Trip Reports
to Allen Telescope Array (ATA)

42231 Bidwell Rd., Hat Creek, California Google Maps
W 121, 28' 13" - N 40, 49' 36" (gate on Bidwell) - 1009 meters altitude
W 121.47028 - N 40.82667
Or in the astronomer's Right Assention nn.nn.nn or something = 121/15.nn.nn
I visited Wednesday June 7, and Wednesday October 31, 2007
by Ed Thelen ed@ed-thelen.org

Corrections,additions,comments have been added since. Latest Sept 2008

Goals

Organization of this document

- Arrival at ATA
- History of Hat Creek in Radio Astronomy
- Our guides :-))
- Visitor Problem - Information Overload
- Commercial Credits
- Antenna, Offset Gregorian Telescope
- Pointing the Telescopes, Brakes
- Cooled Amplifier, Fiber optic link to a Node
- a Node, then to central building
- Splicing optical fibers
- Central Electronics
- Visit with Rick Forster (Resident Astronomer)
- Roaming the Telescope Area
- Assembly Building and Jigs
- Noise
- A big e-mail of corrections/enhancements Sept 2008

Arrival at ATA

I (Ed Thelen) have been to ATA twice, with friends. The main contact at the site is Susie Jorgansen (tel: 530 335-2364) (site manager).


Paul Allen 		The second trip was delayed one week to avoid the excitement of the formal dedication of the Allen Telescope Array.
OK, I wasn't invited :-| This day was for the wheels and heavy hitters -
Image from April 9, 2008 presentation by Tom Pierson, CEO of SETI, available at http://www.seti.org/csc/lectures.php

ATA is about 13 miles from Burney, the route is well marked. Driving along Bidwell Road we notice an antique looking 6 foot dish atop a weathered looking 30 foot wooden tower. This place seems to have quite a history. A quarter mile further on we arrive at a lockable gate stating the visiting hours.

An associate sets up chairs to form a little theater in the entrance hall and visitors are shown two introductory movies from a DVD about the history of the Hat Creek facility and the current ATA - Much of the following section is from History of HCRO.

History of Hat Creek in Radio Astronomy

In 1962, an 85 foot radio telescope was constructed here, and several very significant discoveries took place here - including first interstellar maser, detection of molecules of water and ammonia in "outer space".

In the 1970’s Berkeley astronomers developed a two dish centimeter wavelength radio interferometer at HCRO. This evolved into an array of 10 dishes operating at wavelengths of 1 to 3 millimeters. The 10-element array was called BIMA, named after the consortium of universities that operated and helped fund the instrument (Berkeley, Illinois, and Maryland Association).

Around 1993, a wind of over 100 MPH lifted the 85 ft. antenna out of its supports and dumped it crumpled on the ground.

Much of the BIMA array was moved to CARMA in Cedar Flat at 7,200 feet in the White Mountains near Big Pine, CA in 2003 to minimize water vapor absorbing the high frequencies. View in Google Earth All except this old warrior - abandoned as just too old and worn.

SETI and Berkeley Radio Astronomy Lab proposed a big (350 antennas) array of 20 foot (6.1 meter) radio telescopes connected in an array - with up to 16 simultaneous studies going on at once - basically 16 somewhat independent data streams coming from the electronics - limited to a circle in the sky of about 5 degrees diameter at any one time - the (frequency dependent) beam width of a single antenna. The above is a simplification of many options, including splitting the array into parts for more data streams.

Paul Allen (of Microsoft fame) seems to have paid $12.5 million to initially get things going, and committing another $12.5 million depending upon progress.
Nathan Myhrvold (also of Microsoft fame) apparently donated $1 million for the central electronics to do the magic of forming and utilizing the somewhat independent 16 data streams (or 8 if using both horizontal and vertical polarities).
U.S. Navy has donated $1.5 million in return for usage time.

