1. Introduction

The EPI (Exploring the Physics of Inflation) project, Ref. CSD2010-00064 within the CONSOLIDER-Ingenio 2010 program framework, has the main engineering goals of developing a new instrument at 41 GHz (35 – 47 GHz band), called FGI (Forty-Gigahertz Instrument), and to build a second 3-m telescope to complement the capabilities of the QUIJOTE project, which is under operation with the MFI (Multi-Frequency Instrument, 10 – 14 GHz and 16 – 20 GHz) and the TGI (Thirty-Gigahertz Instrument, 26 – 36 GHz).

The new instrument at 41 GHz, safely below the 60-GHz oxygen absorption band, will significantly increment the sensitivity of the QUIJOTE project to detect the r parameter (tensor-to-scalar ratio). The reason for this is not only the significant reduction of noise due to the number of polarimeters that will incorporate but also the lower synchrotron signal from our galaxy expected at these higher frequencies.

Figure 1. Sky temperature expected at the observatory location (Izaña, Tenerife, Spain) with the bands of the different instruments defined with colored bars.

The QUIJOTE experiment operates at Izaña observatory, Tenerife, Spain, (see Fig. 2) with the first telescope and the MFI mounted on it so far. The second telescope, designed within the EPI project, is already installed in the observatory and it is waiting for the completion of the TGI. It is expected that the FGI will be installed in one telescope when the instrument will be completely assembled by the end of the EPI project. By that time, the MFI will have finished its measurements campaign and will be able to be exchanged.


Figure 2. Projects dome with the two telescopes (left) and the MFI operating at Izaña observatory (right).

2. Principles of Operation

The main scientific goal of the EPI project is to combine the data from the ESA Planck mission, the instruments of QUIJOTE project, and the new instrument at 41 GHz (FGI) in order to study the physics of the inflationary period of the universe. A special emphasis is put on the detection of the primordial Gravitational Wave Background (GWB) with the goal of reducing the uncertainty of the r parameter in about an order of magnitude.

These scientific goals demand the design and operation of very high sensitive polarimeters (receivers capable of measuring the polarization state of the incoming electromagnetic signal). Therefore, suitable polarimeter schemes were designed for the MFI, the TGI and finally the FGI. In the case of the FGI, the receiver scheme (hereinafter called pixel) is shown in Fig. 3.

Figure 3. Pixel scheme of the FGI.

The polarimeter obtains the signal polarization through the measurement of the so-called Stokes parameters I, Q and U simultaneously. The parameter V = 0 since it is assumed that the microwave background is linearly polarized. The linearly polarized incoming signal passes through the pixel feedhorn and reaches the polarizer and the orthomode transducer (OMT). At the OMT outputs there are two noise-like orthogonal signals proportional to the right-hand and left-hand components of the incoming signal, Er and El respectively. These two signals are amplified in the cryogenic low-noise amplifiers in the Front-End Module (FEM). These amplifiers are the key components of the receiver in order to determine the low noise performance and therefore to obtain a very sensitive instrument.

In the Back-End Module (BEM), which operates at room temperature, the signals are further amplified and filtered to define the operational bandwidth. A phase-switch module before the detection stage introduces different relative phase differences between pixel branches which help to minimize systematic errors in the receiver. Finally, the signals reach the detection module where they are correlated in two hybrid couplers and detected in Schottky diode detectors before being amplified with video amplifiers to accommodate the signals levels to the Data Acquisition System (DAS).

According with the scheme of Fig. 3, and assuming that the relative phase between pixel branches is zero, then

These signals, which are easy to obtain with simple mathematical calculations in the DAS, are proportional to the Stokes parameters, defined in a circular reference system a

Therefore, it is clear that

When the phase switches change their states in the module sixteen different states appear, four different relative phase differences repeated four times each. These changes produce that the Stokes parameters are obtained from the combination of different outputs through the whole cycle of the sixteen states, adding redundancy and therefore making the pixel less sensitive to systematic errors.

