Minutes of the LArg Front-end Optical Links Meeting
CERN
28th January 1998
Present :
M-L.Andrieux - ISN
E. Auge - LAL
B.Dinkespiler - CPPM
A.Gara - Nevis
F.Henry-Couannier - CPPM
J.Lundquist - KTH
O.Martin - CPPM
R.Mommsen - BERN
A.Patti - ISN
M.Pearce - KTH
K.Strahl - Montreal
JJ.Veillet - LAL
MINUTES
~~~~~~~
o O.Martin - CPPM
Final Results from a complete analogue link demonstrator
Olivier presented final results from a complete analogue
link using inverted-polarity Sandia VCSEL's. Results from
a static test demonstrated that there was a -1.8 to -2 dB
attenuation in the VCSEL light output with a 1m pigtail
connected to a 70m length of fibre (compared to a free
space measurement). For the VCSEL array presented, the threshold
current was approximately 0.7mA for all elements with a
light output of approximately 1.5mW for a bias current of
5mA. A noise equivalent power (NEP) of approximately 1uW was
demonstrated for two of the VCSEL elements for bias currents
up to about 6mA - the NEP increased to approximately 1.5uW
for bias currents up to 8mA. These measurements were carried out
with an oscilloscope. Straight-cut MT connectors were used
through-out. There was no longer any evidence for the slope of
the VCSEL light output changing (onset of a new lasing mode).
Finally, Olivier presented results from dynamic measurements.
He indicated that a dynamic range of 9 bits was easily possible
with 10 bits `a likely conclusion'. These measurements were for a
signal/noise of 3. The differential non-linearity was less than
+-5% across the dynamic range - but more work was needed to reduce
the error bars, especially in the lower part of the dynamic range.
o J.Lundquist - KTH
Gamma and Neutron Irradiation Results for VCSEL's, Fibres
and G-Link Chips
Johan presented results from neutron irradiation (at SARA - Sept
1997) and gamma irradiation (at Karolinska Hospital in Stockholm - Jan
1998) of candidate digital link components. Both these facilities
and the test procedures were detailed by M.Pearce in the previous
meeting. Results from the neutron irradiation of VCSEL's (from
Honeywell and Mitel) indicated that the threshold currents do not
change and that there was a light output attenuation of approximately
10%. Annealing was observed after approximately 12 hours
for a bias current of 5-6 mA (Mitel) / 7-8 mA (Honeywell).
The dose received by the VCSEL's was approximately
(2.0+-0.4)e13 n(1 MeV GaAs) / cm**2. A 10m length of standard fibre was
irradiated to (0.7+-0.2)e13 n(1 MeV SiO2) / cm**2. For a 10m length the
resulting attenuation was almost 20%.
Gamma irradiation results for VCSEL's also indicated no change
in the threshold current. For a `10 years at LHC' dose the VCSEL
light output decreased very slightly. A 10m length of standard
fibre suffered an attenuation of almost 90%, however this scales
to approximately 40% for a 2m length. As observed during the
neutron irradiation, the G-links chips (PLL monitored) did not
behave abnormally during gamma irradiation. Bernard reported on a
communication with an HP engineer who had told him that the G-link
is manufactured in a `sea of gates' technology with 25 GHz transistors.
The resulting small feature size may help to explain the observed
radiation tolerance of the chip.
In conclusion, Johan
stated that VCSEL's appeared resistant to both gamma and neutron
radiation. In order to understand the radiation implications for
fibres (especially for gammas) more information on possible routing
scenarios is needed. Whereas gamma radiation is more damaging for
fibres, neutrons are more damaging for VCSEL's. In the limited
tests of the G-link PLL to date, no adverse effects have been
observed for either type of radiation. Finally, it was pointed out
that the dose rates in both the irradiations are highly unrealistic
at 8500 (1500) times that expected in ATLAS for the gamma (neutron)
irradiation.
o M.Pearce - KTH
Gamma Irradiation Studies at BNL (J.Ye and R.Stroynowski)
Mark reported on results communicated by J.Ye and R.Stroynowski from
SMU. A gamma source at BNL had been used to irradiate Honeywell
transceivers (VCSEL's and PIN's) and 50/125 GRIN fibre from Corning.