There are also a number of other donors of significant cash and useful products. Susie mentioned that Xilinx has donated several million in the form of FPLAs (Field Programmable Logic Arrays) and technical help. (Both LaFarr and I had purchased Xilinx chips and development systems. :-))

Each antenna costs about $150,000 installed. There are a number of donors of $150,000, each of whom get to have his/her name placed on an antenna.
Reading:

History of HCRO
wikepedia
an AAS meeting summary

Commercial Credits

Other than some help in return for telescope time by the ONR (Office of Naval Research), this ATA is largely privately funded. (University of California Berkeley also helps) Here are the commercials posted by the Computation Center:
Vendor Donation
Agilent A signal generator and spectrum analyzer used for testing and monitoring ATA protoype test array signals
Sun Microsystems Sun workstations, LCD monitors, disk arrays and tape units used in SETI's Prelude system, and the massive storage system for ATA imaging data
Trimble Navigation The differential GPS positioning system used to establish the ATA antenna locations for the 350-element array
Xilinx FPGA (Field Programmable Gate Array) chips for the ATAA's BeamFormer, Imaging Correlator and SETI PDMs Programmable Detection Modules)

Our guides :-))

Since our announced purpose of visiting was to view techie stuff, Susie Jorganson always had a technical person available to answer our questions :-))
Rick Forster (Resident Astronomer, with overview of technology) at his desk overlooking the array. The top screen shows the web cam image which he can point and zoom for a view of outside and antennas. Local temperature and wind is also available. Jeff Chero, [jeffchero at berkeley dot edu] - electronics, our October guide, in the processing room. His other lair is the electronics work room. (not shown, Jeff Kaufman, Power Electronics Technician, our June guide.) I also talked with Geoff Bower [GBower at astro dot berkeley dot edu] in Campbell Hall in Berkeley.

Visitor Problem - Information Overload

The major problem for a techie visiting ATA is information overload. There are so many specialty fields necessary to make phased array radio telescopes point, receive and process data that the mind boggles at such a listing. Every subsystem seems a career specialty - antenna design, antenna pointing, low noise amplifiers, laser transmitters and receivers, fiber optics, highly stable frequency adjustable local oscillators, rf mixers, radio interference measurements/remedies, ADCs, digital radio techniques with digital down converters and filters, FPGAs, ... image synthesis, computer programming, and on and on ... most of which have had continual major upgrades in the past 20 years. And being a LINUX guru is almost necessary.
When I was a kid, astronomy was kind of a back water - learn a little optics, a book or two of astronomy, some statistics, a thesis (that you could explain to the man on the street) and you were employable? Now everything seems to be on the bleeding edge of progress in so many new fields. :-(( ;-))
I have no clue how a current astronomer has time for astronomy. :-((

Antenna, Offset Gregorian Telescope

The antennas are offset Gregorian Telescope. The big dish (the primary reflector) is 6.1 meters (20 feet) in diameter and is a segment of a paraboloid. The center of the paraboloid in approximately at the bottom edge of the dish - just like DishTV. A big advantage of the offset is that the secondary reflector (the smaller dish (about 6 feet in diameter) does not obstruct the view of the primary. Radio "light" bounces off the primary onto the secondary, then off the secondary to the Log Periodic antenna, not shown, but at the V of the blue lines. An advantage of the large size of the secondary is its ability to reflect and focus lower frequencies than the usual radio telescope. Note that the aluminum is not shiny - as a condition of usage, the aluminum is made dull by "soda blasting". Dimensioned drawing

The above is a bit of a schematic diagram. The actual telescopes have added components.
  1. Sheet aluminum under the area between the main and secondary reflectors to keep the radiation of the earth under the antenna from leaking into the antenna system. (just visible)
  2. A canvas appearing cover above the above mentioned sheet aluminum to keep leaves, birds, snow, ... from accumulating on top of 1) above. (just visible)
  3. A Log Periodic Antenna to convert radio waves into electrical signals in wires. (triangle structure in front of bottom of big dish)

from Acr48F.pdf

This is the motor control box on the back side of each antenna. This communicates with the Control Building via fiber optic lines. Nice way to protect the Control Building incase lightning hits an antenna ;-)) It receives azimuth and elevation targets, and transmits back actual antenna pointing, and a large number of other status values.