3. Number of Pixels

As stated before, the achievement of the scientific goals requires the use of very high sensitive receivers. In order to improve the instrument sensitivity even more, the number of pixels needs to be increased, since the number of pixels, N, and the instrument sensitivity are closely related as shown in (11), the radiometer’s equation.


Where K is a constant of proportionality which depends on the receiver configuration, Tsys is the system noise temperature, B is the pixel bandwidth, and t is the integration time, which is the time that the receiver is taking measurements.

From (11), it is clear that the larger the number of pixels is the smaller the temperature difference can be measured, that is, the higher the sensitivity is. Following this, the FGI instrument is equipped with 29 pixels, which is the maximum number of pixels that can be accommodated in a QUIJOTE-like cryostat maintaining the hardware compatibility between QUIJOTE and EPI projects. This compatibility is advisable to reduce the design and manufacturing costs of the project.

4. Pixel Subsystems

4.1. Waveguide components before the Front-End Module

4.1.1. Feedhorn

The first component of the pixel, according with Fig. 3, is the feedhorn. It is a corrugated horn antenna designed to have more than 20 dBi of gain, very low cross-polarization values, more than 25 dB of return losses at the input port, and a Guassian profile with low side-lobe levels.


Figure 4. Artist view of the EPI feedhorn cross-section with main dimensions.

Figure 5. Measured data from a typical FGI feedhorn: performance vs. frequency (a); radiation patterns at 41 GHz (b).

4.1.2. Polarizer

The input signal at the feedhorn passes through a polarizer, which is a section of square waveguide provided with suitable designed ridges in its internal walls in such a way that the orthogonal components of the signal are 90° out of phase at the polarizer output. If the polarizer is placed within the pixel with its reference axis rotated 45° regarding the OMT reference system, then the orthogonal components of the input signal at the feedhorn output are converted into the right-hand and left-hand circular components at the polarizer output.

Figure 6. Artist view of the EPI polarizer cross-section with main dimensions.

Figure 7. Measured performance of one manufactured polarizer: port reflection following orthogonal axis (a); and phase difference between orthogonal signals at the output (b).


4.1.3. Orthomode Transducer (OMT)

An orthomode transducer with in-phase outputs separates the left-hand and right-hand circular components of the incoming signal providing suitable signals to the receiver for the calculation of the Stokes parameters.

Figure 8. Artist view of the OMT internal configuration.

Figure 9. Measured performance of one manufactured OMT: rectangular ports reflection and isolation between these ports (a); and insertion losses (b).


The previous subsystems are connected using suitable designed octagonal-shaped square-to-circular waveguide transitions in a very compact way. Fig. 10 shows two sets of subsystems fully assembled and 20 units of polarizers and OMTs.


Figure 10. Two sets of feedhorn, polarizer and OMT assembled (a) and 20 units of polarizers and OMTs (b).

4.2. Cryogenic Low-Noise Amplifiers (Cryo-LNA)

A key component in the instrument performance is the cryo-LNA. It has to provide very high gain to minimize the noise contribution of the subsequent components while its own noise has to be kept as low as possible. For this reason, these amplifiers are cooled to cryogenic temperatures, around 20 K (-253 °C).

The cryo-LNAs designed for the FGI are assembled with two MMIC LNAs from the IAF (Fraunhofer Institute, Freiburg, Germany) and a microstrip equalizer in between. Each MMIC LNA is a four-stage design processed in mHEMT technology with 100 nm gate length. These devices are placed in a gold-plated aluminum chassis provided with WR-22 waveguide ports.




Figure 11. Pictures of some cryo-LNA prototype: 20 units assembled and ready for delivery (a); close view of the RF cavity (b).

Figure 12. Noise temperature of 60 units of cryo-LNAs measured under cryogenic conditions.


