***** Minutes revised on 15-Apr-1997 concerning NIEL calculations *****
Agenda for SCT links meeting
The meeting was held on Thursday, February 27
at 14:00 in room 40-R-D10.
14:00 Irradiation studies. Stephen Oglesby 15'
14:20 Irradiation and lifetime studies. Juerg Beringer 25'
15:00 Results from prototype optical link Tony Weidberg 10'
15:15 Shielded twisted pair tests. Stefan Marconi 15'
15:30 Coffee
16:00 VCSEL packaging John Dowell 10'
16:10 DC coupling scheme for optical links Tony Weidberg 15'
16:30 Planning for links technology choice Bob Kowalewski 20'
17:00 Working meeting for links section of TDR 60'
*********************************************************************
Stephen Oglesby presented results on neutron irradiation of GEC
packages containing PINs and LEDs. 4 LEDs in 2 packages and 3 PINs in
3 packages were irradiated at the Birmingham dynamitron with 1 MeV
neutrons at dose rates of 1-5*10^9 n/cm^2/s for a total fluence near
1.7-2.1*10^14 n/cm^2. All of the devices had been previously
irradiated to 5.5*10^13 p/cm^2 (24 GeV protons) and subsequently
annealed.
The LEDs recovered to relative light outputs (RLO) of 35-40% after the
initial proton irradiation. Annealing was done at room temperature
and with currents of 10 to 50mA. A DC current of 10mA was maintained
throughout the irradiation period. The RLO (normalized to the value
after annealing the proton induced damage) dropped to about 15% after
the neutron irradiation; annealing of these devices is in progress.
Stephen calculated damage constants of 6.53*10^-14/cm^2/proton and
1.15*10^-14/cm^2/neutron; the ratio of n:p is 1:5.7. He calculates
and expected ratio using NIEL stopping powers of 1:5.1.
**revision as of 15-Apr-1997
Juerg and Stephen discussed the difference between the NIEL equivalents
they calculate, and of the comparison with observed damage. The NIEL
ratio of 1 MeV nuetrons to 24 GeV protons is 1:5.3 according to Juerg,
while the ratio for ISIS neutrons to 24 GeV protons is 1:3.2. He
expects the value appropriate for the neutron spectrum from the
Dynmatron to be in between these values. He also pointed out that
the damage constants should be measured before any annealing takes
place, and that the irradiations done with LEDs under bias do not
satisfy this condition. This could explain the differences observed
for the damage constants.
**end of revision
The PINs had responsivities of 0.3-0.55 A/W before proton irradiation
and were annealed to values from 0.3 to 0.45 A/W. In the subsequent
neutron irradiation the 3 devices showed a marked loss of
responsivity, down to values in the range 0.02 to 0.05 A/W. In
addition the dark current rose sharply from <20nA to 300-500nA.
The PIN behaviour under proton irradiation was as had been expected
based on information from GEC. However, the neutron irradiation
results are much worse than expected. Several new GEC packages
with 3 new-style epitaxial silicon PINs (these have faster rise
times at 5V than the devices in the packages irradiated to date)
should be available within 1-2 weeks. Tony expressed concern about
possible rate effects (the dynamitron is a very hot source) and will
arrange an irradiation at ISIS as soon as possible.
ACTION > > > > Irradiate new PINs at ISIS au plus vite
* * *
Juerg Beringer presented further results on irradiation and lifetimes
studies of GEC LEDs at Bern. Improved statistics are now available,
as are new results from neutron irradiation. The fluences in the
neutron irradiation were 0.7*10^14 n/cm^2 and 4.4*10^14 n/cm^2. The
devices have been annealing at 50mA and -10C for 8 days. RLOs at
present are near 85% and 50% for the two batches, respectively.
Juerg showed that the GEC LEDs that have a high initial light output
appear to anneal more rapidly than initially weaker devices; the
asymptotic behaviour is however, not yet known. Juerg also finds
good agreement between proton and neutron damage constants and the NIEL
calculations, but in his case he finds n:p=1:3.1 and calculates a
ratio of 1:3.2 after folding in the neutron energy spectrum from ISIS.
The aging studies are in progress. The aging acceleration factor,
partly from continuous DC operation (i.e. time spent at high current)
and partly from temperature effects (test being run at 50C), is at
least 120 (12 from continuous operation, >=10 from temperature). A
few LEDs (3/119 ABB and 1/31 GEC) can be considered dead, but all of
these failures occurred before the high temperature phase of the
lifetime study. The duration of the lifetime test now corresponds to
20-30 years operation in the SCT. About 80 additional GEC LEDs will
be added once they have finished annealing.
The aging tests should be finished in mid-April. Reirradiation of
ABB LEDs, GEC LEDs, and VCSELs is in progress to study the effect
of interim annealing on the RLO versus total fluence.
VCSELs:
63 Sandia VCSELs have been operating for the same period, with one
post-irradiation failure. The aging acceleration factor in this case
is not yet known. About 80 additional devices will be added once
they have finished annealing.
