junction, the input noise contribution is overestimated by about 25 percent. This is due to the summation of the contributions due to the vacuum fluctuations, which is the highest contribution, the mismatch at the intermediate frequency (IF), and the fact that the mixer output noise is not completely independent of local oscillator (LO) power. At 810 GHz, due to a much stronger loss-induced mismatch of the SIS junction, the mixer output noise varies considerably with LO-power. The large loss of approximately 8 dB directly in front of the mixer enhances this contribution to the measured RF input loss. The contribution of the IF-mismatch is likewise enhanced. Together with the contribution of the vacuum fluctuations, this can lead to an overestimation of the RF input noise by more than 100 percent, depending on the actual mismatch of the mixers. We therefore conclude that, for the severe mismatch that occurs in Nb based SIS mixers above 700 GHz, the method of intersecting lines gives incorrect results and the full analysis should be applied.
Accurate knowledge of radio-frequency (RF) losses in front of superconductor--insulator--superconductor (SIS) mixers is essential for optimizing millimeter- and terahertz heterodyne receivers. However, such losses are difficult to extract from conventional hot/cold-load measurements because the measured receiver noise combines contributions from the input optics, the mixer, and the intermediate-frequency chain. In this work, we present a numerical implementation of the straight-line, or intersecting-lines, method for estimating input-referred RF noise and optical loss from standard Y-factor measurements. Hot/cold IF spectra are acquired while sweeping the local oscillator (LO) pumping level from under-pumped to over-pumped operation, so that each IF frequency bin gives a family of linear relations between receiver output power and input load temperature. We evaluate all pairwise intersections between these lines and analyze the crossing cloud to obtain the effective RF input noise. We repeat the analysis across the IF band and multiple LO tunings which provides a frequency-resolved characterisation of the input losses over the full band. Our implementation turns the straight-line technique into a reproducible framework for receiver characterisation and relative optical-loss comparison in low-loss front-end configurations.
In the assessment of a superconductor-insulator-superconductor (SIS) mixer, the breakdown of the overall receiver noise into intrinsic mixer noise and noises from other components is indispensable. The input noise due to losses in front of the mixer is conventionally measured by assessing the receiver noise as a linear function of mixer conversion loss at low local oscillator power and extrapolating the line to an infinite gain. However, it has been observed that this method tends to overestimate the input noise. To address this, the underlying theory of the input noise measurement method was revisited. It was found that the product of mixer gain and noise, i.e., the output noise of the mixer, is not a constant with respect to LO power. Therefore, extrapolating the linear relation to a large gain, where the input noise is inferred, becomes problematic. Numerical simulations were conducted to validate this analysis. Finally, a two-step measurement approach is introduced, in which an attenuator is inserted in front of the SIS mixer at the 4-K cold stage, to obtain the actual RF input noise.
A 600-700GHz SIS mixer with the twin-junction tuning circuit has been designed and fabricated [1] The surface impedance of thin-film superconducting microstrip lines, based on Mattis-Bardeen theory [2], is included in the optimization of RF impedance matching. The receiver noise temperature measured in the frequency range of 630-660GHz is below 200K and the lowest receiver noise temperature of 181K is achieved at 656GHz. The FTS response of the SIS mixer shows a good RF coupling from 600GHz to 700GHz. Both the noise performance and the FTS response can be quantitatively described by numerical results using the quantum theory of mixing. Some detail considerations on the mixer model calculation, such as spreading inductance around the junction tuning structure and mixer's embedding impedance, are discussed. Introduction Nb-based superconductor-insulator-superconductor (SIS) mixers have been used very successfully in sub-mm wavelength detection. Their receiver noise temperature has reached to three times the quantum limit, —3 hfrkB, below the junction's gap frequency. However, at high frequency bands, Nb SIS mixers need to be designed carefully because the photon energy is close to or higher than the gap energy of Nb, A-1.4meV. When the frequency band covers the gap frequency of Nb, —670GHz, the RF loss of Nb superconducting transmission lines increases dramatically and the surface impedance has strong frequency dependence. Usually, the RF properties of a superconducting transmission line can be described by the Mattis-Bardeen theory [2] in good accuracy. From results reported so far [4-9], the best receiver noise temperature is close to 5 hi7kB in the band of 600-700GHz. Despite of the input loss contributed by the experimental setup, some other properties of SIS mixers should be taken into account. In this paper, we characterize the performance of a 600-700GHz mixer by comparing the experimental and simulated results.
We report on a 850-GHz superconducting-insulator-superconducting (SIS) heterodyne receiver employing an RF-tuned niobium tunnel junction with a current density of 14 kA/cm^2, fabricated on a 1-µm Si_3N_4 supporting membrane. Since the mixer is designed to be operated well above the superconducting gap frequency of niobium (2Δ/h ≈ 690 GHz), special care has been taken to minimize niobium transmission-line losses. Both Fourier transform spectrometer (FTS) measurements of the direct detection performance and calculations of the IF output noise with the mixer operating in heterodyne mode, indicate an absorption loss in the niobium film of about 6.8 dB at 822 GHz. These results are in reasonably good agreement with the loss predicted by the Mattis-Bardeen theory in the extreme anomalous limit. From 800 to 830 GHz, we report uncorrected receiver noise temperatures of 518 or 514 K when we use Callen and Welton's law to calculate the input load temperatures. Over the same frequency range, the mixer has a 4-dB conversion loss and 265 K ± 10 K noise temperature. At 890 GHz, the sensitivity of the receiver has degraded to 900 K, which is primarily the result of increased niobium film loss in the RF matching network. When the mixer was cooled from 4.2 to 1.9 K, the receiver noise temperature improved about 20% 409-K double sideband (DSB). Approximately half of the receiver noise temperature improvement can be attributed to a lower mixer conversion loss, while the remainder is due to a reduction in the niobium film absorption loss. At 982 GHz, we measured a receiver noise temperature of 1916 K.
Abstract We developed a low-noise, compact, balanced superconductor–insulator–superconductor (SIS) mixer, operating in the 787–950 GHz radio frequency range. A waveguide mixer block was designed to integrate all the key components, such as a radio frequency (RF) 90° hybrid coupler, two identical SIS mixer chips, bias-tees, and an intermediate frequency power-combiner. The RF waveguide 90° hybrid coupler consists of branch lines with wide slots optimized by numerical simulation, for ease of fabrication. The balanced mixer was installed into a cartridge type receiver, originally developed for the Atacama Large Millimeter/submillimeter Array Band 10 (787–950 GHz). The receiver demonstrated double sideband noise temperatures of approximately 200 K for most of the band, without any correction for loss in front of the receiver. The local oscillator noise rejection ratio was estimated to be more than 15 dB within the measured frequency range.
A good understanding of input radio-frequency (RF) noise contribution is of particular importance to the development of high-sensitivity superconducting hot electron bolometer (HEB) receivers. In terms of simulation results based on the hot spot model, we found that the output noise of HEB mixers is nearly linearly decreased with the mixer's bath temperature when the HEB mixer is operated at its optimum DC-bias regime under appropriate local-oscillator signal pumping, and the effect of this decrease on the receiver output power can be well corrected. Thus, we used an intersecting lines technique to measure the input RF noise temperature of a quasi-optical superconducting NbN HEB receiver, and the measured input RF noise temperature of the HEB receiver is around 300 K at 850 GHz. For comparison, we also evaluated the losses and equivalent noise temperatures of the quasi-optical components in the RF signal path. The evaluated input RF noise temperature is found to be smaller than the measured one.