Minutes of the LArg Front-end Optical Links Meeting
CERN
8th July 1998
Present :
M-L.Andrieux - ISN
B.Dinkespiler - CPPM
J.Lundquist - KTH
R.McLaren - CERN
M.Pearce - KTH
P.Schwemling - Paris
R.Stroynowski - SMU
MINUTES
o Report from CPPM - B.Dinkespiler
Copper link status.
Bernard reminded us that 10 FEB -> mini-ROD
cables have already been installed including the flat conversion
cables running to the mini-ROD's. In addition, 4 copper link VME
boards (P2 connectors) have been built and tested and another 10
have been ordered. Bernard reiterated the need for careful planning
and infrastructure (cable trays) if more copper links were to be
installed reliably.
MUX board.
A MUX board has been built to allow the 32 bits from the
FEB to be fed into the 16 bit G-link input. This board has been sent
to Al Gara (together with optics) to be tested with a FEB and
mini-ROD. Initial results are very encouraging and have been useful
in understanding noise problems in the FEB -> mini-ROD system.
Multi-channel BERT (bit error rate tester).
Work is starting in earnest on this project now
with a student involved at CPPM. It is hoped to be able to have
results in time for the LEB Rome workshop.
o Report from ISN - M-L. Andrieux
Orleans Neutron Irradiation Facility.
The facility will replace the
SARA irradiation facility in Grenoble. It is located outside of
Paris. Full information from :
isnwww.in2p3.fr/atlas/andrieux/CERI.html. The facility can accommodate
both cryogenic and room temperature irradiations.
Amphenol Project.
Marie-Laure reported on the status of the Amphenol
project. Many of the previous problems have now been solved and
Amphenol were expected to present a prototype working link at a
meeting in their laboratories on July 9th. A second meeting is planned
with Amphenol on July 29th at which time they must have produced
two complete demonstrator links which fulfill the original specifications.
New personnel.
Marie-Laure announced that there was a new engineer
working at ISN on the link project. His first task would be to
investigate the possibility of driving VCSEL's directly from the
G-link output - thus removing the need for a laser driver.
o Report from KTH - M.Pearce
Report from the ROD meeting.
A number of points raised during the ROD
meeting (7th July, CERN) are relevant to the FE-links group :
o Location of ROD's (USA-15 or surface)
- Cost input from link group (underway MP)
- Check if 300m transmission is OK (SMU have fibre)
o Optical splitting is an elegant solution to trigger tower remapping
o ROD demonstrator board will have 2 links per board (one link per
FEB)
o Final boards may have as many as 4 Gb/s links per board
Fibre installation document.
In collaboration with R.McLaren's group
at CERN, a document is under preparation detailing installation
issues for LArg optical fibres and connectors (including serial
control links). The difference in cost of housing ROD's in USA-15
or on the surface is covered. At present price quotes come from one
company (Europtics) but a world-wide price survey is being launched.
Fujikura fibre.
Mark reported that the SCT group have approached
Fujikura for a quote on approx. 15 km of radiation hard fibre. This
fibre has shown negligible radiation-induced attenuation for
2*10**14 n/cm**2 and 240 kGy (LArg requires 2*10**13 n/cm**2, 800
Gy). The fibre has a 50/60/125 structure (pure silica core, fluorine
doping) and is step index. Mark
explained that if a combined order were to be placed with the SCT
group a decision was needed within a month. There are several issues
to address :
o How much fibre ?
o Can we use the same ribbonisation (ie: 12-way) as the SCT ?
o Can this step index fibre be joined to graded index fibre (ie:
radiation hard step index fibre used in the crack region and coupled
to cheaper graded index fibre which runs to USA-15) ?
o Can the step index fibre support 1.6 Gb/s, ie: dispersion issues ?
Mark showed that simple calculations together with simple dispersion
measurements performed by the SCT group made it look unlikely that a
fibre length over 20 metres would support 1.6 Gb/s data flow. In
order to proceed Mark suggested that a demonstrator link be tested
with the step index fibre (40 m in Stockholm) and step to graded
index coupling be investigated.
Laser driver news.
A replacement (cheaper !) laser driver and VCSEL
combination has been made at KTH to replace the commercial
transceiver units used to date. The circuit is still under test and
results will be available in a few weeks. The laser driver uses a
sub-micron BiCMOS process and its radiation hardness will need to be
evaluated (total dose and SEE).
Mitel News.
Prototype VCSEL and PIN (4-way) arrays have been received
from Mitel. The devices are based around the MT-ferrule. The PIN's
are being evaluated by the Oxford SCT group. The VCSEL's will be
tested electrically by Oxford and at Gb/s speeds (with the laser
driver) by KTH. Mitel are happy to accept input on future packaging
ideas.
Irradiation facilities in Sweden.
Mark briefly reported on two
neutron facilities near to Stockholm. One is based on a reactor
source and the other on an ion beam and target. Both facilities
offer the possibility of quick and easy evaluation of devices such
as the laser driver being developed at KTH and will be visited
during August with a view to arranging access times etc.
A 32 bit ATLAS data generator (Presented by J. Lundqvist).
Johan explained the current progress with using an Analog devices SHARC
DSP to generator 32 wide ATLAS data words to be fed into Bernard's
BERT. The device could send a pre-generated list of data words
stored in memory (32 K) but would need a faster DSP to realise truly
pseudo-random word generation at 40 MHz.
o Report from SMU - R.Stroynowski
Transceiver autopsies.
Richard summarised results from the dissection
of the Gb/s transceivers irradiated during the March neutron tests at
SARA. In all transceivers (Honeywell, HP, Methode) the VCSEL's survived
the total dose (10 yrs LHC). In the Honeywell transceiver, the CMOS
driver chips died. The most promising transceiver from Methode was
full of a potting compound which proved hard to dissolve for
further investigations.
Alcoa-Fujikura fibre.
Richard reported that a second batch of the
Alcoa-Fujikura fibre had been purchased. This fibre is completely
standard and meant for interconnects between Gb ethernet transceivers.
Previous tests
(http://www.cern.ch/Atlas/GROUPS/FRONTEND/links/minutes/may15_fibre.html)
showed that the fibre was remarkably radiation tolerant. A further
gamma test on this new batch has reproduced earlier results. The same
fibre will also be neutron irradiated at SARA. There was some
discussion over the results and in particular how to determine why
the fibre was so radiation resistant even though it was not designed
to be so. In particular, it was suggested that the dopants in the
fibre could be deduced using an appropriate spectrometry technique.
Optical splitters.
Richard showed samples of optical splitters from
various manufacturers. The insertions losses measured were
compatible with those in the data-sheets. Tests of the splitters
within a Gb/s link were planned in the near future (during the
forthcoming SARA irradiation test of 15/16 July ?). As mentioned above
optical splitters are an elegant and highly cost effective way of
realising trigger tower remapping between FEB's and ROD's.
o A.O.B.
Robert described a new type of low-profile connector called an MT-RJ
which could be useful where a fibre fibre density is needed, ie:
patch panels, ROD etc. (http://www.hp.com/HP-COMP/sff/)
M. Pearce
(July 17th 1998)
M.P.