==> Next meeting: tuesday, may, 12th, 1998, 9:00 h
 

                        Minutes
                        -------

Present:        I. Crotty       CERN
                Z. Hajduk       CERN
                B. Hallgren     CERN
                R. Hammarstrom  CERN
                P. Jarron       CERN
                M. Letheren     CERN
                F. Linde        CERN-NIKHEF
                P. Farthouat    CERN
                W. van Sprolant CERN
                G. Stefanini    CERN
                R. Richter      CERN-MPI
 
 

F. Linde: N-irradiation test of RASNIK and CAN temperature sensors

The integrated n-dose in 10 years of LHC running (at peak lumin.) in the
region of the ATLAS muon chambers will vary between about 3 * 10**11 and
3 * 10**12 n/cm**2 depending on whether one is at large angles close to
the cavern walls or in the forward area close to the innermost muon
chambers.- The gamma dose in the most unfavourable location is about 1
krad.

A test irrad. of el. equipment was done at the n-irrad. facility PETTEN
(NL). This reactor typically delivers about 10**13 thermal, 10**11
epithermal and 10*11 fast neutrons/cm**2 per day. 60% of the fast neutrons
are between 1 and 3 MeV.- The dose of gamma rays during irrad. is not
known at the moment.
 

Irrad. of RASNIK CCD and R/O electronics:

The RASNIK alignment system was assembled and operating during
the irradiation. During irrad. the contrast of the image recorded on the
CCD was constantly degrading.
Data seem to indicate that the transfer eff. in the CCD was decreasing.
A total dose of 2.7*10**13 thermal, 5.1*10**11 epithermal and 6*10**11
/cm**2 fast neutrons was used.

After irrad. recovery took place and is still ongoing.

Discussion: It is not clear whether the damage observed can be
unambiguously assigned to neutrons and not mainly to accompanying gammas.
CCDs, from their construction, are quite sensitive to ionizing radiation.
The dose of gammas received in this test may be well beyond expected
total dose in the ATLAS cavern.
 

Irrad. of a CRYSTAL-CAN module connected to a temp.- and B-sensor:

Temperature and B-field sensors were constantly read out during the test
(however there was no magn. field). Temperature climbed from an initial 21
to final 28 deg Celsius. After a total dose of 2.7*10**13 n/cm**2 the
contact via the CAN-bus failed.

After a cool-down of several weeks the module was inspected. The reason
for the CAN-bus failure was that the DC/DC converter (type MHP
245S2000) had gone from nominal 5 V to 7.3 V, which blew the fuse. The
CAN-bus opto-couplers and driver (Philips 82C250) however were still
working correctly. It was also discovered that the ADC's voltage reference
had changed by about 9%, which partly explains the observed increase in
temperature.- The voltage regulator (LM2936Z-5.0) had shifted from nominal
5 V to 4.88 V.

Further tests are forseen for RASNIK and CAN-module.

Discussion: with the exception of the DC/DC converter the module survived
a n-dose far beyond the expected total dose in the ATLAS cavern. This is
quite encouraging and shows that a systematic irrad. test program for COTS
is necessary, justified and promising in order to demonstrate the
surviability of standard electronics in the LHC-caverns. However, more
"statistics" on components is needed. One may well limit future irrad.
measurements to the identified critical components to save cost.

Components from several vendors and production batches should be
compared to see whether there is a strong dependance.- The remark on gamma
rates (see above) is of course also valid for this test. If the damage
picture of this CAN-module test is confirmed in future tests, designers
may well consider giving up DC/DC converters for small loads in favour of
rad-tol voltage regulators (see next presentation).
 

P. Jarron: RD49 COTS activities progress report

Pierre announced a workshop on rad-tol issues in Oxford RADECS 98 (sept.
14/15th). Rad. effects on COTS and other equipment will be covered as
used in the nuclear power industry, space- and high energy physics
applications.
 

Rad-tol voltage regulators:

This development project is pursued together with SGS-Thomson. Pierre
reported on the present status.

Aim is rad. tolerance to doses in the calorimeters, which is far above
doses in the caverns (region of muon spectrometers and power supplies).
For this purpose the HF2BiCMOS process has been identified as most
promising. Typical active and passive components have been tested with
gamma doses up to 500 krad. Irrad. with neutrons doses between 5*10**11
and 10**13/cm**2 are foreseen.

The 500 krad irrad. shows:

        - good tolerance of NPN- and PNP- power devices, while min. size
          devices are somewhat more sensitive (though still operative)
        - no problem with resistors
        - no obvious low dose rate effects at 100 rad/s resp. 10 rad/s
        - after irrad. annealing is observed
        - variation between devices: 15% typical
        - apparently large variation between prod. lots

On the basis of this process a voltage reg. will be designed (july 98) and
produced (nov. 98).- For cost reasons a limited set of device types can be
actually produced and LHC experiments have to agree on parameters for
these devices. Pierre proposed to clarify the following questions in the
LHC electronics community:

        - max. output current: 1 A to 3A ?
        - max. output voltage: 5 V to 8 V ?
        - low drop voltage (to save power): 0.5 V to 1 V (@ 3 A) ?
        - flat pack ?
 
