BINGO: Report for USP on Year 1 of Mike Peel’s fellowship

Mike Peel

1 Introduction

This document summarises some of the work I have done in the first year (Nov 2016–Nov2017) of my FAPESP Young Investigator fellowship on the project “BINGO: An ultra-sensitive survey of hydrogen and radio continuum emission to measure the Dark Universe”,based at IFUSP. This work has been done in collaboration with people at both IFUSP andINPE, Sao Jose dos Campos. This is not the same as my scientific report to FAPESP,which is due in April 2018.

The key points are:

• Significant work has gone into locating a new site for BINGO, which will now bebuilt in Paraıba.

• As a consequence, my other work on BINGO (such as the prototype receiver) hasbeen delayed, although it is now moving forward.

• I am concerned that there have been no opportunities for me to find potential studentsto work on BINGO and other radio astronomy projects this year.

2 Activities

A sample of my activities this year are below.

2.1 BINGO RFI measurements

One of the main BINGO activities that took place this year was the search for a newsite, after issues arose with the original site in Minas de Corrales, Uruguay, and politicalissues that made it favourable to have the telescope in Brazil instead of Uruguay. Asa result of this, I participated in a number of visits to potential sites to measure theirRFI environment. These included an INPE site in Rio Grande do Sul, a military site inArerungua, Uruguay, several possibilities near Cachoeira Paulista, in the state of So Paulo,and finally an number of different options across the state of Paraıba, northern Brazil.As well as participating in the site visits, this included technical work to fix issues with

1

the RFI measurement equipment (including issues with the amplifiers and the spectrumanalyser configuration), and most significantly, the analysis of the measurement data. andcomparison of the different options. Some example plots from this work are shown in Figure1, and the site selection work for BINGO will be the subject of an upcoming publication.

Full reports on this work are available on the BINGO wiki (or by request).

Figure 1: RFI measurements at a noisy site (Cachoeira Paulista, top-left), the bottomof the slope in Arerungua (top-right), at Vao do Gato (bottom-left) and Urubu (bottom-right). The red line shows the maximum value recorded (i.e., transients), while the blueline shows the average. Significant spikes seen in any line are a potential issue.

2.2 Aircraft route maps

Automatic dependent surveillance – broadcast (ADS-B) is an air traffic control signaltransmitted by modern airplanes at a frequency of 1090MHz. Signals are transmittedvery frequently, and include information such as the position, flight number, altitude andheading of the aircraft. Not all aircraft are currently equipped with ADS-B transmitters,but most commercial flights use them.

In order to look at the airplane densities around the potential sites in Paraıba, we usedADS-B USB dongles with small 5-inch antennas, attached to raspberry pi 3’s running the“piaware” software package and a small script to save the ADS-B data to disk. These are

2

Figure 2: Top: A heat map of planes detected using ADS-B measurements in Paraiba.The colour scale is logarithmic. Vao do Gato (left) and Urubu (right) are shown as blackdots. Bottom: Culmulative plots showing the fraction of time that a plane is within agiven distance from Vao do Gato (top) and Urubu (bottom).

3

the same systems used as a distributed network by various websites that track airplanesaround the world, however these may not be complete, and the fees to access their dataare large, which are not an issue with dedicated receivers. We installed two of these inParaıba, one on a rooftop in Sousa, and another on a rooftop in Campina Grande. Figure2.2 shows the results from 2 months of data overlaid on a map of northern Brazil.

To work out the fraction of time that we will have an aircraft within a given distanceof the site, we use the full dataset to calculate the closest airplane in every three-minuteperiod (sampled every minute) and then plot a normalised histogram of the result, whichis shown in Fig for Vo do Gato and Urubu. From this we can estimate that for 5% of thetime we will have an airplane within ∼30km of the site, and for 20% of the time there willbe one within ∼70km. However, this depends on flight routes, which may change in thefuture.