This is inside the protective cover in the antenna. The LogPeriodic antenna inside in about 2.5 feet long. It can efficiently receive a VERY wide range of frequencies - from about 400 megahertz (wavelength about 30 inches) to 11,000 megahertz (wave length less than 3 centimeters, about an inch) The little fins at the pointy end are about 1/4 inch long. This converts the radio waves reflected by the dishes into electrical signals. The base of the LogPeriodic is on the left, and the tip is connected to the Low Noise Amplifier inside the copper structure. Notice that fins are in planes 90 degrees apart - this permits receiving both vertical and horizontal polarized waves. After amplification, both vertical and horizontal RF is converted to laser light and transmitted to the Control Building. more pictures

This box is sitting on the base of the LogPeriodic antenna. Inside the LogPeriodic is a little cryogenic unit to make *VERY* low temperatures, to cool the Low Noise Amplifier. Apparently this (or every??) antenna has a vacuum monitor for the pump/refrigerator cooling the Low Noise Amplifier. Inside of the base is a little motor to extend or retract the LogPeriodic a range of about 10 inches for optimal focusing - see next lower box.

The optimal position of Log Periodic antenna is somewhat frequency dependent a focusing environment, such as a Radio Telescope. The elements of this extreme log periodic are spread over a length of over 2 feet. So, depending on general frequency you wish to receive, a stepper motor (Focus Drive) can slide the correct part of the antenna into the focal point. Fortunately the position is not very critical. Rick Forster says that improvement in setting gives a maximum improvement of about 20% :-))
An Overview of this "front end" is given at the SETI web site Technical Overview and diagram Here is a local copy

Pointing the Telescopes, Brakes
I asked about the change from stepping motor antenna drive to an AC drive that Jill Tarter (SETI) had mentioned. Jeff showed a smoked connector on a stepping motor. Apparently the stepping motor currents were much higher than the connectors and drive circuitry could reliably handle. He then showed the drive motors now used to drive the antennas in azimuth and elevation.

Each motor (above) drives a new (1950s) type "gear box" called "harmonic drives". Inventor's page This relatively new invention provides very high gear ratios - on the order of 30-1 to much higher, very compactly and with almost no backlash/hystersis. This photograph does not show the electronic position sensors.

(Local copies of harmonic drive documents)
page 1,
page 2,
page 3,
diagram,

This is an overview of the test and checkout station in the Electronics Shop (possibly called the Myhrvold Development Lab). The various subsystems under test/repair can be placed in this local convenient environment for quicker access and easier monitoring.
Here is Jeff working on a board. the two black wires from the bottom left center to the board are actually fiber optic lines for communication to/from the board. He has identified the failing chip and will send the board back to Berkeley for repair.
This is the inside of a motor control and communication box. The black edges of the box are radio isolators to help keep radio frequency noise inside the box and away from the sensitive receivers.
When all else fails, hit the big red button ;-}} (or maybe the computer reset?). Actually, the antennas move rather slowly, the main danger that the walkers may assume the antenna has not moved, and bump his/her head. The slew rate is on the order of five degrees per second. It takes over 16 seconds to move 1/4 of a circle - about a minute to do a circle. (Reference?)
Interface with the outside world. The copper tubing is cooling air, you can see the PC style power socket, and the fiber optic communication cables. The encoders give serial streams of angles to the box for feed back control and transmission back to the Control Room.
One of the angle position encoders. Catalog which says that this series has a maximum resolution of 200,000 or about 5 arc seconds or 24 microradians or about 18 bit resolution.

And here is the standard drive brake - about the size of the palm of your hand. With no electricity, springs force the center rotor (sintered metal) to contact the steel and cause a breaking action - to a limit. If external forces (such as wind at 100 mph) occur, the brake will slip - and in azimuth permit the antenna to point down wind. If you wish to move the antenna, you supply current to the break.

Cooled Amplifier, Fiber optic link to a Node

This image of the cooled (to about 40 Kelvin) low noise amplifier comes from an on-line SETI lecture given by Jill Tartar Feb 16, 2008, available at http://www.seti.org/csc/lectures.php. With I presume tongue-in-cheek, she said it amplifies "... basically from DC to daylight" ;-)) She then said that with the log periodic antenna (above) (feed?) they receive 500 megahertz to 11 gigahertz - a range of 22 to 1.

A refrigerator is used to permit the Low Noise Amplifier (I did not get a picture) to have even lower noise, the Low Noise Amplifier is cooled to well below the boiling point of liquid nitrogen. The refrigerator (called a Pulse Tube Cooler) is contained inside the pyramid of the log periodic antenna above. The general idea is similar a household refrigerator, compressing and expanding a gas, with but with changes for smaller size and longer life. This is a reference to a Pulse Tube Cooler, which unlike a Sterling cooler, has no moving parts on the cold end. also. A SETI Block Diagram.

