Ideal Backend for E-ALFA

Desired Specifications Current/Extended WAPP capabilities Loss from not Achieving Desired Specs Comments
200 MHz Backend 100MHz/200MHz One of the proposed projects - the Very Deep Survey - cannot be performed without 200MHz coverage. Additionally, the other deep survey would benefit considerably from the increased bandwidth coverage. The deep surveys are considered by many members of the consortia to be defining projects for ALFA, in that they will push ALFA's limits more than the other proposed surveys. These surveys will, amoung other goals, allow for examination of the HI mass function with redshift and environment. It should be emphasized that no other surveys have been able to probe the HI universe to the sensitivity ALFA will, and this will finally allow us to investigate the higher redshift HI universe. Without 200MHz this will not be possible.
9-level/ at least 12-bit sampling
100MHz WAPP 200MHz WAPP
(MHz/chan) (MHz/chan)
100/4096:
3-level/4bit
200/4096:
3-level/4bit
100/2048:
3-level Stokes
200/2048:
3-level Stokes
≤50/8192:
3-level/4bit
≤100/8192:
3-level/4bit
≤50/2048:
9-level/4bit
≤100/2048:
9-level/4bit
≤50/4096:
3-level Stokes
≤100/4096:
3-level Stokes
Degradation in the sampling will results in considerable bandwidth loss due to RFI. The need for high sampling could be alleviated somewhat with the presence of:
  • Narrow band filters &/or blankers which block the known RFI (however, the restrictions these put on the usable bandwidth should be considered)
  • Flexible scheduling so that observations can take place when there is no RFI outside the filters/blankers
Minimum of 8192 channels/polarization (at max bandwidth, sampling) 4096/8192
chan/polarization
(at Max BW-> 5.2km/s res. at 1420MHz)
Inability to obtain desired science results & decreased usable bandwidth This is need for two reasons:
(1) This channel resolution (e.g. 5 km/s at 1420MHz for 200MHz bandwidth) is the minimum needed to get reasonable science results
(2) Any lower resolution will simply increase the band wiped out by RFI.
2 Polarizations per Receiver (simultaneaously observed) 2 pol/2 pol Observations would require 1.4x longer to acheive the same sensitivity This if for both RFI excision and noise reduction
<1 ms blanking of data No Banking capabilites planned (but could be put in place separately) Time / sensitivity will be lost for all experiments as well as observing efficiency <1 ms blanking allows for considerably more efficient RFI excision. While we are not interested in storing <1ms data, if we cannot excise RFI at this rate could be forced to dump considerably more data than is necessary. Additionally, even with RFI filters, fast dumping would allow for observing to occur even when there is unfiltered RFI in the spectra (assuming it doesn't saturate the IF, of course).
1-2 Hz dump rate for data 1 Hz/1 Hz Time/sensitivity could be lost for all experiments as well as observing efficiency 1-2 Hz dump rate (combined with the <1ms blanking) would allow for efficient processing of the data, and excising of any 'leftover' RFI in the data without losing too much of the useful data.
Cross-products & Reference horns for RFI mitigation No Cross-products available
No Reference Horns Planned
Loss in bandwidth and time efficiency To properly allow for interference mitigation would require a minimum of 2 reference horns with all cross-products between reference horns and ALFA feeds, and could be even more effective with 2 reference horns and cross-products between everything.