A total dose of 200 Gy was given (2.5 ATLAS years) with a dose rate
11,000 times that expected in ATLAS. The tests were totally passive
with no measurements or biasing during the exposure. Before and
after irradiation the transceiver exhibited clear eye-diagrams at 1.24 Gb/s.
In both cases a bit error rate of O(10e-12) was observed. To investigate
the effects of radiation on the fibre, an optical time domain reflectometer
(OTDR) was used. The OTDR operated at 1310nm.
Before irradiation a fibre attenuation of 7.3 dB/km was
measured (included two couplers). After irradiation, no measurement was
possible - the fibre appeared opaque. The attenuation was however measured
using a calibrated laser source (1310 nm) and power meter. An attenuation
of 617 dB/km was measured ! This means the received signal falls below
the transceivers sensitivity. Note that a 2*60m length of fibre was
irradiated - which is unrealistic for ATLAS applications.
o B.Dinkespiler - CPPM
`Poor Man's' Bit Error Rate Tester
Bernard review plans for the construction of a Gb bit error rate tester.
The device would allow on-line testing of links transferring 32 bits at
40MHz with single bit errors detectable. The testing system would have
in-built modularity to allow up to 16 links to be tested simultaneously.
At present a Tektronics pattern generator is used to produce the
32 bit input words - although this could be easily replaced by a simpler
(home-made) dedicated device. The receiver boards will be compatible
with the mini-ROD's to allow compatibility with test-beam electronics.
The initial system will be based on G-link serialisers and Honeywell
transceivers. There is also the possibility to plug in separate optics
boards. For technical details and schematics see Bernard's transparencies.
As an aside Bernard reported that the G-link receiver has an `error-pin'
which would enable simple error checking to be carried out very easily.
An error condition is set if the control word added to the data is
illegal, ie: due to a single event effect. This would probably form the
basis of the first round of radiation tests in March.
o M-L.Andrieux - ISN
Aging Tests at ISN and Kerma Factors
Marie-Laure presented details of two ovens (53 l and 275 l) available at
ISN for aging tests of optical links. Both ovens can be computer
controlled.
She also reported on a revised calculation for the Kerma factor for
SiO2. This work is being carried out in collaboration with Claude Leroy.
The value used to date, 1.3 +- 0.3, seems to have been incorrectly
calculated - this is being looked into.
o M-L.Andrieux - ISN
Amphenol Prototype Status
Marie-Laure started by reporting that CHEVAL had drilled the VCSEL
ceramics and these had been passed by Amphenol for mechanical suitability.
In total, 30 ceramics had now been metallised and 6 had been passed
to Amphenol for electrical tests. The remaining 24 will be sent to
Elmatek for bonding with a set of VCSEL's and PIN's. Amphenol have
made significant progress with the optical connectors as these are now
able to mate correctly. The target of <1 dB connection loss has almost
been met but the outer fibres in the ribbon suffer a loss of approximately
1.5 dB. It is hoped to meet the target by improving the plug moulding and
polishing. The VCSEL will be aligned provisionally using a video camera
coupled to a microscope. It is hoped to release a complete link by
March 1998 in time for radiation testing.
o B.Dinkespiler - CPPM
Status of Testbeam Copper Links
Bernard presented the status of the copper links for the testbeam.
The PCB's are designed and tested and can be used with FEB's
and mini-ROD's as they become available. Six hundred cables salvaged
from CPLEAR have been left at CERN. Plans for the final production
of the links need to be defined with some urgency.
TRANSPARENCIES
~~~~~~~~~~~~~~
Copies of transparencies will be sent to :
o A representative from each of the participating
groups (CPPM, ISN, KTH, SMU)
o The ATLAS secretariat
M. Pearce
(February 12th 1998)
M.P.