After the cooled Low Noise Amplifier, the signal is strong enough to send to this Amplifier box via coaxial cable. Here it is further amplified and routed via coax to the board with the laser. The laser is connected to the blue fiber optic shown for transmittal to a node box which handles 12 antennas (24 channels) Note that there are 2 data channels in this box, one for the vertical polarity radio waves, and the other for the horizontal. The big green board in the middle is the amplifier control (suitably shielded.

Node, then to central building

The control signals and data signals from each antenna pass through a "Node", which is an organizing convenience, no known technical use? The current organization is 12 antennas are 'serviced' by one Node. There is more physical capacity in each node if the need arises. The chimney is for a ventilating system.
Jeff points out the optical fiber entry and exit. and lots of spare power and space.
Optical signals in the Central Building This might be called an Optical Time Domain Reflectometer :-)) How about that for a mouthful ;-)) (as opposed to frequency domain). It is useful in determining many characteristics of a fiber optic cable - such as how good are any splices or junctions, where are they - as measured in time converted to length, various other imperfections, and very vital in this application, how long in time to the other end as accurately as possible. With the game being to phase match every thing, this is vital knowledge for initial setup of delay parameters to point the electronically steerable sub-beams with in the physically steerable main beam.

Splicing optical fibers

There are miles of optical fibers already in the 42 antenna array. There are several splices in each fiber between the control room and each antenna. Each splice must be carefully made or additional signal attenuation and reflections will be introduced, causing potential or real problems. To make a low attenuation, low reflection splice requires considerable care. The two faces of the splice are cut at 7 degrees from perpendicular, carefully cleaned and polished, and then fused together.
The RF carrying fibers are of a special glass with especially low loss and low thermal expansion (delay change) characteristics. The actual signal carrying cross section is (7 mills?) so cleanliness and alignment are especially important.
The left hand image is a good clean surface, ready for fusing
The right hand is a dirty surface, will cause trouble.

Central Electronics - now things get "interrrresting" :-|

Here we are in techie & mathematical heaven -
There are things going on in here I probably never understand :-(( And this version may have serious flaws :-(( - getting corrected :-))
However, lets get started - This is where the field fiber optics, carrying the vertically and horizontally polarized RF information, come into the building and are joined to the patching system.
And onto distribution (I think)
And somehow the data (both polarities) gets into these Radio Frequency Converter Boards1 "RFCB"- sometimes called "mixers" - - with inputs from 4 separate local oscillators to begin generating the 16 separate data streams available for users.
Detail on the mixers
RFCB1
Band Pass Filters1
(A correction)
The cards pictured here: - are actually IBOB boards with 2 adcs attached. They use a very similar FPGA to the BEE2 but are actually a different board. Much more info available at:
http://casper.berkeley.edu
Cheers,
Andrew Siemion
Dear Friends,
I'm so confused -
Also see Beam Forming.
Clearly I've lost it :-((
The unfortunate part of all of this is that I came up to find out about the delay lines and correlators, as instructed by the Berkeley folks, but that expertise seems to be elsewhere :-((
This is a weather reporting station and GPS Trak clock. (Sidereal time is derived from GPS ( or ?) by a card and steers array and time tags data flowing into storage. The site has a rubidium clock - about 10^-12. A cesium clock (10^-15) would be for VLBA work - not currently planned.) And precision pulses to ... ... racks are for the fiber optic input, LO, sample clock, 1 pulse-per-second generation and distribution, network infrastructure, support computers. The SETI Prelude racks are deep blue in color, sealed up (for airflow) and located just to the left of the racks in the picture above.1
(Hot Update, Better, clearer, more up-to-date diagrams here.)
I found this diagram, specifying a 42 antenna system so I figure it is relatively current - I doubt that a year ago they knew that this phase would be 42 antennas. The image here is large (124 K bytes) but still hard to read but maybe worth the struggle if you have a tutor handy. I was particularly confused as it indicates that the output of the digital beam former is analog, and goes to more ADCs. But then again, I'm guessing at the whole thing anyway -
But then I found this representation of probably brightness at some radio frequency of M-33 on the wall. If you get tired of O-H radicals, Doppler shift, atmospheric absorption, calibration problems, people problems, interference problems, scheduling problems, funding problems, environmental problems, and all the other hazards of life, you can gaze at this and feel good :-))

Visit with Rick Forster

Susie Jorgansen then introduced us to Rick Forster (Resident Astronomer).
We talked astronomical things, including that these telescopes were "fast". A "fast" telescope implies a short focal length and a large field. Fast, however, is a term borrowed from photography (an f/5 telescope can take a photograph with one-fourth the exposure time of an f/10 instrument).
Rick started then talked about the nearby antenna pedestal. Apparently nothing in life is perfect, and sometimes the "little things" can turn into "big things".

