---

Agenda:

• 1- Approval of the previous set of minutes.
• 2- Tests and results from ECP on cooling in power racks.
• 3- Proposal from St. Petersburg for production of LCS racks for ATLAS.
• 4- AOB.

 

---

Short Summary:

Results from the tests carried out by ECP/ESS/GI show that if necessary (environment, thermal capacity, etc.), direct liquid cooling can be employed with the electronics racks, but there are certain details to be worked out in practice before full-scale use can be contemplated.

With its long experience in Leakless Cooling System, PNPI proposed to produce electronics racks for ATLAS with their cooling system and was ready to provide prototypes before the end of the year.

---

Detailed Minutes:

• 1. Approval of the previous set of minutes:

No comments.

• 2. Tests and results from ECP on cooling in power racks:

G.Dumont delivered a full report on initial results from ECP's tests on direct water cooling of electronics boards. For reasons already discussed under item 2 of the preceding meeting (meeting#6 16/04/97) ATLAS had requested the ECP/ESS/GI section to study a system of liquid cooling as applied to standard electronics modules, to compare its effectiveness with that of conventional forced air ventilation and finally, to study its impact on such modules. Potential users of the application were all the LHC experiments, on the racks in the control rooms and the main underground areas, but also on the electronics "mounted" on the detectors.

• 2.1. Equipment:
  

To run these tests the section fitted a 19"-38U standard rack with a Leakless water cooling system, which distributed the water over cooling screens via a primary exchanger ( Figure7-1 and Figure7-2 ). Two standard CERN VMEbus Type V430 crates were installed in the rack, to which were connected module boards with a 62[W] charge, i.e. with a thermal flux of approx. 0.2[W/cm2]. The boards were fitted with temperature probes in different positions, as indicated in Figure7-3 . On each board a DATE type aluminium cooling screen was mounted, secured on it by spacer studs ( Figure7-4 ), water being circulated over the screens from in front ( Figure7-5 ). Water was circulated at approx. 16deg.C for an ambient temperature of 23deg.C. The water flowrate on each board was 0.5[l/mn].

• 2.2. Measurements made:
  

-Tests without ventilation systems

-Tests with standard ventilation systems

-Tests with water-cooled screens:

- Screen on electronic components side:

-Without thermal paste

-Varying distance

-With paste

-Variable paste thickness

- Screen on soldered side:

-Without thermal paste

-With thermal paste

• Figure 7-6 shows the effect of conventional cooling by fans which, apart from the constraints referred to above, is very satisfactory for the heat flux tested on the board with a maximum of 60deg.C.
• Figure 7-7 compares the effect of the cooling screen's position depending on whether it is located on the electronics components side or the soldered side (without heat seal). It can be seen that its effectiveness will be greatest when the screen is on the components side. This measurement holds true for one board but will have to be confirmed for all boards and screens when the latter are mounted side by side.
• As shown in Figure 7-8 , and logically, the effectiveness of the screen is directly proportional to its distance from the board, and at distances greater than 1 mm hot spots of greater than 80deg.C. can be seen to be caused.
• The measurements quickly led the team to employ a heat seal between the board and the cooling screen and Figure 7-9 clearly demonstrates its great effect; it is important to note in particular the difference between Delta T of the water from the screens which really shows the difference in efficiency, with or without the thermal paste. Two types of paste have been tested: Gap Pad™ from Bergquist and DATE paste; both have a Silicone base with aluminium additives and give the same results (thermal conductivity of approx. 1.5[W/m.K] and electric strength of 20[kV/mm]). The difference relates to their appearance and mode of use: Gap Pad™ being a laminated 60 shore foil available in different thicknesses whereas DATE comes in the shape of a malleable paste that can be spread over the components.
• With the introduction of heat seals, the problem of pressure from the cooling screen on the paste arose; since the screen was mounted on the board by its four corners it naturally tended to buckle and its contact was not uniform; moreover, the same phenomenon also appeared to have an influence where the configuration did not include heat seals. Tests were therefore performed using a clamping system to keep the pressure of the screen against the board uniform -- the results are very important in demonstrating the scale of the phenomenon ( Figure 7-10 ).

