Friday February 27 at 9:00 in 40-R-D10.
1. Minutes of last meeting.
Approved.
2. Measurement in magnetic field. Status of the set-up. William Van Sprolant
The test set-up will be available mid-MArch for measurement. The ESS
group will test fan trays and standard power supplies. Gilbert Dumont (gilbert.dumont@cern.ch)
should be contacted to get access to this facility.
3. Updated specifications of the inner tracker power supplies. Z. Hajduk
ZH presented a new version of the common
specifications. Only the basic element is specified and one should
now go ahead and see whether it's possible to specify the system which
will house these elements.
A few remarks were made:
- having a single basic can save money (CAEN).
- the frequency spectrum of the ripple must be specified.
- the cooling is not addressed. One should just specify that the racks
housing the PS must be closed and cooled.
ZH will recontact the manufacturers in order to have more feedback.
In particular it would be of interest to have a relative quotation of the
different parts.
4. HEC requirements. J. Fent (MPI)
Jurgen presented the requirements of the hadronic end cap calorimeter
(HEC). They need three voltages for the cold electronics (8V/400
mA, 4V/500 mA, 2V/100mA) and three for the warm electronics (8V/800 mA,
4V/400 mA, -3V/100mA). It is still not yet clear whether they can use switching
power supplies.
5. Summary of the radiation test meeting. R. Richter
Neutron- and gamma-rates in the caverns of ATLAS and CMS are of the
same order
of magnitude. Though these levels are much lower than the ones in the
inner
cavity and in the calorimeters, they still are of concern w.r.t. the
survival
of standard components (COTS=components-off-the-shelf). The RD49-collabo-
ration has shown that doses as low as 10**12 n/cm2 can virtually destroy
certain bipolar transistors.
While many measurements are available on gamma-irrad. (as documented
e.g. in
the data bases of space agencies) little is known on n-irradiation.
There was
agreement that n-measurements have to be done in the months to come.
A first
round of measurements should scan a wide range of doses, e.g. 10**10
to 10**13
n/cm2 in powers of 10. P. Jarron volunteered to inquire on available
irrad.
time in Saclay. There is also the spallation source ISIS at RAL, while
the
PROSPERO facility in France delivers n's with a very low background
of gammas.
J. Elias said that there was access to a clean n-beam made from Cf-spallation
plus moderation (to obtain a well defined n-spectrum).
P. Jarron pointed out that there might be a possibility to obtain access
to
the database of SANDIA, which contains information on neutron irradiation
damage.
There were 3 presentations given:
P. Jarron: General aspects of radiation damage
----------------------------------------------
Pierre summarized the findings of RD49. CMOS and bipolar devices have
quite
different damage mechanisms under irradiation. Power devices are particularly
sensitive. CMOS because they are qualified to rel. high voltages (20-50
V) and
therefore need a "thick" gate oxid of about 50 nm, which gets polarized
by
ionizing rad. (Modern CMOS-logic circuits only have a gate thickness
of about 6
nm and consequently are much less prone to damage.)- Bipolar transistors
need
large base widths of the order of 1 to several micron.
Pierre gave a list of potentially sensitive device classes:
- linear bipolar ICs (volt. reg., amplifiers, comparators)
- optocouplers, which sometimes
are "embedded" in ICs, are very
sensitive to
n-displacement damage
- CMOS is generally less
sensitive to n's. But there are exceptions, so
screening in
necessary.
These issues have to be kept in mind for power supplies. Under certain
conditions information can also be obtained from Matra Marconi Space
and from
CEA-DEIN. Pierre will further look into this possibility.
G. Stefanini: CMS tracker LV power distribution and cable cooling
-----------------------------------------------------------------
As an example of the issues to be considered, Giorgio brought up the
case of the PS-system of the CMS inner tracker (taking into account
Si and
MSGC strips only). Similar considerations would also apply to other
subdetectors.
In the layout currently studied, it is assumed that the 12 million
(approx.) strips are subdivided into 1500 (approx.) groups, each
groups of around 8k strips being connected to an independent LV power
supply (+2V, -2V and return rails).
The segmentation is required for reliability reasons and limited by
cost
considerations. The supply currents for each group are -9.6A and +3.6A,
respectively. The total power consumption (at the FE) is about 40 kW.