For instance, when the sun shines on one side of the steel tube center post, it can tilt the top up to 1/2 degree (the angle of a full moon). Rick called this effect "sunflowering", as in flowers following the sun. This may not sound like much, but it is 1/5 of the average beam width of the reflector telescope (much greater error if working at higher frequencies. There are other aspects of this that contribute more errors, such as shifting the focal point of the antenna on the order of a wave length. As this whole array tries to have errors less than 1/8 wave length, this one problem makes a big contribution to the entire error budget of the array.

Rick said that this center post tilt error is difficult to model, and therefore difficult to correct in software or by other means. I imagine that "random" breezes, patchy clouds moving about, and similar effects are most difficult to measure and work into a model. I suggested that painting the gray (galvanized steel) center posts and supports with titanium dioxide paint (very white and reflective) might help - The Nike precision tracking antennas I worked with so many years ago used titanium dioxide paint to help fight this same problem. Maybe the reason this is not done is that the glare white, desired for engineering reasons, is very noticeable in the surrounding woods (a National Forest?) and offensive to the natural beauty of the area??

Of course we had to ask Rick how so many antennas could be used (in phase) together - it sounds like a mind numbing problem. Rick said that you basically pair each antenna with every other antenna in the group, and that you want as many different distances as practical. That this was why the field of antennas looks so random. The goal, in part, is to place antennas so there is as large as practical range of differences in distances. There is of course some minimum practical difference between two antennas, and a maximum practical distance between two antennas - but you really do *NOT* want the antennas to be placed like rows of corn or a checker board or any regular order. For a very helpful article on the placement of the antennas and why it isn't as random as it appears, read Dr. Seth Shostak's "Spreading Antennas Around". They even use heavy duty computing to help do a proper job.

Roaming the Telescope Area

Rick said that LaFarr and I could roam the telescope field, which is near the office and near the western edge of the planned complex. (Expansion will take place to the east.) "Just be careful if the telescopes start to move". Since their max angular speed is 4 degrees per second, I wasn't worried about being smacked too hard. Movie from SETI web site
This is the back view of the ATA style antenna. The "dish" is "hydroformed" in Idaho by a firm that specialized in this operation. As part of the deal, the "dish" is "soda blasted" to remove the shine. This has two benefits
a) less glare in the National Forest.
b) less worry about focusing the sun's rays and the resulting heating.

Concrete pads which support the antenna weight and are designed to keep the antenna in place in a 100 mph wind. These pads are poured into (I forgot how deep?) holes drilled by contractors. Nothing is simple. There are two sizes of pads. Susie told us drilling contractor stories. Notice the ground wire bolted to the antenna foot and leading into the ground. (Grounding and lightning protection is another specialty.) All signaling to and from the antenna is by fiber optics, but I imagine that a lightning hit could cause interesting power supply problems. The dark thing on the top left is the aluminum preventing radio radiation from the warm earth from getting into the receiver. Part of the white antenna assembly building is visible in the upper left.

This is the steel pipe that comes up from the ground to the center of the center post. All power, signals, and even cooling air come into the antenna through here. Detail of the foot and hold down screw. Note the square threads, and the jam nut on top to prevent loosening.

Going up, here is what we think is an electric ?motor? control box. We saw a little bird fly out near the thing that looks like two eyes. There must be more practical problems than I wish to think about. Like trying to keep the fiber a constant length - there is apparently a cooling air flow from the node (12 antennas each) to keep the optic fiber as close to rather constant earth temperature as practical. Here is another view of the metal to keep radiation from the earth out.

Side view of the antenna Detail of the rear. The wires going from strut to strut provide wind vibration damping. Also visible is the azimuth drive at the top of the center pole, and the elevation screw.

