Present: P. Bonneau, M. Bosteels, F. Dalla Santa, S. di Pietro, G. Dumont, P.
Farthouat, L. Feld, P. Fontaine, A. Fucci, J. Godlewski, R. Gregory, S.
Grohmann, M. Hatch, A. Hervé, A. Onella, R. Pintus, W. J. Postema,
J. Pothier, R. Principe, B. Righini, A. Smith, W. Van Sprolant,
M. Wilhelmsson and W. Witzeling.
 

1.      Power Supply Cooling
 A. Fucci gave a summary of the responsibilities of the EP-ESS group which
includes the Electronics Pool. He indicated how power supply tests had been
performed in magnetic fields to simulate the situation in the LHC caverns of
ATLAS and CMS. The overall aim was to augment the efficiency of power supplies
by improved cooling which should also provide better radiation tolerance.
B. Righini presented more details of tests carried out by members of the power
supply section. Some results are already published on the web including the
specification for a VME crate and its power supply. He gave details of a water
cooler to be mounted between VME modules so as to provide cooling in regions
where the presence of a strong magnetic field  would preclude the use of
conventional fans. The cooler is 1.5 mm thick which allows it to be mounted in
the space normally available for the circulation of cooling air. However, if
the VME specifications are not accurately maintained by a manufacturer, there
might not be sufficient space for such coolers.  He presented data which
showed that the efficiency of the cooler was better when it was mounted on the
solder side of a PC board, was greatly improved by adding a layer of
conductive rubber between it and the board, and could be further increased by
clamping it to the board. The water connections are made on the front of the
VME module but an acceptable (small) water connector has not yet been obtained
though Staubli may be interested in producing one provided that a sufficiently
large order can be guaranteed. Development of this system has been halted
until potential users are found. A. Hervé later commented that there was not
much enthusiasm in CMS for this system but it would be kept in mind for
possible use where the magnetic field would give problems for conventional cooling.
B. Righini continued by presenting a system for fanless cooling of power
supplies. After initial tests by ESS members, this had been further developed
in conjunction with a German firm who had been given a contract to produce a
prototype. This prototype contains a primary water cooled radiator in thermal
contact with secondary radiators (coolers) which cover the individual power
supply units (5V, 5.2V, 12V etc). The water connections are made with easily
affordable plastic self-sealing connectors. The power supply had a nominal
power of 1 KW, but the cooling effficiency was sufficiently high to enable it
to be tested up to 1.22 KW. The water cooling efficiency remained close to 90%
as the suppply was run up from zero to 1.22 KW in 250 W stages.
 
2.      ATLAS Rack Cooling Plans and Tests
S. di Pietro explained that he had started to work on conventional racks with
fans and water cooled heat exchangers with the aim of decreasing the noise
level and also decreasing the heat loss to the atmosphere. Two-speed fans had
been obtained and several types of heat exchanger had been considered for use.
A 4-crate rack would have 4 exchangers with the last being placed at the top
so that the air recycled to the bottom of the rack would be cool as it passed
down the walls and hence radiated less to the atmosphere. It was pointed out
that, as racks are mounted in rows, there will only be 2 outer walls per row
so that this might not lead to a great increase in cooling efficiency. Many
suggestions were put forward for measurements and modifications to be included
in the series of tests which were planned. The CCG will follow up the results
of these tests.

3.      CMS Rack Cooling Tests
R. Principe presented a summary of results from tests carried out on the
closed rack systems of ALEPH, DELPHI and L3 (OPAL has open rack cooling).
After the first tests were made on the racks, some improvements were made and
further tests carried out. The power dissipation in each rack was typically
between 4 and 6 KW. The ALEPH and L3 racks could be cooled with 85% efficiency
which rose to about 90% for the DELPHI racks which were better sealed. As a
result of these tests, CMS believed that 90% cooling efficiency could be
obtained relatively easily and that this would be sufficient for their needs
in the counting rooms where most racks would be situated. ATLAS was aiming to
have 95% efficiency, but it was suggested that this would only be needed if
all their racks used maximum power.  Thus, it was likely that the ATLAS
requirement was based on an overestimation of their total power and that an
efficiency of 90% might be acceptable to them as well.

4.      Rack Cooling for COMPASS
M. Bosteels had been developing a cooling system which could be adapted for
use in the many different types of racks which COMPASS was likely to assemble
from its institutes which are spread over many countries. The system should be
capable of being applied to both closed and open racks. He believed that heat
exchangers should always be placed at the top of crates where the air was hot
and would lead to higher efficiency. Cooling water at 2C above the dew point
would be used in an underpressure system supplied by a pump operating at 2
bar, with the return water being in free flow to a reservoir kept at 500 mbar
absolute. He presented a standard commercial collector with each in/out
channel having a valve on the input and a pre-setable valve on the return. A
7-channel collector costs only 250 CHF and up to 12 channels may be specified
on a standard collector. He also presented an efficient design of heat
exchanger of 1 U height and with parallel vanes. A typical flow rate would
involve a pressure drop of about 800 mbar across this heat exchanger. The low
pressure in the return reservoir could be used to check that the system was
efficiently leak tight before start up. Only a small reservoir is needed for a
water cooling system as it can be drained for maintenance, but use of a
fluorocarbon coolant would require a reservoir sufficient to hold all the
coolant during maintenance or modifications.  It was estimated that such a
cooling system would cost about 40 KCHF in total for a typical control room.

Conclusions:  Rack cooling seems to be under control for both ATLAS and CMS.
Solutions have been identified for fan-less cooling of electronics and power
supplies, but these will only become available commercially if sufficient
orders are placed.

5.      A.O.B.

There was no other business. As progress with evaporative cooling is being
closely monitored in a series of special ATLAS SCT meetings, the agenda of
next CCG will focus on liquid cooling.

 
Back to Home Page.