Technical Aspects of a FASR Design

This telescope is very unlike most other radio telescopes. In particular:

Sensitivity is not a driving force.

Take advantage of this to allow a

SIMPLER design, implying HIGHER RELIABILITY

A FASTER design to build and debug: optimization for sensitivity is not needed, allowing compromises in the design that will have no affect on the science

A CHEAPER design

RFI

Likely to be dominated by satellites at many frequencies, both GSO and non-GSO.

Problem of the GSO belt. For several days, twice a year, the sun may be unobservable at certain frequencies as it passes behind the GSO belt, and strong GSO satellite signals come into the main beam of the FASR antennas.

RFI excision: before or after correlation?

Does RFI drive the digitization precision?

Dynamic Range

High imaging dynamic range is required

This is the scientific output

The source emission has high variability: >10000:1

The receiver will have to cope with very high values of interference signals

Slide 5

Slide 6

Signal Processing & Transmission

Send back signals from the antennas as analog or digital? As sub-bands?

Limited dynamic range of analog fiber-optic transmission.

Digitization of the strongly varying signal:

1. Apply AGC before the digitizer, calibrate later, or

2. No AGC, keep gain constant, but allow enough bits.

There’s a lot to be said for 1-bit digitization. (Loss of S/N isn’t an issue.) Can we get away with it?

Calibration

Needs care. 1% precision over years, to measure secular changes, will be difficult.

Calibration needs may dictate the size of antennas.

Not a show-stopper, but needs to be thought through carefully.

Field of View and Size of Antennas

If primary beam of antenna is >> solar diameter, then imaging is easy.

If primary beam of antenna is << solar diameter, imaging of small regions in the disk is relatively easy.

If primary beam of antenna is comparable to the solar diameter, good imaging is hard.

Pointing is most critical

Short spacing measurements become very important for good imaging

Short spacing measurements are hard to obtain in this regime (can’t get the antennas close enough together).

Calibration may set a minimum size of antenna.

An important design choice, tied to array configuration.

Noise

Antenna A: rx noise voltage A, signal voltage a

Antenna B: rx noise voltage B, signal voltage b

After correlation: (A+a).(B+b) = A.B + a.b + A.b + B.a

Only a.b may have a non-zero expectation value. This is the desired, correlated signal. The other terms give purely “random” noise.

In conventional interferometry, A>>a and B>>b, so A.b and B.a are negligible compared with A.B . The random noise on the image is dominated by A.B . The noise is statistically constant.

With FASR, usually a>>A and b>>B, so A.B becomes negligible, and the noise on the image is dominated by A.b, B.a & a.b .

Since a and b are also highly variable with time, this can give noise in an image strange properties. Some imaging algorithms may not work as expected.

Biggest potential show-stopper of a FASR Design:

This telescope is very unlike most other radio telescopes. In particular:

Sensitivity is not a driving force.

Take advantage of this to allow a

SIMPLER design, implying HIGHER RELIABILITY

A FASTER design to build and debug: optimization for sensitivity is not needed, allowing compromises in the design that will have no affect on the science

A CHEAPER design


This telescope is very unlike most other radio telescopes. In particular:
	Sensitivity is not a driving force.
		Take advantage of this to allow a
			SIMPLER design, implying HIGHER RELIABILITY
			A FASTER design to build and debug: optimization for sensitivity is not needed, allowing compromises in the design that will have no affect on the science
			A CHEAPER design


	Likely to be dominated by satellites at many frequencies, both GSO and non-GSO.
	Problem of the GSO belt. For several days, twice a year, the sun may be unobservable at certain frequencies as it passes behind the GSO belt, and strong GSO satellite signals come into the main beam of the FASR antennas.
	RFI excision: before or after correlation?
	Does RFI drive the digitization precision?


	High imaging dynamic range is required
		This is the scientific output
	The source emission has high variability: >10000:1
	The receiver will have to cope with very high values of interference signals


	Send back signals from the antennas as analog or digital? As sub-bands?
	Limited dynamic range of analog fiber-optic transmission.
	Digitization of the strongly varying signal:
		1. Apply AGC before the digitizer, calibrate later, or
		2. No AGC, keep gain constant, but allow enough bits.
	There’s a lot to be said for 1-bit digitization. (Loss of S/N isn’t an issue.) Can we get away with it?


	Needs care. 1% precision over years, to measure secular changes, will be difficult.
	Calibration needs may dictate the size of antennas.
	Not a show-stopper, but needs to be thought through carefully.


	If primary beam of antenna is >> solar diameter, then imaging is easy.
	If primary beam of antenna is << solar diameter, imaging of small regions in the disk is relatively easy.
	If primary beam of antenna is comparable to the solar diameter, good imaging is hard.
		Pointing is most critical
		Short spacing measurements become very important for good imaging
		Short spacing measurements are hard to obtain in this regime (can’t get the antennas close enough together).
	Calibration may set a minimum size of antenna.
	An important design choice, tied to array configuration.


	Antenna A: rx noise voltage A, signal voltage a
	Antenna B: rx noise voltage B, signal voltage b
	After correlation: (A+a).(B+b) = A.B + a.b + A.b + B.a
		Only a.b may have a non-zero expectation value. This is the desired, correlated signal. The other terms give purely “random” noise.
		In conventional interferometry, A>>a and B>>b, so A.b and B.a are negligible compared with A.B . The random noise on the image is dominated by A.B . The noise is statistically constant.
		With FASR, usually a>>A and b>>B, so A.B becomes negligible, and the noise on the image is dominated by A.b, B.a & a.b .
	Since a and b are also highly variable with time, this can give noise in an image strange properties. Some imaging algorithms may not work as expected.