        - remote sensing needed?
        - voltage setting: preset in factory or adjustable?
        - voltage sign: also neg. voltages needed?

It is clear that component cost will be significantly higher than for
rad-soft COTS, even at larger quantities. While COTS cost typically 0.5 $,
rad-tol regulators may cost several $ a piece.

Discussion: question whether process should further be optimized to stand
up to 2 Mrad. This is not obviously achievable and will cost time. Nominal
calorimeter and cavern doses (already with safety factors) are OK with 500
krad. Attempt to make rad-HARD volt. reg. may fail, drive cost. Aim of
RD49 was rad-tol development.
 

Pierre than outlined the NA49 concept on COTS:
 

"Blind" rad. tests are problematic because of tracability of components
and production batches. Problem of procurement strategy, risk asessment
and variability between different vendors. Space agencies select COTS on
the basis of long-term experience documented in databases. Qualification
and rad. tests done w.r.t. whole lots, purchased in advance ("single buy
strategy").

Other access to rad. data is collaboration with Sandia and/or CEA. The
Sandia group undertakes large effort ($ 5M/year) to screen and qualify
COTS w.r.t. rad. tolerance. CERN plans collaboration at a cost of $ 50 k
to gain access to Sandia knowledge. Initial period is july 98 to june 99
with the aim to explore knowledge transfer on a good faith basis. However,
no new scientific inquiry by Sandia will be performed under this effort.
For this project an agreement from LHC experiments is needed by may 98.
 

B. Hallgren: Report on the accelerator division's Irrad. Facility

Bjorn described the TCC2 radiation area in LAB II, where the accelerator
division intends to test CERN recommended field-buses. The group is
chaired by Raymond Rausch SL/CO. A typical dose of 50 Gy (5 krad) and
5 * 10**10 n/cm**2 can be obtained in a 6 week running period.

The access conditions are difficult, i.e. cable runs between irrad. zone
and surface counting room are several 100 m long. Shorter cable paths are
foreseen for 1999, which makes tests with powered electronics practically
possible.- Physical access during accelerator down-time is limited to
brief interventions because of a rad. level of 18 mSv (reaching the
allowed human annual dose in 1 hour).

Bjorn showed a list of commercial modules which have already been
successfully tested with doses of 5 Gy. Further irrad. of these units with
doses of up to 100 Gy will be performed. Bjorn plans to test the CAN chip,
the CAN Local Monitor Board and the CAN Special Cable in the same setup.
He also showed a list of potentially critical components which are used in
the CAN electronics (voltage regulators, volt. references, CAN-
transceivers and controllers as well as optocouplers).

Discussion: facility not well suited for LHC needs because of too low dose
rate and infrequent access. To reach equiv. of 10 years of the LHC caverns
needs 1 year in the test area. May be useful for specific low dose rate
tests. Access for signal and power cables mandatory to monitor electronics
during irradiation.
 

General discussion on organisation:

Ph. Farthouat pointed out that there have not been many new results since
june 97, when meetings on rad-tol power supplies were started (with the
notable exception of the RASNIK/CAN test -- see above). This is partly due
to lack of sensitivity of experimentalists to the rad-tol problem. While
Inner Tracker people have always assigned a high priority to rad-HARD
components and have executed an impressive test and certification program,
the problem of rad-tol components in the cavern is largely ignored by
subdetector communities. It must also be noted that subdetectors refuse to
allocate resources considering it a general detector- respectively LHC-
problem. Response from subdetectors w.r.t. voltage and current range needs
was not always satisfactory.

Some organisational activity has been reported at a higher level but
didn't come down to people involved in technical work (certainly not
to Ph. F.). At the moment it is unclear which support is available for
this research w.r.t. manpower and money.- If situation doesn't improve
the question has to be asked whether this meeting (and the more specific
one on power supplies) should be continued.

P. Jarron commented that the funding of RD49 was inadequate given the fact
that 8 LHC experiments are to be served.- There is nobody responsible for
central procurement (of electronics) in the LHC community.- NA49 can give
guidelines for tests and collect information from agencies but has no
manpower for a broad test program.- Pierre also expressed some scepticism
about the LHC-experimentalist's capability to agree on parameters e.g. on
a small number of rad-tol voltage regulators (see above).
 

Possible scenario for rad-tol tests:

Perform n-irrad. tests of typical COTS for power supplies at levels
required for the cavern. N-irradiation has to be carefully monitored
for gamma background to make results meaningful. Determine rad. level
where 50% of components fail. Compare different vendors and different
fabrication batches. Determine safe rad. level. Compare with
industry data bases if available. Create list of certified components and
base PS design on certified components. Vendors have to comply with
certified components in final design.

This procedure avoids the problem that vendors would not disclose
detailed diagrams of their design long before tendering and/or order.

==> Next meeting: tuesday, may, 12th, 1998, 9:00h

Robert Richter