Figure 7-11 summarizes all the results from the different configurations and the table below, which ranks them in order of effectiveness, produces no major surprises: direct cooling of the electronics boards using water screens is effective with a heat seal or a system using screen pressure and much more effective than ventilation systems using both.


	Cooling methods	T°C maxi	T°C mini	Delta T

1

No cooling

121.5

92.5

29

2

Without Silicone rubber

No pressure

82.2

57.1

25.1

3

Forced air cooling

60.6

38

22.6

4

With Silicone rubber

No pressure

60.1

26.8

33.3

5

Without Silicone rubber

With pressure

56.3

35

21.3

6

With Silicone rubber

With pressure

54.5

27.6

26.9

• 2.3. Power supply tests:

As the aim of the liquid cooling technique is to keep heat dispersal to the ambient air to a minimum, ECP also joined up with the power supply manufacturers (Wiener and WES-Crates) to develop water-cooled units. As a first prototype Wiener simply took the power module cooled by indirect exchange ( Figure 7-12 ) and tested it on the rack's LCS circuit ( Figure 7-13 ). Numerous temperature probes to measure the various transfers are positioned in the power supply which is placed in an insulated enclosure; to appreciate the true effectiveness of the absorbed system an evaluation was made of the power input. Results given in Table 7-14 and Figure 7-14 show that the cooling efficiency is only 57.5%, which is far too low. A second Wiener prototype has been delivered and is under test, with the aim of achieving complete cooling of the supply and an efficiency of 90%.

• 2.4. Conclusion and continuation of the tests:

The tests show that the direct cooling technique for electronics boards is, as forecast (see item 3, minutes of meeting#6 16/04/97 ), really effective and can be used to replace the system of cooling by fans. At the same time the tests have also raised certain problems regarding implementation of the technique - leaks onto screens and thus buckling of the screen, how heat seals are to be used and contact with heat seal components; before the approach can be used on a large scale these problems must be resolved.
As a next step, ECP must now test boards in their true configuration (components of various sorts with non-uniform heat flux) and determine the best configuration for cooling screens (soldered or component sides, with or without heat seals). Also in the programme are tests on modified Wiener and WES suppplies and tests in magnetic fields.
As mentioned above, the purpose of the exercise is to succeed in drawing off 90% of the heat from a rack in water and at all events to provide those in charge of air conditioning in the control rooms with exact figures for dispersal to the ambient air.
In the subsequent discussion, a number of points were raised:

• The problem again arose of how to measure the temperature inside electronics component housing and to find out its maximum permissible temperature.
• On heat seals, it would be obviously be ideal if each board could be produced with its own molded seal; this is technically feasible with a silicone paste that could be poured and polymerised but that means developing the boards from the outset with that aim in view.
• Experience shows that only certain components dissipate heat on a board, so allowance could be made to identify them and cool them down individually using small-scale screens.
  
• 3. Proposal from Petersburg Nuclear Physics Institute (PNPI ) for production of LCS racks for ATLAS:

A.Krivchitch began by pointing out St. Petersburg 's long experience with Leakless cooling systems. It had supplied all the L3 racks built following the model using air/water exchangers supplied with Leakless and distributed between the crates; and had also built cooling systems for DESY with mixed pressure/Leakless systems. In the future, PNPI proposed to produce electronics racks for ATLAS with their cooling system and was ready to provide 3 prototypes (1 single and 1 double rack) before the end of the year.
The statement related more specifically to the structure of control systems designed for managing 1 or more racks (see Figure 7-16 ).

• 4. AOB:

 

---

The next meeting is to be held on Tuesday 17/06/97, Building 40, room S-A01 from 2.p.m. to 4.pm.

Agenda:

1- Approval of the previous set of minutes.

2- Water cooled power cables.

3- Tests and results on heat transfer of ATLAS Si.

4- Pipe cabling.

5- AOB.

---

Back to CWG Home Page

CERN/P.BONNEAU/02/07/97