4.3. Gain and Filtering Modules

The first subsystem in the BEM is a module that provides further amplification and filters the signal to define the pixel effective bandwidth. The filter has been designed in microstrip technology on Alumina substrate. The microstrip technology enables the band-pass definition with low sharpness, which helps to compensate the bandwidth limitations of other subsystems in order to maintain the required effective bandwidth.



Figure 13. Picture of the filter prototype assembled with Jmicro transitions (a); and measured performance (b).


These gain modules are completed with two commercial MMIC LNAs from OMMIC and an equalizer that enables to obtain a positive slope in the gain curve so some of the frequency dependent losses of the whole receiver chain can be compensated. In Fig. 14 a picture of three units of these gain modules is presented, together with the measured gain performance of 60 units.




Figure 14. Picture of three units of the gain modules (a) and measured gain of 60 units (b).

4.4. Phase Switches Module

A total number of 30 units of Phase Switches Modules has been assembled and tested to be included in the BEM racks. These modules includes two branches with a home-design four phase state circuit (0º, 90º, 180º and 270º) in each one. The modules provide full switching capability by using TTL drivers. Fig. 15 presents a picture of some units of this module and the phase difference performance of one unit.



Figure 15. Picture of 27 units of the Phase Switches Module (a); typical phase difference performance of one unit (b); picture of the four-state phase shifter circuit(c).


4.5. Correlation and Detection Module

This module provides the output detected voltages (Vd1 to Vd4) correlating the input signals, proportional to Er and El, and detecting those signals using Schottky diode detectors. The module is completely designed in waveguide technology but the detectors, which are designed in microstrip technology on Alumina substrate.

The electrical scheme of the module was shown in Fig. 3. The input hybrid couplers separate the input signals with same magnitude and 90° out of phase. Therefore, the remaining input in each hybrid coupler is loaded with an absorber material (Eccosorb MF124) suitably shaped to act as a waveguide load. Two additional hybrid couplers in the module produce the correlation operations, sum and subtraction, between the input signals. A 90° phase shifter in one branch enables to obtain the signals that are required to calculate the Stokes parameters according with Section 2.

Figure 16. Picture of a correlation and detection module.

The module contains four microwave detectors based on HSCH-9161 Schottky diode from Agilent Technologies developed in microstrip technology on Alumina substrate. They convert the radiofrequency signals into DC measurable voltages. The detectors have been designed and manufactured using 50 Ohm/square resistive layer transmission lines as a solution to simultaneously provide a flat sensitivity response and good return loss over the operating bandwidth of the receiver. A single unit of the detector and the sensitivity response in the frequency range from 33 to 49 GHz are shown in Fig. 17.



Figure 17. Picture of the Schottky diode detector prototype (a); and measured sensitivity performance with an input power of -30 dBm (b).


After the detectors, the detected signals are amplified using video amplifiers (not shown in Fig. 16, they are assembled in the bottom side) in a differential configuration. Each video amplifier follows the electrical scheme presented in Fig. 18. A variable resistor (potentiometer) within the circuit enables to vary the gain of the video amplifier, which helps to accommodate the signal level to the DAS requirements.

Figure 18. Electrical scheme of the video amplifier for each output.

A frequency sweep test has been carried out to the correlation and detection modules finding the typical results shown in Fig. 19, together with a picture of 16 units of them. For this test, the inputs were two linearly polarized noise-like signals with same amplitude and 90° out of phase; therefore, the detected voltages definition are exchanged regarding (1) – (10).



Figure 18. Measured detected voltages at the correlation and detection module outputs (a); simulated (ADS) detected voltages with a phase error of 20° in the 90° phase shifter (b).


5. Power Budget and Output Voltages of the Radiometer

Following, the signals power levels and subsystems contributions are used to calculate the expected voltages at the pixel output working in a real environment in order to check the receiver configuration and its suitability to obtain adequate levels that can be used to extract the Stokes parameters; that is, the power budget. This calculation is based on realistic expected input signals and measured subsystems performance.

