Daily Image
20-12-2021.gif)
The heat is on!
| Submitter: | Arno Schoenmakers and the LOFAR2.0 Station Team |
| Description: | The new LOFAR2.0 station will deliver a lot more processing power compared to a LOFAR1 station. This is because LOFAR2.0 is designed to have a factor of 3 more input channels than LOFAR1 as it needs to process all LBA and HBA signals in parallel. This will result in more sensitive observations over a larger bandwidth, but this also implies that more hardware will be switched on, both analog and digital. More hardware implies more heat production, especially inside the subracks where the signal processing boards are concentrated. When all equipment is switched on, and processing signal data, we expect ~0.8 kW of heat production per (low-band) subrack. Since it is vital that the electronic boards are kept within their operating temperature range, we efficiently transport all that heat away from the boards, out of the subracks, and ultimately to the ambient environment of the LOFAR cabinet. Only when the heat dissipation chain is setup correctly, the required > 10-yr lifetime of a LOFAR station can be achieved. All Dutch LOFAR cabinets have a passive cooling system with heat exchangers built into the in the cabinet doors. As a result, the temperature inside a cabinet will always be higher than the ambient temperature. This poses a major challenge for cooling the equipment inside the cabinet, especially in the summer. In the mechanical design of the LOFAR2.0 subrack, special attention has been given to cool the subrack efficiently. For each board in the subrack, the so-called hotspots have been identified and measures have been taken to be able to effectively cool these hotspots. The airflow through the subrack has been carefully balanced to match the amount of heat dissipation, and where needed, air ducts and cooling ribs have been added, such as on the FPGAs. But are all these measures sufficient? To find out, we need to run tests. Tests of the cooling capacity of the cabinet itself have already demonstrated that the temperature inside a LOFAR cabinet can reach ~50 degree centigrade on a calm warm sunny summer day. To find out how the boards inside a subrack fare under those circumstances, we have put our test subrack inside ASTRON's climate chamber, which can be heated to these high temperatures. All kinds of safety measures were put in place to prevent the system from overheating or becoming damaged, and constant monitoring of the system at several critical points would provide the data. After careful preparations, the test was executed last Friday. We gradually increased the temperature in the room, while monitoring the temperatures in the subrack. In the images, you can see the test setup and some of the measurement results. The test showed that the designed cooling setup suffices to keep the vital points in the subrack below or at allowed maximum temperature even when the environment is heated up to 55 degrees Celsius, which is 5 degrees above the required threshold. That means that we have a system that indeed can operate inside a real LOFAR environment! On the other hand, we noticed quite large differences in the temperatures, of up to 10 degrees, between the separate FPGAs on a single Uniboard, which is a difference that we couldn't explain immediately. No blocking issues were found, and we haven't lost our only test setup. This means that we can continue testing with the subrack and ultimately will deliver a LOFAR2 system having high temperature robustness and thus improved availability in summer conditions. |
| Copyright: | ASTRON |
| Tweet |  |