Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.naic.edu/~tghosh/vfuture/whyMK5.txt Дата изменения: Thu Nov 20 17:17:10 2003 Дата индексирования: Sun Apr 10 04:15:03 2016 Кодировка: Поисковые слова: scientist

Q1. A Brief description of the Mark 5 system, what it is and
what does it do to enhance our science?

An Overview (from A. Whitney's documents on the subject)
---------------------------------------------------------

(The file Fig1.jpg has a picture)

The Mark 5 system is being developed by Haystack Observatory as a Gbps
VLBI data system based on magnetic disk technology. Incorporating
primarily low-cost PC-based components, it can record data at a rate of
up to 1024 Mbps on to an array of inexpensive, removable IDE/ATA
disks.

At the heart of it lies the so called StreamStor disk interface card,
specially designed (by Conduant Corp.) for high-speed real-time data
collection and playback. It supports the following three physical
interfaces in a triangle of connectivity as shown in the following
schematic diagram:

1. Front Panel Data Port (FPDP) 2.Disk Array
/\
This is a two-way 32-bit card-top bus / \
through which realtime data may enter / \
or leave the system. All 32 bits of / \
the bus are always active. / \
1. FPDP bus ---------- 3. 64-bit/66MHz
2.This port supports up to 16 IDE disks PCI bus
in 8 master/slave pairs.

3. This is the standard connection to the host PC operating under a
Linux OS.

The path exercised for regular VLBI is between 1 & 2. Path 2 - 3 is
used to read the data off of the disk, and 1 - 3 can be used in eVLBI.

The I/O panel (located at the rear of the unit Fig1), contains the same
set of connections as a Mark 4 or a VLBA tape drive, so that the Mark
5A is very easily substituted for a tape drive. Up to 64 (or 32 for
VLBA formatter) tracks of formatted data can be directly recorded. The
StreamStor card accumulates 32-bit 'words' (corresponding to the 32-bit
wide FPDP bus) into 64-kB blocks and writes them sequentially in a
round-robin fashion to the disk array. This alleviates the needs for
"barrel-rolling" and "data-modulation" by the formatter.

For the near future, VLBA will still be operating using Tape recorders.
Hence, if Arecibo were to get a MARK 5A unit within the next few
months, the most useful configuration for a Mark 5A unit would be the
so-called "pig-wire" mode whereby both the Tape recorder and the Mark 5
disk-recorder can be kept wired to the formatter in parallel, providing
the opportunity to record up to 256 Mbps on tape or disk at any time.
In this configuration, Arecibo would be able to participate in both
VLBA (tape-based) and EVN's disk-based runs at short notice. For higher
data-density runs, the choices would be to either, reconnect all 64
tracks from the formatter to the Tape recorder or to the Mark-5 disk
recorder, providing 512 and 1024 Mbps data-rates respectively.

In its "straight-through" mode, Mark 5 will be able to record any 32-bit
digitized signal, a feature that can be utilized by other AO backends
as a fast recorder.

Mark 5B
--------

This system will have a re-designed I/O board facilitating internal
data formatting. It will be fully VLBI Standard Interface (VSI)
compatible. With Mark 5B, it will be possible to record data at the
rate of 1024 Mbps from both Mark-4 and VLBA systems. Among other
improvements, it will also have much simpler a '80-pin' Input/Output
interface. The expected date for its full implementation is early
2004.

Upgrading from Mark-5A to B would involve changing the I/O board,
and (most probably) the I/O ports. Cost of this is as yet undefined.

Mark 5 Software
----------------

Much of the software for VLBI uses is already in place. The Mark 5 PC
has its stand-alone Linux-based software, and Mark 5 units operations
are also included within the 'Field System' (developed and maintained
by NVI Inc.) that runs Arecibo's current VLBA4 recorder.

