You may want to look at the analogue links we are using on MeerKAT. ~
2GHz BW and good dynamic range.
J
On 17 Dec 2009, at 8:57 PM, Dan Werthimer wrote:
hi john, tom,
attached is a paper on the analog fiber links used in the ATA.
the work well at the ATA, which transmits the RF band over
a single mode fiber. there are no mixers at the antenna - only
a low noise sandy weinreb amp and a fiber driver to transmit
the entire band, from 0.5 to 11.2 GHz.
i don't have expertees in fiber links,
if you are interested, i suggest talking with jack welch or dave
deboer.
mel wright and other ATA folks have
phase and amplitude stability measurements.
best wishes,
dan
John Ford wrote:
Hi Tom,
To save money on analog fiber transmitter/receivers, perhaps you
could
transmit your full 10GHz band on a single
analog fiber, or as you suggested, convert the 10 GHz band to a
pair of
5 GHz IF's, and transmit on two analog fibers.
After the analog fiber receiver (located in the electronics lab) you
then can break the 5 or 10 GHz bands up into smaller bands
and digitize the smaller bands, or consider digitizing the full band
using a 20 Gsps ADC (but you'll have to wait for
this ADC board, or help us develop it - we are looking for
partners to
drive the development of this board).
We've decided (Maybe prematurely?) that wide-band analog links are
not the
way to go, for single-dish, at least. The stability we need was
not there
the last time I looked, when you factor in the twisting of the
fibers, the
diurnal temperature variations, etc. Can you give me more info on
these
wideband analog links? It would be much easier for us if they really
worked well at these bandwidths of ~10 GHz.
In any case, I agree with Jouko that you should bury as many fibers
as you
have money for. The cost of the WDM systems far overshadows the
cost of
fiber at 500 meters. If you were trying to reuse existing
infrastructure,
it would be a different story.
John
Best Wishes,
Dan
Tom Kuiper wrote:
Dan Werthimer wrote:
Can you transmit the RF or IF through analog fibers, and then
locate
the digitizers in
an electronics lab, where they are easy to service, upgrade, keep
cool, shield from the receivers, ...?
We can, in principal. The receiver being built first for
Canberra by
Steve Smith in Sandy's group at Caltech re-uses much of the GAVRT
design and technology. 17-27 GHz will be converted to baseband in
five 2-GHz bands. The conservative assumption is that we will
clone
this for Goldstone. Of course, technology is always changing.
At the ATA we use analog transmission over fiber.
Each ~15 GHz bandwidth fiber driver/receiver pair cost $2500.
All
the digitizers
and digital backends are in the same room, making it easy to
upgrade
and service.
40 x $2500 = $100K is about the cost of the receiver. This is also
the reason why Sandy is bringing down only half of the IFs that he
generates in his GAVRT receiver. Do you think that, given the
receiver design, we would want to upgrade the ADCs?
Currently, the design is to bring the baseband out to connectors on
the receiver box, to be connected with hardlines to the ADC box.
Steve understands the need for shielding well. He works
elbow-to-elbow with Glenn, after all, and they have tested both
their
analog and digital gear together in close proximity.
ATNF digitizes at the receiver on the AT so it can be done,
carefully.
The fastest current CASPER digitizers is 6 Gsps (8 bit); we
plan to
develop
a 20 Gsps digitizer over the next year or so. Their are a some
80
Gsps digitizers,
but we don't have any plans to use them yet, as we can't handle
the
data rate.
When you finish the 20 Gsps digitizer, we may want to revisit the
receiver design. Currently, with available mixer chips, we cannot
cover the whole 17-27 GHz band but we could break it into two.
Jouko's recommendation seems like a good idea. Fiber cables are
cheap. If I understand it correctly, the single mode fibers will
support analog if we need to go that way.
Best regards
Tom
Best Wishes,
Dan
Tom Kuiper wrote:
During the upcoming Goldstone 70-m down-time, we have an
opportunity
to put fibers dedicated to radio astronomy in the antenna. The
receiver we envision would, when fully built up, have 40 2-GHz
ADCs. To be able to use our signal processing equipment with
other
antennas as well, we would want to put it in the Signal
Processing
Center, which is about 0.5 km away from the receiver. I assume
that
we would have a 10 GBe switch in the antenna. Could someone
give me
an estimate of how many fibers of what kind would best serve our
needs?
An alternative scheme would be what GAVRT DSS-28 does, which
would
be to bring 40 IFs to the SPC and digitize there. That way the
ADCs
could serve other antennas. However, this appears to me to be
more
expensive and more problem prone, and ADCs are cheap.
Thanks and regards
Tom
<ATAPhotonicsPaper.pdf>
Justin Jonas
[email protected]