2.3 BINGO receiver prototype

The BINGO receiver takes the signals from the horns, and provides sufficient amplificationto allow the signals to be digitally recorded by the DAQ. At the same time, it has to bevery stable (very low 1/f noise), and has to be low noise while still operating at roomtemperature (unlike most radio receivers, which use cryogenic receivers, which are muchmore costly). In order to do this, we use newly-available amplifiers with very low noiselevels that operate at room temperature, and we nominally use a correlation receiver toprovide the necessary 1/f stability. The default receiver configuration is shown in Fig. 3for the waveguide components, and Fig. 4 for the RF components.

Figure 3: The layout of the waveguide components for BINGO, starting with a horn, whichleads to a polariser, which leads to a pair of magic tees around which the receiver moduleis then constructed (see Fig. 4).

4

Figure 4: The nominal BINGO receiver layout. A correlation receiver is used to amplifythe signals from a horn and from a colfet (acting as a reference signal). The horn, polariserand first hybrid (magic tee) are shown more clearly in Fig. 3.

The original concept was to use reference horns to provide a reference signal for thecorrelation receiver. However, the change of site also means a change in receiver design,as reference horns can no longer be used at the site in Paraiba. As such, the receiver willhave to either use a colfet as a reference for a correlation system, or alternatively we usea single radiometer chain rather than a correlation receiver. Work to determine whichsolution would be best here is ongoing; it also depends on the level of 1/f performancethat we need to achieve with the receiver, and the amount of money available to spendon this component of the project (since the correlation receivers are more expensive thansingle radiometer chains, both due to the magic tees and increased components needed, andbecause two output signals have to be measured rather than just one). Tests are ongoingin Manchester to determine this (see below).

The Brazil prototype receiver is being built around a pair of prototype BINGO filters,

5

Figure 5: The Brazil receiver prototype in November 2017, comprising filters for the BINGOfrequency band and amplifier chains.

produced by Phase 2, and a series of Minicircuits amplifiers. For the first stage we areusing Minicircuits ZX60-P33ULN amplifiers, which have a very low noise figure of 0.38dB,however these also have a low gain of 15dB. They offer remarkably similar gains and phasesbetween different amplifiers (see Fig. 6), which is essential for a correlation receiver.

As the second stage amplifiers, we then use Minicircuits ZX60-P162LN+, which havea medium noise figure of 0.5dB and around 20dB of gain. For later stages, we then useZX60-V63+ amplifiers, which have high noise figures of 3.6dB (which is fine at this pointin the receiver chain) but with 25dB gain; 2–3 of these will be used to provide the fullamplification needed. All of these amplifiers are from the same family, however they differin price from £85 for the first, to £70 for the second, to £40 for the third, making thisarrangement the most cost-effective while preserving the RF performance. These amplifiersare considerably cheaper than other options, such as Miteq or Kuhne, and consume verylittle electrical power.

The amplifiers are connected together using RG402 SMA cables. Tests are ongoing todetermine whether isolators or attenuators are needed between the amplifiers to providethe best matching between the components (necessary to avoid oscillations and standingwaves in the receiver).

This provides two single radiometer chains. In order to run tests of the full correlationreceiver, hybrids are necessary. The first hybrid has to be a magic tee, i.e., a waveguidehybrid. While prototypes of these are available in the UK, manufactured by Sylatech, forthe Brazil prototype we want to try to build these either in Brazil (possibly at Metalcardor Calfair), or in China, in order to reduce the costs. Negotiations for the design for

6

Figure 6: The S parameters of the ZX60-P33ULN amplifiers. Amplitude on the left, phaseon the right. The S12 parameter (gain) and the phases are impressively similar between 6measured devices. The BINGO frequencies are shown by the green shaded area.

these (from Phase 2) and for the manufacture of them are ongoing. The same applies tothe other waveguide components, such as the polariser, transitions, and also the BINGOfilters. We expect that Chris Radcliffe from Phase 2 will visit USP and INPE early in 2018to collaborate with the potential manufacturers of these.

While that is ongoing, the temperature stabilisation of the receiver will be worked on.This includes monitoring the temperatures of the different components, and constructingan insulated box to hold the receiver, and actively controlling the temperatures of parts ofthe receivers. Additionally, drivers for the phase switches (which are necessary to alternatethe outputs of the correlation receiver) are under construction.

