Cooling System Overview
The Cisco CRS 16-slot line card chassis cooling system includes the components and control system that draw ambient air through the system to dissipate heat and keep the system operating in a desired temperature range. The complete Cisco CRS 16-slot line card chassis cooling system includes:
- Two fan trays
- Two fan controller cards
- Temperature sensors distributed on cards and modules in the chassis
- Operating software that controls the cooling system
- Air filter
- Inlet and outlet air vents and bezels
- Impedance carriers for empty chassis slots
- Power module cooling fans (fixed configuration only)
16-Slot Line Card Chassis Airflow
The airflow through the FCC is controlled by a push-pull configuration (see the following figure). The bottom fan tray pulls in ambient air from the bottom front of the chassis, and the top fan pulls the air up through the card cages and exhausts warm air from the top rear of the FCC.
1 |
Front (SFC side) of chassis |
6 |
Power shelves |
2 |
Room air |
7 |
Air exhaust |
3 |
Bottom fan tray |
8 |
Upper card cage |
4 |
Air filter |
9 |
Lower card cage |
5 |
Top fan tray |
10 |
Rear (OIM side) of chassis |
Note |
The Cisco CRS 16-slot line card chassis has a maximum airflow of 2050 cubic feet per minute. |
The bottom fan tray pulls in ambient air from the bottom front of the chassis and the top fan pulls the air up through the card cages where the warm air is exhausted out the top rear of the chassis.
The chassis has a replaceable air filter mounted in a slide-out tray above the lower fan tray. The Cisco CRS 16-slot line card chassis air filter, shown in the below figure , plugs into the rear (MSC) side of the chassis.
You should change the air filter as often as necessary. In a dirty environment, or when you start getting frequent temperature alarms, check the intake grills for debris and check the air filter to see if it needs to be replaced. Before removing the air filter for replacing, you should have a spare filter on hand. Then, when you remove the dirty filter, install the spare filter in the chassis.
Note |
A lattice of wire exists on both sides of the air filter with an arrow that denotes airflow direction and a pair of sheet metal straps on the downstream side of the filter assembly. |
Cooling System Operation
The fan control software and related circuitry varies the DC input voltage to individual fans to control their speed. This increases or decreases the airflow needed to keep the line card chassis operating in a desired temperature range. The chassis cooling system uses multiple fan speeds to optimize cooling, acoustics, and power consumption. There are four normal operating fan-speeds and one high-speed setting used when a fan tray has failed.
At initial power up, control software powers on the fans to 4300 to 4500 RPM. This provides airflow during system initialization and software boot, and ensures that there is adequate cooling for the chassis in case the software hangs during boot. The fan control software initializes after the routing system software boots, which can take 3 to 5 minutes. The fan control software then adjusts the fan speeds appropriately.
During normal operation, the chassis averages the temperatures reported by inlet temperature sensors in the lower card cage (or in the upper card cage if the lower cage is empty). To determine the appropriate fan speed for the current temperature, the fan control software compares the averaged inlet temperature to a lookup table that lists the optimal fan speed for each temperature. The software then sets the fan speed to the appropriate value for the current temperature. The temperature ranges in the lookup table overlap to ensure a proper margin to avoid any type of fan speed oscillation occurring between states.
Note |
When there are no active alarms or failures, the fan control software checks temperature sensors every 1 to 2 minutes. |
Thermal Alarms
Local thermal sensors (on individual cards) monitor temperatures and generate a thermal alarm when the cooling system is not cooling properly. A temperature sensor might trip in response to elevated ambient air temperature, a clogged air filter or other airflow blockage, or a combination of these causes. A fan failure causes a fault message, but if no thermal sensors have tripped, the fan control remains unchanged.
When a thermal sensor reports a thermal alarm, the sensor passes the fault condition to its local service processor (SP), which then notifies the system controller on the route processor (RP). The system controller passes the fault condition to the SP on each fan controller board. The fan control software then takes appropriate action to resolve the fault.
When a thermal sensor trips, the fan control software tries to resolve the problem (for example, by increasing fan speed). The software performs a series of steps to prevent chassis components from getting anywhere near reliability-reducing, chip-destroying temperatures. If the fault continues, the software shuts down the card or module to save components.
Quick-Shutdown Mode
The fan controller cards and fan trays have a quick-shutdown mode that kills power when a card or fan tray is disengaged from the chassis midplane. The quick-shutdown mode minimizes inrush current during a hot swap or OIR. In normal maintenance conditions, the software gracefully shut downs the power to the failed part, allowing ample time for capacitors to discharge.
Fan Controller Redundancy in the Line Card Chassis
The main feature of the Cisco CRS 16-slot line card chassis cooling system is fully redundant fan control architecture. This architecture, which systematically controls the speed of the fans for various chassis-heating conditions, is redundant from both a power standpoint and a cooling standpoint. The architecture supports a redundant load-sharing design. The Cisco CRS 16-slot line card chassis cooling architecture contains:
- Two fan trays, each containing nine fans
- Two fan controller cards
- Control software and logic
The chassis is designed to run with both fan trays in place.
Both fan controller cards work together to provide fully redundant input power and control logic for fan trays and fans. Each fan controller card receives its input power (-48 VDC) from both the A and B power shelves. The fan controller card then provides one fan tray with input power from the A bus and provides power to the other fan tray from the B bus. This feature ensures that the upper fan tray is powered from the A bus on one fan controller card and from the B bus on the second fan controller card.
In a fully redundant system-one that is equipped with dual power feeds, dual fan controller cards, and dual fan trays-the cooling system can withstand the failure of any one of the following components and still continue to properly cool the chassis:
- Fan tray—If one fan tray fails or is removed, the other fan tray automatically speeds up to the maximum limit and provides cooling for the entire chassis. (If multiple fans in a single fan tray fail, the remaining fans in the two fan trays provide cooling for the entire chassis.)
- Fan controller card—If one fan controller card fails, the other fan controller card provides all of the power to the fan trays. In this mode, the single remaining fan controller card provides a maximum of 24 VDC.
- Power shelf or power module—If one power feed fails, the other power feed provides all of the power to the fan trays.
In the single-failure cases described in this section, the rotational speed of the remaining operational fans changes automatically according to the cooling needs of the chassis.
A double-fault fan failure involves two fan trays, two fan tray boards, two fan controller cards, two power shelves, two power modules, or any combination of two of these components. When a double-fault failure occurs, the system can automatically power down individual cards if the cooling power is insufficient to maintain them. The chassis remains powered on unless both fan trays have failed or thermal alarms indicate a problem serious enough to power down the entire chassis.
Note |
When a cooling system component fails, it should be replaced as soon as possible. |
For information on the rotational speeds of the fans in revolutions per minute (RPM), see the following section.