Assembly Building and Jigs

We made a large loop of the existing antennas and eventually came to the assembly building - a large canvas tent structure with vertical walls up maybe 20 feet.

This is a rolling center pole upon which a partially completed telescope can be placed for further assembly or "system integration" ;-)) We did not understand the cut out near the top.

This circle (and more) posts bolted to the concrete floor are carefully leveled so that the hydroformed 20 foot primary reflector will not be deformed during further assembly. A ring of aluminum is welded to the outer edge and supports for the struts welded to the ring. The goal is to have the final reflecting surface with an error of less than 1/8 wave length at 11 gigahertz - less than 1/8 inch from a paraboloid sector over the 20 foot diameter surface.

This orange jig probably helps lift the horizontal partially assembled 20 foot primary reflector for further processing. Note the white ?Teflon? pieces on the jig which probably help protect the reflector from scuffing. There are two little white pieces (one hidden by the on the foreground post) mounted on a rotating shaft controlled by a screw arrangement. This likely secures ?something? during movement. We did not figure out how the reflector gets rotated from horizontal to vertical for mounting on the black vertical post.

This is the fork lift that carries the assembled primary, secondary, struts, log periodic antenna, wiring, azimuth drive, elevation drive, ...
to an already prepared center post bolted to the foot pads. This primary assembly is bolted to the top of the center post in the telescope array. (likely on a very still morning free of wind gusts ;-)) The final wiring through the center post is completed and I presume azimuth and elevation calibration/alignment can begin :-)) And the focusing and the fiber optic length determination and the noise levels and the ... and final acceptance as a functioning part of the array. The casting with the hole on top is part of the dish mount structure. We couldn't figure what the straight sided triangle thing is for.

I took these images of photos hung on the walls of the control building

The great outdoors (and lightning) is powerful stuff. This is ground wire that connects a steel leg of each telescope to the earth ground. Looks like 19 strands of 4/0-1C soft drawn copper. Both LaFarr and I had past experience with lightning. LaFarr was standing near a tractor when it was hit, and I had strung an informal interbuilding squawk box system through a tree that got hit. (We forgot to ask details of the earthing.)

Noise

A reason for selecting this location for a radio telescope is the relatively low radio noise leaking in from civilization. The surounding hills and mountains largely block noise from the earth outside. The rural location, and being in a National Park provides low ambient noise. As you come into the ATA area, you are asked to turn off your cell phones - they all operate in the frequency band (500 mHz to 11 gHz) observed by ATA. Unfortunately, EVERYTHING radiates radio waves, including you and me.

Carl Sagen is quoted as saying that all the energy ever received by all the radio telescopes in the world is less than that of a snowflake gentle falling onto your hand. Keeping external noise out, and using low noise techniques inside a radio telescope is vital to successfully observing the very remote signals that are so weak arriving at the earth. Noise is often quantified as an equivalent temperature in Kelvins - from absolute zero. Liquid nitrogen at 77.2 K (-320 degrees F) is hot and too noisy.

1 A big e-mail of corrections/enhancements from Matt Dexter

Matt also supplied these two lovely data flow diagrams in .pdf format that are easy to view - to aid understanding. Each is about 135 KBytes can be magnified by your brouser for easier viewing. :-))
  1. MDexter_fx32_iBob_single_2008sep18.pdf
  2. MDexter_fx32_iBob_2008sep18.pdf 

Thanks for writing

I *almost* have a picture of you at
    http://casper.berkeley.edu/team.php     ;-))


----- Original Message ----- 
From: "Matt Dexter" 
To: 
Sent: Thursday, September 18, 2008 4:19 PM
Subject: trip report and struggling on with ATA


> Hi Ed
> 
> I just came across
> http://ed-thelen.org/ATA/HatCreekATA.html
> and wanted to congratulate you on a great write-up.
> Well done!
> 
> If I may offer a minor correction or two...

    or many more   ;-))
 
> I think your "mixers" are called RFCB or Radio Frequency Converter Boards.
> And those aren't low pass filters @ http://ed-thelen.org/ATA/Mixer-03-.jpg
> Those are actually bandpass filters.

Probably very true,  a DC component, bias, is probably worse than useless.
   :-|

> The F3dB [3 db down?] of those BPFs is about 15.2 GHz -> 15.8 GHz .
> The allen head screws get hand-tuned on a network analyzer to define the
> filter response.  Not easy for beginners to get right.

