RFI shielding and mitigation techniques for a sensitive search for the 327 MHz line of Deuterium Alan E.E. Rogers, Joseph C. Carter, Preethi Pratap M.I.T. Haystack Observatory Westford, MA Marcos A. Diaz Department of Electrical and Computer Engineering, Boston, University, Boston, MA Why look for Deuterium? D/H ratios tell us about density of material in the early Universe → open vs. closed scenarios Optically, H and D are too spectrally close to differentiate between the two Overall view of the 25 5x5 crossed dipole stations of the Deuterium array RFI monitor Technical Developments Digital receiver developed for this project Allows deep integration Antenna design – crossed active dipoles 5x5 array of crossed dipoles on each station – 24 stations View of a part of one station with the RFI monitor (mounted on the trailer) with 12 active Yagis covering the horizon with 30 degree spacing, to detect and locate the source of the RFI. A portable “handheld” receiver with various antennas, including an active Yagi is also used to localize sources of RFI. Reduction of internal RFI Electronics put in and enclosed box Analog electronics within internal boxes Ferrite filters on USB, ethernet cables “Double” filtering of A.C. power Fiber optic data and control path 48 channel receiver for each station of the array – shown with cover removed Analog down converters Local Oscillator Digital boards Filtering to reduce intermod and images Stub filters on active antennas – rej. low freq 327 MHz output filter on preamp > 50 dB rej. 327 MHz input filter in receiver > 50 dB rej. 50 MHz I.F. filter > 50 dB rej. Digital down converter filter > 100 dB rej. rej. = out of band rejection Measures to reduce external RFI Minimized horizon response of array Dipoles are over a horizontal ground plane Parasitic directors (as seen in in photo) added to reduce gain at horizon by about 10 dB Clutter fence considered but large and expensive Negotiated with neighbors to replace certain consumer products which have clock harmonics close to MHz. Shielded buildings within 500m of array. This included shielding some windows. Coaxial stub filters form an integral part of the low noise active dipole antenna Antenna temperature from dipole over ground plane pointed at the zenith Frequency MHz noise from Galactic plane noise from Sun sky model 9 May 2002 Calibration of active antenna arrays using a sky brightness model, Rogers, Pratap, Kratzenberg, Diaz, Radio Science vol. 39, RS2023, 2004 RFI environment at Haystack Observatory RFI noise temperature near Haystack BW = 1 MHz integration 100 s RBP 8 Dec 03 (noise floor is limited by noise figure of spectrum analyzer) Sources of RFI at 327 MHz PC motherboard > 100 dB shielding needed Fiber optic ethernet converter > 100 dB req. Some digital answering machines with 5 km Other PC and electronics within 500 m. Continuum transients mostly of unknown origin. These have spectral features due to multipath. modems, display of spectrum analyzer, inadequately shielded network equipment, 100base-Tx to 100base- Fx ethernet converters, UPS power supplies Data processing techniques to mitigate RFI Transient excision - excise data for which any spectral or continuum signal exceeds the 8 sigma level in the RFI monitor. Spectral exclusion – exclude any 244 Hz frequency channels for which RFI is detected above 8 sigma in 24 hours integration. Weighted least squares smoothing to 2 kHz resolution is used to “interpolate” over the excluded frequency channels for each day’s integration. Days 2004_167 thru 2004_180 of array data – average of spectra from all elements of the array as a test of RFI amelioration. The line at the bottom of the plots show the amount of integration for each frequency. Note that the very deep integration corresponds to about 30 years observing with a single dipole. Sensitivity to detect* Continuous Wave (CW) RFI (in EIRP at 100m from array) RFI monitor active 12 dBi Yagi (Tsys = 200K) in 24 hours: dBm Array active dipole (Tsys = 100K, -10 dBi at horizon) in 24 hours: - 97 dBm Average of all 24x48 dipoles : dBm All dipoles in 10 days: dBm * assumes 10 sigma detection and resolution of 244 Hz EIRP = effective isotropic radiated power This project is funded by a grant from the National Science Foundation Why is it hard to detect? An example of the extreme sensitivity of Deuterium array to RFI Expected signal temperature: 500 microK in 10 kHz BW = -192 dBm Antenna Gain in sky direction: 22 dBi Antenna gain at horizon: ~ -10 dBi i.e. any RFI at -160 dBm or greater will be stronger than the Deuterium So that an RFI source at 100m LOS must be less than -97 dBm - which is 48 dB below the part 15 limit of -49 dBm into an isotropic ant. FM radio VHF TV UHF TV Ch 66 radar cellular HF digisonde TV The images show the RFI occupancy in two different domains: The image on the far left is the occupancy by frequency channel as a function of day number, the image on the right is a depiction of the transient detection as a function of time and day number. Frequency Day UT Time RFI (CW and transient) detection in 2003 No excision CW not detected by RFI monitor transients excised transients and CW removed continuum transients produce baseline ripple