The LAr group is still investigating Sandia VCSELs as an option and
have been studying packaging (with Amphenol). The relevant physicists
will be here during the week of March 10-14. Their meeting is Tuesday,
March 11 at 9:30. Bob will attend, and other interested parties are
encouraged to do so.
ACTION > > > > Bob will attempt to arrange for a private discussion
with those colleagues working on VCSEL packaging.
ACTION > > > > Bob will contact Sandia about VCSELs.
* * *
Tony Weidberg reported on results using a prototype link. The link
uses an existing GEC pigtailed package with the LEDs driven by a CMOS
rad-soft version of the LDC, and receives the signals on a fast PIN
diode followed by a low noise transimpedance amplifier (from Pierre
Jarron) and a separate commercial discriminator. In a production
system the amplifier and discriminator would be combined into one
chip, and a slower PIN with higher responsivity would be used. Tony
showed a plot of signal to noise as a function of the amplifier
rise time (which can be tuned), and found a broad maximum with S/N>200
for rise times between 8ns and 25ns. He used this measurement to
estimate the safety factor of LED-PIN links after irradiation. The
table with the results will appear in the TDR; the final S/N after
irradiation is expected to be of the order of 50. It was agreed to
show the distribution of S/N one expects after irradiation based on
the distribution of RLO for the GEC LEDs measured at Bern. This plot
will be included in the TDR.
ACTION > > > > Juerg will produce a plot of GEC LEDs irradiated with
approximately 2*10^14 p/cm^2 as a function of RLO (and
expected S/N).
* * *
Stefan Marconi presented information on the twisted pair option.
Measurements of stp transmission properties have been made using the
board designed by Peter Lichard of the TRT. Signal properties over
10m of twisted pair cables show good performance at both 40 and 80
Mbit/s, both for the full 400mV LVDS voltage swing and for
transmission at 1/2 LVDS levels. The cables contained 27 pairs, 20
with signals and 7 dummies for measuring cross talk and noise. The
cross talk was at the level of 5mV r.m.s., to be compared to signal
swings of 1/2 LVDS levels (approx. 250mV). The connection point for
the shields had no noticeable effect on the transmission. Common mode
signals (from DC to 40MHz) of 1V, injected between the driver and
receiver circuits, had no effect on the transmission. Common mode
levels of 2V still allowed acceptable transmission. Plots can be
seen at "http://wwwcn.cern.ch/m/marconi/www/public/stp/100stp.html"
Measurements on longer (20m-80m) lengths of standard gauge (AWG 28)
cables indicate that compensation will be needed for lengths beyond
50m (see "http://wwwcn.cern.ch/m/marconi/www/public/stp/fullink.html").
Stefan mentioned plans for full prototype twisted pair links, which
will comprise a 10m length of 24 pairs of fine stp, a repeater based
on existing LVDS-TTL, TTL-LVDS chips, and 80m of standard twisted pair
cable. A compensation circuit for the 80m section will be necessary,
and is under study. Three 24-pair links will be made, and will be
available for use in the TSP-97 program.
His calculations of the stp cable material budget in the inner
detector indicate a worst-case total of 0.9% X0 up to the last SCT hit
and 2.5% X0 to the last TRT hit (see
"http://wwwcn.cern.ch/m/marconi/www/public/stp/budget.html").
In addition the space needed in the cable runs within the inner
detector was calculated, and found to be about 2 cm^2 per 36 modules
(i.e. 6 24-pair cables). This exceeds the space reserved for optical
fibres (0.4 cm^2) in the current engineering layout.
* * *
John Dowell reported on discussions with GEC-Marconi regarding the
mounting of VCSELs in place of LEDs in the existing packages. The
alignment would be done to <10um via solder bump bonding. A revised
package using less silicon and more ceramic is proposed to reduce
cost. The VCSEL packaging could proceed as soon as VCSELs are made
available.
ACTION > > > > Bob will determine if Sandia can supply VCSELs before
cutting to allow post-processing by GEC, and will also
inquire about possibilities for VCSEL packaging from
Sandia.
* * *
Tony Weidberg proposed a scheme for a compensated DC-coupled system
for the data receiver. The use of an NRZ scheme rules out pure AC
coupling, and pure DC coupling makes unrealistic demands on the
stability of the amplifier. A scheme using a microprocessor to do
a calibration was proposed (the idea comes from Alan Rudge). It
will be described in the TDR. Some discussion ensued regarding
the timescale on which calibrations would need to be performed.
It was pointed out that such a system provides monitoring "for free."
* * *
Bob Kowalewski proposed a procedure for arriving at a technology
choice for the SCT data transmission. The proposal was slightly
modified after discussion, and can be found at
"http://wwwcn1.cern.ch/~kowalews/SCTlinks/choice.html"
Further comments are welcome.
*********************************************************************
These minutes can be found from web page
http://www.cern.ch/Atlas/GROUPS/INNER_DETECTOR/sctnew/Electronics/links/