The input power to the FEM is calculated as:

Where k is the Boltzmann’s constant (1.38·10-23 J/K), Tsys is the system noise temperature which includes all the contributions, and Beff is the effective bandwidth, which has been calculated with the available data so far, obtaining around 12.6 GHz.

The calculation of Tsys requires considering different contributions: the sky temperature, the antenna spillover, the noise of the waveguide components due to their losses, the cryo-LNAs noise temperature, and the BEM contribution which is greatly minimized by the cryo-LNAs gain.

According with the plot in Fig. 1, which was made with data taken at Izaña observatory, the sky temperature at the center frequency is around Tsky = 15 K. The antenna spillover has been estimated in 5 K; the contribution of all the waveguide components cooled at cryogenic temperatures has been calculated to be around 5 K; the cryo-LNAs have a mean noise temperature of 23 K and a gain higher than 45 dB which makes the contribution of the subsequent components negligible. Therefore, the Tsys is around 48 K.

The obtained input power is around Pin = -81 dBm. This power is amplified by the FEM gain and the BEM gain, whereas the coaxial cables of the FEM-BEM connection and the phase switches module introduce noticeable losses resulting in a power level around -20 dBm at the correlation and detection module input. Considering the module losses, the detector sensitivity (around 1100 mV/mW), and the gain of the video amplifier, the expected voltage at the module output, ready to be digitalized in the DAS, is around 1.5 V (this voltage corresponds to the output with maximum value).


6. Integration in the Back-End Racks

Once that all the subsystems of the BEM are assembled, they are connected together forming a receiver chain. This receiver chain is integrated in a dedicated PCB that enables to bias all the subsystems and to extract the detected voltages. In order to connect these modules, waveguide sections were acquired from a hardware provider. A total number of 122 WR-22 waveguide sections were bought to this provider. A picture of all these sections is shown in Fig. 19 together with a picture of two receiver chains fully assembled on a dedicated PCB.



Fig. 19. Picture of the 122 WR-22 waveguide sections (a); picture of two BEM receiver chains assembled on the PCBs (b).

The connection between the cryostat and the BEM is carried out using coaxial cables. Since the BEM inputs (BEM Gain Modules inputs) were designed in waveguide technology (WR22 type), coaxial-to-WR22 transitions were required during the assembly. Due to the high cost of commercial transitions and the large number that were needed (60 units), a decision to design and assembly these transitions was taken. The transitions were manufactured at the IFCA workshop and assembled and tested at DICOM. Fig. 20 shows a picture of several units of these transitions and a plot of the input matching performance of all the transitions.



Fig. 20. Picture of 40 coaxial-to-WR22 transitions (a); input matching performance of all units (b).


Due to the larger thickness of FGI modules compared with the TGI modules (WR22 flange dimensions are larger than WR28 flange dimensions), the total number of receiver chains that could be installed in a single 19-inch rack was 10 (instead of 16 of the TGI). Therefore, for the FGI, three BEM racks were needed, assembled and delivered to the IAC. These racks also incorporates the same thermal control capability that was requested for the TGI ones. On the other hand, the FGI racks include some improvements regarding the TGI units, for example they have individual current test points and individual TTL signal connectors, avoiding interferences and mechanical constraints. Fig. 21 presents some pictures of these racks.



Fig. 21. Picture of the front panel of rack#3 (a); picture of the rear panel of rack#3 (b); picture of the internal configuration of rack#2 (c); detailed view of some temperature sensors and thermal mats within the rack (d).


Before being sent to the IAC, these racks were tested at the DICOM laboratory under representative conditions, i.e. input signals are similar to those expected during normal operation except for the power levels that are much higher than in the real conditions. In Fig. 22 a picture of the measurement setup is presented and a plot of the typical performance under these conditions is shown.


Fig. 22. Picture of the BEM rack measurement setup (a); plot with the detected voltages (b).



7. Data Acquisition System


8. FGI Calibration