Problems in the cavern are magn. field of up to 800 G and limited
accessibility. Problems of rad. have been mentionned before. The potential
failure modes of PS's must be studied carefully in order to protect
FE electr.
from destructive malfunction.
If PS's were located in the counting room an additional mass of 44 tons
of
Cu-cables would be necessary. This would dissipate an additional 32
kW of power
and a water cooling system would therefore be necessary. However this
would avoid all problems of radiation damage, magnetic field and
accessibility.
Giorgio gave a list of key issues for the power supplies at LHC. E.g.:
linear
vs. switching, wallplug efficiency, floating grounds/voltages, load
protection
and a conclusive concept for safety, control and reliability.
Discussion: a detailed comparison of cost (and feasibility) should
be
done (despite obvious difficulties) of PS in the cavern vs. PS in the
counting room.
J. Elias recalled new electro-magnetic compatibility (EMC) regulations
requiring additional circuitry in PS's to prevent el. perturbation
to the
outside. This is already done at CDF. This circuitry is another candidate
for screening of rad. tolerance.
B. Hallgren: Critical components in real-life circuits
------------------------------------------------------
a) VME-PS's
-----------
Bjorn showed the diagrams of a commercial VME-PS. He identified a number
of
potentially critical components like rectifier diodes, N-channel power
MOSFETs,
OP-amps, opto-couplers and micro-processors plus related logics and
memories.
It would of course be advantageous to test these components separately
(i.e.
not inside an together with the whole PS) because locating the faulty
components in a defective PS may be very difficult.
b) the ATLAS DCS LMB (local monitor board)
------------------------------------------
This module is used in the region of the ATLAS Muon chambers for measuring
voltages, temperatures etc. It is based on the CAN-bus standard, which
was
originally developed for the car-industry. Standard COTS are used,
e.g. the
voltage regulator ADP3301, ADM660, a 16-bit ADC (of which 12 bits are
used) and
opto-couplers.
Bjorn mentioned the test which NIKHEF is currently performing at a Dutch
nuclear test reactor. A CAN microcontroller and a Crystal Semiconductor
ADC
are exposed to the neutron source during 8 hours, which is supposed
to be
more than the equivalent of 10 years of ATLAS operation in the Muon
chambers
(up to 10**13 n/cm2). The CAN node is operated during the test and
results are
logged continuously, so the mode of failure, if any, will be documented.
General discussion:
-------------------
Main aim is to verify whether LV-PS's (and perhaps HV-PS's) can be used
in
the caverns. Therefore critical components must be identified and tested.
Need agreement on strategy inside collaborations. A workshop is foreseen
for
may, 4/5/6 (?) to get input from larger community.
Define program for tests, use potential input from SANDIA (P. Jarron).
==> Next meeting: monday, march, 30th, 1998, 9:00h
5. Preliminary results of tests of CAN interfaces in radiation. F. Linde
Some very preliminary results of a CAN interface irradiations were shown.
A full system has been put in a neutron beam to receive approximately 5
times the expected dose for a 10-year period in LHC, in about 10 hours.
The actual dose the system received is not yet perfectly known (calibration
to be done).
After 3 hours (3 years of LHC) the system failed (broken connection).
It is not yet known which part of the CAN interface died.
6. Status of the LArg power supplies. M. Citterio
Mauro presented the work made in BNL. They are using DC-DC convertors
from VICOR which can provide up to 50 W per cubic inch and be assembled
like briks to build high power PS. They accept high voltage DC input (400
V). In addition a ripple attenuator module (called VI-RAM) associated with
the DC-DC converter allows very low ripple level (0.25 mV peak to peak
for a 5 V 20 A device). The noise degradation of the front-end electronics
is very small at that level (equivallent input noise goes from 51.3 nV
when linear power supplies are used to 53.9 nV with these DC-DC convertors).
The behaviour of these elements in a magnetic field depends on the field
orientation. It works up to 180 Gauss in the worst direction.
To power a full front-end crate (3 kW) a volume of 21 liters is necessary.
This can be housed inside the detector and would solve the problem of the
services for the HEC. In addition, at that location the magnetic field
is very low and these convertors would work.
Radiation tests are in progress as well as water cooling studies.
7. Next meeting
Monday March 30 in conjunction with the radiation test meeting.