Functional and scientific enhancement:
----------------------------------------

Tape recorder-based VLBI systems require heavy maintainance being
rather delicate systems. Headstack alignment and proper calibration is
also very important for the success of the observations. For playing
back, the correlators need a separate playback unit. Tape-based
recording is also more error-prone when compared to any disk-based
system. All these factors make the Mark 5 a much easier and reliable
system to use.

In addition to the operational ease, a move to Mark 5 systems will
enable Arecibo to provide 256-MHz bandwidth with 2-bit sampling, a
factor of 2 improvement over its current capability. The data
acquisition rack for Arecibo's VLBA4 system has 8 double-sided baseband
converters, each of which can be set to a maximum bandwidth of 16 MHz.
This configuration implies a maximum possible data output rate (for
2-bit sampling at the rate of 32 Mega-samples per sec, i.e. 32 x 2 x
16=) of 1024 Mbps. However, to date we have been limited to record only
at maximum rates of 256 Mbps with the VLBA and 512 Mbps with the EVN
stations even with our dual-headstack tape recorder. A move to a Mark
5 system would enhance the maximum sensitivity possible for baselines
involving Arecibo by a factor of 1.4 without requiring any other
changes in the DAS.

Although such a factor may not at first seem impressive, this
enhancement will open up possibilities of conducting wider-band VLBI
with the world's biggest antennas where Arecibo's presence is essential
for both detecting the weakest sources and to form better synthesized
beams for high dynamic range maps of weak features.

Another advantage of Mark 5 systems is that the data can be read out
using the same unit (equivalent to "play-back units" for tape-based
systems), or sent to a correlator directly via the 1-3 pathway
mentioned above. This opens up the possibilities of e-VLBI for
Arecibo.

A few research areas that will benefit immensely are:

1. Host galaxies of low-luminosity AGNs can be studied by detailed
mapping of their ISM and nuclear regions via both their continuum
emission and atomic/molecular absorption effects.

2. Investigations of weak-cored sources, the low-luminosity Seyfert and
LINER galaxies, the most distant radio galaxies, and the search for
gravitational micro-lensing. Observations of such extragalactic
phenomena provide fundamental input both for testing the unification
schemes of AGNs, and for estimating cosmological parameters.

3. Polarization VLBI has already yielded information of the highest
resolution on the magnetic fields within AGNs. As polarised signals
are typically only about few percent of the total intensity, Arecibo
with its enhanced sensitivity baselines is needed to make many
observations even possible.

4. Afterglows of Gamma Ray Bursters (GRB) are usually very weak at
centimeter wavelengths. VLBI imaging efforts for these engimatic
objects, where researchers have already traced out jet-like emission
would benefit immensely from the highest possible sensitivity from
Arecibo.

5. Imaging of weak SNRs in other galaxies and especially in nearby
starburst (ULIRG) galaxies.

6. Within our Galaxy, VLBI is crucial for the study of flare stars,
radio-emitting X-ray binaries, and other classes of radio stars, to say
nothing of the galactic superluminal sources. This is also the case
for the imaging of galactic maser sources of OH and CH3OH.

7. High-sensitivity VLBI permits the astrometry of pulsars, and the
investigation of the interstellar scattering of their radiation during
propagation which, together with observations of low-latitude AGNs and
galactic masers, can provide unique information on turbulence in the
interstellar medium and the solar wind.


Most types of objects observed in radio astronomy have a wide range of
intensities -- if only because their distances vary. Increasing the
sensitivity of an observation by a factor of even 1.4 increases the
volume of space within which sources of a given luminosity can be seen
by a factor of about 1.7. Often, source properties are linked with
luminosity, and lowering the detection threshold at VLBI resolutions,
will not only increase the number of observable sources but also help
generate fresh clues for outstanding astrophysical and cosmological
questions.



Q2. A simple block diagram of how Mark 5 interfaces with AO

--- fig2.ps file is such a block diagram

Q3. May need major changes to the AO system (IF frequency/Bandwidth etc)?