3 Developments

In September 2017, I co-organised a SPAnet Radio Astronomy Workshop, which took placeat IAG. This event had 50 radio astronomers attending from across So Paulo and Brazil.The workshop was led by Isabel Aleman (IAG), with myself and Pedro Beaklini (IAG) asco-organisers. Full information about the workshop, including a report on the workshop,is at http://www.radiastranomia.net.

4 Final remarks

BINGO activities in this first year have gone substantially slower than planned, due to thechange of sites for the telescope and the necessary work to investigate the new options thatwere proposed. Ultimately a site in Paraıbahas been decided, and work is now progressingwith that site in mind.

As part of my fellowship application, I proposed a PhD student project working on the

7

BINGO receiver work. To date that has not been possible, as neither IFUSP or INPE hasoffered any opportunities to find PhD students. There have also been no opportunitiesto find masters students, or undergraduate students working on BINGO-related projects.This is of significant concern and I hope that opportunities become available in the comingyear.

5 Publications

Most of the work this year has been preparatory, hence why I have not published manypapers this year, although a large number of them are in preparation to be published nextyear, including:

• Dickinson et al., “The State-of-Play of Anomalous Microwave Emission (AME) Re-search” (submitted)

• BINGO collaboration, “The BINGO Telescope: Site selection and RFI measure-ments” (in prep, eta 2018)

• C-BASS collaboration, “The C-Band All-Sky Survey (C-BASS): Design and capabil-ities” (in preparation, eta 2018)

• C-BASS collaboration, “C-Band All Sky Survey (CBASS): Commissioning Observa-tions from the Northern telescope” (in preparation, eta 2018)

• C-BASS collaboration, “The C-Band All-Sky Survey (C-BASS): Constraining diffuseGalactic radio emission in the North Celestial Pole region” (in preparation, eta 2018)

• Planck collaboration, “Planck 2017 results. I. Overview, and the cosmological legacyof Planck” (in preparation, eta 2017)

• Planck collaboration, ”Planck 2017 results. II. Low Frequency Instrument data pro-cessing” (in preparation, eta 2017)

• Tibbs et al., “Planck Observations of M33” (in preparation, eta 2018)

6 Conference Proceedings

As most of my travel funding this year was spent on visiting potential BINGO sites, I onlyattended a single conference, SAB 2017, where we presented a series of three posters, andAlex Wuensche gave a keynote talk. The proceedings for these are in the process of beingsubmitted, and are:

• Abdalla et al., “The BINGO Telescope: The Science Case”, SAB 2017

8

• Fornazier et al., “The BINGO Telescope: Pipeline Description”, SAB 2017

• Peel et al., “The BINGO Telescope: Instrument Description”, SAB 2017

• Wuensche et al., “The BINGO Telescope: a new 21 cm window for exploring theDark Universe and other astrophysics”, SAB 2017

7 Seminars

• “The rules of BINGO: The technical requirements and how to meet them”, SPAnetRadio Astronomy Workshop, 28 September 2017

• I am scheduled to give an IAG colloquium on BINGO in early 2018.

8 Other activities

• I have worked through a lot of bureaucracy!

• I reviewed several papers for MNRAS and Science.

• Observations with ALMA (magnetic dust around L2 Puppis) and VLA (Anomalousmicrowave emission in NGC 2023) are pending, and should take place in the nextfew months.

• During the first part of this year I served on the Wikimedia Foundation’s FundsDissemination Committee, which included meetings in November 2016 and May 2017.My term on that committee has now come to an end.

• I attended the Wikimedia Conference in Berlin, Germany, in March 2017, funded bythe Wikimedia Foundation, to attend the Wikimedia strategy track.

• I attended Wikidatacon in Berlin, Germany, in October 2017, with a scholarshipfrom Wikimedia Deutschland. At the conference I presented on “Wikidata-poweredinfoboxes on the Basque, Catalan and English Wikipedias”.

• I gave a presentation on “Wikidata lists using Listeria” at Neuromat, USP, on 16November 2017.

9