140 likes | 154 Views
Cooling Tests of the. Sector Collector racks. M. Dallavale, C. Fernández, C. Willmott. CMS Week. February 2007. Cooling tests SC rack. CMS Week. February 2007. Sector Collector Racks. Power consumption. X2J22. X2J02.
E N D
Cooling Tests of the Sector Collector racks M. Dallavale, C. Fernández, C. Willmott CMS Week. February 2007.
Cooling tests SC rack. CMS Week. February 2007. Sector Collector Racks. Power consumption. X2J22 X2J02 With a water flow of 10 l/min, it should imply a difference in temperature of the water in and out of 1.27ºC (if the system is 100% efficient). 2
Cooling tests SC rack. CMS Week. February 2007. Test Setup. At B904, one rack was filled with: - 2 LBB crates, - 2 SC crates, - 12 final ROS boards, - 2 TIM final boards, - 12 Dummy TSC boards with the same power consumption that the finals and similar heat distribution. Room Tª 2/3 Heat Exchangers, water cooling, turbine, air deflector, etc. similar to the final system. ROS up.Top SC. TSC up.Top SC. ROS down.Top SC. 12 PT-100 sensors for temperature reading. TSC down.Top SC. ROS up. Bottom SC. TSC up.Bottom SC. ROS down. Bottom SC. TSC down.Bottom SC. Water out Water in Air on top of LBB crate 3
Cooling tests SC rack. CMS Week. February 2007. Water flow. It turned out that first days we had been running with a water flow of 2.5 l/min. Effectively we could run to a stable temperature with one SC crate on, but not with two. One SC powered Two SC powered 52ºC 40.5ºC 3ºC 520 W 4ºC 690 W A filter in the pipe of the water in was almost totally obstructed. It was repaired, and from that point we had a flux of 10 l/min. Water out Tª – Water in Tª = 1,2ºC => we were dissipating 847 watts. We must ensure that in the final system the water flow is at least 10 l/min 4
Cooling tests SC rack. CMS Week. February 2007. Rack thermal isolation. BEFORE AFTER Water out temperature (as well as the rest of the temperatures in the rack) follows much less room temperature. Isolate the rack as much as possible. Insure air tightness. -Heat exchangers (part out of the air circuit). -Separation between crates, air deflector and HE mainly in the bottom part. -Rack sides, the air was escaping. 5
Cooling tests SC rack. CMS Week. February 2007. Rack thermal isolation. HE bottom HE top Difficult access to the HE in the racks YB+2 and YB+1, where RPC cables are already installed. 6
Cooling tests SC rack. CMS Week. February 2007. Temperature map. Sensor on ROS PCB (air temperature) 3 HE Sensor on equalizer at ROS ~ 30ºC Sensor on device under mezzanine. ~ 21.5ºC 7
Cooling tests SC rack. CMS Week. February 2007. Comparison between 2 and 3 HE. Maximum temperature on TSC basically does not change. Air is cooled down by ~ 2ºC. 8
Cooling tests SC rack. CMS Week. February 2007. Air impedance. Effect of blocking the air in the back part of the TSC on the Top SC crate. - Upper sensor decreases by 4ºC (42.7ºC to 39ºC) - Lower sensor increases by 1.5ºC (37ºC to 38.5ºC) 2 HE It seems the air is escaping through the back part of the board since its impedance is lower. Forcing the air to go through the front part helps on dissipating, by means of increasing slightly the temperature of the lower part of the TSC. Equilibrate air impedance in the TSC. 9
Cooling tests SC rack. CMS Week. February 2007. Alignment Crate. Low Power consumption but high air impedance. Covers sensor Alignm. Alignm. ROS And TSC 6U boards 6U boards ROS And TSC Front view Lateral view 10
Cooling tests SC rack. CMS Week. February 2007. Alignment Crate. ~1 ºC <0.5 ºC Move modules to the left. On top of 6U boards. Insert Align. crate Insert Align. Modules On the right If possible, remove covers and move Alignment modules to the left most part of the crate, on top of the 6U part. 11
Cooling tests SC rack. CMS Week. February 2007. Air flow. This is with 6 TSC and 12 ROS. Need of a DSS interlock fast enough that turns off the power supplies when the temperature increases to high levels. -What action is taken in the DSS system when the temperature exceeds a value? How fast is that action? -Is monitorization of the air flow included? -If air flow fails, what type of alarm is generated? -What is the action taken by that alarm? It turns down the Power supplies? Which granularity? 12
Cooling tests SC rack. CMS Week. February 2007. Air Tª at LBB. Turn on 12 TSCs Turn on 12 ROS Two sensors were located in the bottom of the upper LBB crate. One in the front part of the crate and another one in the rear part. The temperature of the air that comes out from the warmest part of the TSC is 32ºC, however, the air that reaches the bottom part of the upper LBB is maximum 22ºC. The temperature of the air of the lowest LBB crate is 19ºC. 13
Cooling tests SC rack. CMS Week. February 2007. Conclusions. • Make sure that the water flow is around 10 litres/minute. • Isolate rack and Heat Exchangers as much as possible with aluminium plates and tape to improve air tightness. • Three Heat exchangers instead of two does not seem to improve much maximum temperatures but could help to cool down the air. • The biggest impact is the air impedance: • Equilibrate as much as possible in the TSC. Compare air impedance of dummy TSC with the final ones. • Alignment crate increases air impedance. • It seems it could be possible to move modules to the left part. • Remove covers. • DSS interlock system to power off crates on temperatures or any rack alarm. 14