Receivers for

VLBI2010VLBI2010FFRF Tutorial by

Tom Clark, NASA/GSFC & NVI

Wettzell, March 19, 2009

There is no fundamental difference between the receivers for

PRIME FOCUS&

CASSEGRAINExcept for:• the beamwidth of the feed, • the sense of circular polarization,• issues of access, maintenance,• etc

Therefore we can talk about GENERIC receiver architecture. One of the prototype 12M ALMA

antennas at the VLBA site.

There are many new receiver developments relevant to VLBI2010:• The ATA array @ Hat Creek with 42 • 6M dishes covering 0.5-11 GHz,• Several SKA projects like Meerkat,• Sandy Weinreb’s Cryo LNA’s,• The NASA/Haystack VLBI2010 Prototypes,• NRAO’s E-VLA upgrade• etc

The “Old” Generic S/X Mark-3/4/5Geodetic Receiver Chain

DualFreq

XLNAs

Phase & noise CAL

120-900MHz IF

Baseband Converters

CAL

Formatter

FreqFeed

S

LNAs

SmallCryogenic

Refrigerator

LO H-Maser Clock

Recorder

DigitalAnalog

A typical, “old standard” Cryo S/X Receiver

Fairbanks, Alaska 1984

The Next Generation: VLBI2010• Our old, analog hardware has become very

difficult to maintain– The “Unobtanium” Problem - Expensive

• Geodesy wants more precise measurements– Need new, fast telescopes with wider bandwidths

• Many technology advances have come from the • Many technology advances have come from the Radio Astronomy community– Arrays (ATA, SKA, EVLA) - New PHEMT LNAs

– Wideband Feeds - Mark-5 & E-VLBI

– FPGA developments

• Bad RFI (especially @ S-band=2.2-2.4 GHz)

Technology advances (especially digital) have re-defined the “best” approach from 25 years ago !

The ATA is a Showpiece for the technology relevant to VLBI2010

Microwave Fiber Optics

42 6M Antennas

to control Microwave Fiber Optics direct from feed

iBOB FPGA modules used as PFB (Polyphase Filter Bank)

& Correlator

Dual Polarization 0.5-11 GHz feed

to control room

The New Haystack/GSFC Receiver

VLNAs

CAL

DBC = DigitalBaseband

Converters

CAL

Dual Pol’n2-15 GHz

TX

COTS µλ

RF/fiber/RF

RX

UDC = Up/Dn Conv

ADCROACH

H-Maser Clock

H

LNAs

Cryogenic Refrigerator

Mark 5& eVLBI

Digital

Analog

2-15 GHz Wideband

FeedTX

RX

UDC = Up/Dn Conv

ADC

The New Haystack/GSFC Receiver

The NRAO EVLA Receiver

VLNAs

NOISE CAL

PFBSeveral Dual Pol’n

TX

DigitalFiber

RX

ADCROACH

An

alo

g

Up

/Do

wn

Co

nve

rte

r

TX

RX

Master Time/Freq

Clock

H

LNAs

Cryogenic Refrigerator

EVLA Correlator

DigitalAnalog

Dual Pol’nFeeds T

XRXADCA

na

log

Up

/Do

wn

Co

nve

rte

r

PFB

A Few Comments about Phase Cal

Consider a set of very narrow pulses (i.e. delta functions) spaced ∆t in the time domain:

t

In the frequency domain, we will see the Fourier transform of these pulses, a set of “rails” in the frequency domain, with a spacing ∆f = 1/ ∆t

f

Receiver Dynamic Range & RFI

• From Thermodynamics &Boltzman: P=kTB• Where T is Temp in ºK, B is bandwidth in Hz &

Boltzman’s k = 1.38 x 10-23 W/(K*Hz)• Engineers find a more convenient form is

• Power in dBm = -198.6 dBm + B in dBHz + T in dBK

• For Sandy’s Cryo LNA with 100 ºK input over 10 • For Sandy’s Cryo LNA with 100 ºK input over 10 GHz

• PIN = -198.6 + 100 dBHz + 20 dBK = -78 dBm– The LNA has ~35 dB gain, so Pout is ~-48dBm

• Sandy indicates the amplifier has -5 dBm output

Hence we have only ~38 dB of “headroom” before we experience overload!

The Tunnel Diode has a negative resistance region

Curr

ent

I

~200 ~300 mv

Voltage V

t

When a tunnel diode is driven with a sine wave (like @ 5 MHz), this yields a bipolar series of pulses in the time domain:

• We feed the pulse train directly into the front end of the receiver along with the Quasar’s RF energy.

• In the time domain, we can think of these pulses as constituting VLBI’s reference clock which is used to “time tag” the observable geometric delay.

• We must “line up” the phases at different observing frequencies in order to determine the group delay for the radio source.

• The pulses must to be unipolar and they must have fast rise time – if the rise time is 20 psec, then

• The pulses must to be unipolar and they must have fast rise time – if the rise time is 20 psec, then the RF spectrum will be useful up to ~10-12 GHz.

• Alan Rogers “invented” the first suitable pulse generator in the 1970’s using a Tunnel (Esaki) Diode extracted from a Tektronix Sampling Oscilloscope.

• Some groups have used “Snap” or “Step Recovery” diodes, but they have a high temperature sensitivity problem.

• The “raw” pulse train is bipolar (which contains only odd harmonics ! !).

• So it is desirable to get rid of one polarity.

• Also we have normally reduced the rate by discarding 4 + 5 pulses with a series microwave switch. For a 5 MHz input, the gated unipolar pulse rate is 1 MHzpulse rate is 1 MHz

t

The Phase Cal Problem --

• Microwave Tunnel diodes now seem to be made of that rare metal known as “unobtainium”

• The Tunnel Diode mounting fixture had to • The Tunnel Diode mounting fixture had to be tweaked by Alan Rogers by hand.

• The Tunnel Diode pulsers don’t work well above ~22 GHz

• So we wanted a new approach.

The New “Digital” Phase CalAbout 1½ years ago, I learned that Hittite had

announced a new line of 13 Gb/s digital logic. I was interested in the HMC-672 and/nand gate and suggested this scheme which produces unipolar pulses:

The prototype circuit board resulted in a phase cal that looked like this:

• Unfortunately, the output died around 11 GHz.

• So Alan Rogers “married” the old design (with a series µλ switch) to the digital logic, and got good performance, which Chris describes next.

•However, Hittite has just introduced a new 50 Gb/s=25GHz logic family. I'm intrigued with the new HMC-C065 module, details can be seen at http://www.hittite.com/products/view.html/view/HMC-C065.

•Rather than being a logic chip, the HMC-C065 is a fully packaged, connectorized module which apparently costs $4275 in small quantities.

•So, my “all digital” Phase Cal using COTS logic parts still has a chance !