I interpret "F3dB"  to  be a 3 dB  down corner frequency [of a  filter]
   (I had to take filter theory twice, didn't do so good the 1st time ;-((

> The signal path is :
> upmix the entire signal received over the fiber with the 4 different
> programmable LO1s ->
> filter with ~ 600MHz wide bandpass filters ->
> down mix with fixed LO2 = 14.9 GHz  ->
> 600 MHz bandlimited output is now 329 -> 929 MHz

for those even more shaky than I,
   LO = Local Oscillator
 
> 600MHz bandwidth is too expensive to handle in our current digital signal
> processors so there is a 529-729 MHz BPF on every output of the RFCBs. If
> you have eagle eyes you'll see them in
>  http://ed-thelen.org/ATA/TechieWiringHeaven-.jpg

"Eagle eyes", and knowing what to look for  ;-))

> Thus, the analog signal into the iADC+iBob plates, as Andrew discussed
> is actually a well defined band 529-729 MHz. And only the center 100 MHz
> is processed by the current correlator and beamformer backends.
 
> 
> This picture is mislabled.
>  http://ed-thelen.org/ATA/SETI-Processors-.jpg
> 
> Those racks are for the fiber optic input, LO, sample clock, 1
> pulse-per-second generation and distribution, network infrastructure,
> support computers. The SETI Prelude racks are deep blue in color, sealed
> up (for airflow) and located just to the left of the racks in the picture
> above.
 
 
> Regarding
>  http://ed-thelen.org/ATA/HatCreekATA-StrugglingOn.html#BerkeleyOct22
> Our current FPGAs could not fit the interpolating filters to get
> to delay steps finer than the 838.8608MHz sample clock resolution.
> Just not enough gates. 1/838.8608MHz is as fine of delay step
> as is supported at the moment.
> 
> At this time only the FPGAs programmed for the Transient Search, aka Fly's
> Eye, experiment have fine channel gain correction on the 8bit I + 8bit Q
> voltage samples just after the digital down conversion. The FX correlator
> has gain correction on the F engine's output 1024 spectral channels which
> are 4bit I + 4bit Q voltage samples.
> 
> The attachments may, or more likely -  not, help explain part of the
> signal flow.
> 
> all the best,
> Matt Dexter

Adding new finds here, as of June, 2008, new stuff near top (here)

Interesting search keys

BEE2 beam
Interesting web pages My Understanding
  • a delay unit? and if the electronics aren't fast enough (ATA wants 800 megahertz) you can run n units in parallel phased. Not easy to get 350 x n physical units synchronized when the period has a wavelength of 0.37 meters !!

Maybe we can understand their technique of delaying the signals of each antenna to make 16 independent beams. (These 16 beams must be:

- with in the 5 degree beam width (at 1.4 gigahertz) of each 20 foot reflector
- (the focus bandwidth on the log periodic sensor is very broad also - max 20% down from nominal),
adjusting the delays to allow for
  1. the desired angles of very narrow phase generated beam
  2. the earth's rotation moving the phasing of each antenna with respect to all others
  3. the location of each telescope
  4. the pointing vs delay offset of each telescope
    - the focus is not at the center post - hopefully antennas are identical and this normalizes out.
  5. drift in the length (transit time) of the fiber optic
  6. and a host of other effects.

Then of course there is:

- as the earth rotates, anu apparent polarization of the "target" changes
-- what had been "vertical" slowly changes toward "horizontal"
-- same effect in optical altazimuth mounts, you must rotate the sensor or "software".
Is the "Front End" rotated? or it the rotation handled in the electronics or application?
- The earth's rotation causes oblate spheroid and local gravity is partly "centripetal" force.
-- offset in "up"?, "down" does not go through earth's center, and tidal force of moon and sun
- doing SETI

This is serious stuff :-|


HatCreekATA-StrugglingOn
Pointing/tracking the Synthesized Beam "Array"
Information Please
Article SETI OBSERVING PROGRAMS ON THE ATA from "SETI INSTITUTE Explorer" Second Quarter 2008, Volume 5, Number 2

for comments, corrections, additions, ... please contact ed@ed-thelen.org

corrected up through Jan 18, 2008