--- No, as I.F. frequency remains the same, while the Mark 5 maximum
bandwidth of 256 MHz is already catered for.


Q4. Delivery and cost

--- Delivery time is about 2 months, total and the cost of a Mark 5A
unit is $17,500 (as of Dec. 2002). However, it has been a practice
of VLBI "stations" to contribute some tapes to the common pool.
For example, NAIC contributed 20 VLBI tapes in 2000. Likewise, it
would be good to include the price of a number of disk-packs (the
cost of each of which is at least $1750). Please see Appendix 1,
for more details.


Q5. Operational costs -- disks One-time use or multi-use ?


---- The disk-packs are multi-use and are circulated within the VLBI
community following the same methods as the current VLBI tapes.


******************* Appendix 1 ****************************************



-------------- The Mark-5A Quote - (from ~ Dec2002)----------------------------

From: "Alan R. Whitney"
Subject: EVNtech: Mark 5A available from Conduant

Conduant Corporation has agreed to become a builder and distributor for
Mark 5A systems and has asked me to spread the word. Please feel free
to pass this information on. Conduant is offering the following:

Mark 5A system excluding drives/modules - $17,500:
This includes:
(1) TK200 Chassis
(1) Motherboard - SuperMicro P3TDLE (with 1 P3 1.26GHz Processor, 256MB
ECC)
Matrox G45MDVP32DB Video Card)
(1) Antec 550w Power Supply
(1) 40GB System Drive
(1) CD or CD/Floppy Combo
(1) Red Hat Linux
(1) Mark 5 I/O Board
(1) I/O Panel
Cables/Assembly/Test
1 Year warranty on Conduant Products, passthru warranty on remaining
products
-
'8-pack' module with eight 120GB disk drives (assembled/tested); w/o
shipping covers - $1750
-
'8-pack' module with eight 200GB disk drives (assembled/tested); w/o
shipping covers - $3425
-
'8-pack' drive module without disks or shipping covers - $260.00ea
(Quantities 1-20), $238ea (Qties 21-50), $225ea (>50 units) - see Note 3 below
-
Shipping covers for single '8-pack' module - $16 - see Note 4 below
-
Notes:
-
1. All items are ~60-day delivery. Payment is net 30 days
-
2. Note that these prices are *only* for the VLBI community and are much
lower than normal commercial prices. The cost of a 'do-it-yourself' Mark
5A kit is ~$12-13K, so you are paying a relatively modest amount for
integration and testing by commercial standards. Haystack Observatory
would not be able to build and sell these units any more cheaply due to
standard MIT overhead, so I feel that Conduant is giving the VLBI community
a good break.
-
3. Note on '8-pack' modules sold without disks: In order to keep prices
down, '8-pack' modules sold without disks are not functionally tested
before shipment. This is due to the high labor cost involved in testing
modules (they must be loaded with disks, tested, then
disassembled). However, an express swap/replacement is offered for any
non-operating module. In practice, it is expected that the non-operating
rate will be quite low. If you wish to purchase empty tested '8-pack'
modules, please contact Conduant for a price.
-
4. A two-piece metal shipping cover is available which encloses the
'8-pack' module to guard against static electricity, protect the disks
and, to some extent, hide the disks from casual eyes.
-
5. Conduant is also considering offering extended 24 or 60 month warranties
with 21 day repair or replacement (at Conduant's discretion) and would like
to know whether anyone might be interested. They have not yet been
priced. Let me or Conduant know if you might be interested.
-
6. Additional information on the Mark 5A system is available at
http://web.haystack.mit.edu/mark5/Mark5.htm
-
If you have any questions, please please contact me or Conduant.
-
Alan Whitney
MIT Haystack Observatory
Westford, MA 01886
awhitney@haystack.mit.edu
-
At Conduant, please contact:
Tom Skrobacz
Conduant Corporation
tom@conduant.com
Phone: 303.485.2721 ext: 101
Fax: 303.485.5104
www.conduant.com