High data-rate, small form-factor, pluggable, optical transceivers (SFP+) are the industry’s answer for demanding applications of higher bandwidth and port density, including the challenges related to excess heat and EMI emissions. By Matthew Schmitt, Dave Dedonato and Patrick Recce of TE Connectivity
Figure 1 shows examples of SFP+ optical transceivers in single and multiple port configurations, including 1x2, 1x4 and 1x6 cages. The higher performance and density of these devices result in challenges related to excess heat and EMI emissions; however, there several ways to address these concerns.
There are many internal and external variables that affect thermal performance of pluggable I/O products and dictate whether or not a cooling solution is required. Unfortunately, there is no clear answer to this question and, therefore, system architects must consider many options and restrictions when designing end products.
The form-factor of the product must be taken into account, as different configurations have unique airflows. A “pizza box” (either a 1U or 4U) that is mounted in a standard rack may have front-to-back airflow or side-to-side airflow. A blade-style switch or piece of equipment may be mounted vertically within an enclosure that is, in turn, mounted in a standard rack, a configuration that almost always has bottom-to-top airflow.
The number and density of ports mounted on the PCB must also be well thought-out. These ports can be single port cages, 1xN ganged cages, 2xN stacked cages, or a combination of all three. This is in addition to other I/O connectors at the face of the product. The spacing between these cages must also be taken into account, as well as port density, ambient air temperature, airflow, allowable temperature rise and backpressure created by baffling.
Finally, heat dissipation of the optical transceiver itself must be considered. Older SFP optical transceivers usually operate at lower wattages but newer SFP+ optical transceivers operate at higher wattages. Commercially available optical transceivers are rated up to 70°C, but there are extended-temperature-range transceivers that can operate up to 85°C. Newer SFP+ modules that are used in short-reach and long-reach applications are still dissipating 1W or less, but extended reach and fixed DWDM (dense wavelength division multiplexing) transceivers can range from 1.25-1.5W per port.
The industry has found that trying to cool these higher-wattage transceivers in SFP+ stacked cages is quite challenging. Managing the temperature of the inner-lower row of ports that are not exposed to airflow is especially difficult.
Figure 2 shows the results of an experiment done by TE Connectivity (TE) that simulated the conditions in a rack-mounted router with 24 data ports comprised of six 1x4 ganged cages. The cages are stacked in three stacked pairs. Airflow is moving across the entire assembly from left to right. Table 1 provides details of the test setup. The red numbers in Figure 1 show individual SFP+ ports that are running at temperatures in excess of the 70ºC operating limit.
Airflow: 500 LFM (measured by airflow chamber and not by anemometer probe)
Backpressure: 0.25 inches of water (controlled by baffle)
Altitude: Sea level
Temperature: Room ambient (results adjusted for 55°C ambient)
Power: 1.5 watt per transceiver
Air gap above cage: 50mm
Air gap below PCB: 9.5mm
Temperature rise limit: 70° C
Table 1. Thermal test setup parameters
In order to improve the thermal characteristics, TE has added ventilation holes strategically in both the cage body and in the latch plate. Figure 3 shows a side view of the cage assembly that includes side-wall perforations. Figure 4 is a close-up view of the latch plate that includes the two thermal vent holes. The perforations in the cage body were optimized in size and shape in order to minimize the potential for EMI leakage. The two thermal vent holes added to the front of each latch plate still allow for the use of two light pipes.
The results of these simple modifications to the test setup demonstrate a remarkable improvement in thermal performance (Figure 5). None of the ports in the revised test exceed the 70ºC limit under identical test conditions.
Many variables affect EMI emissions, such as leakage from optical transceivers, type of board-level components (integrated chips, power supply module, etc.) and other improperly shielded connectors used in today’s communications equipment.If these EMI emissions are not properly contained within a chassis, then these disturbances may degrade or limit the effective performance of the electrical circuit or prevent the end product from passing FCC emissions standards.These effects can range from a decrease in performance to a total loss of data transmission.
TE has identified two components that improve EMI performance. The first component is the gasket retention plate, shown in Figure 6a. This component acts as a backer plate for the conductive elastomeric gasket that interfaces with the inside of the front bezel. Modifying the gasket retention plate to provide more attachment points to the cage body reduces EMI emissions between the cage body and gasket retention plate. The redesigned “right-angle” gasket retention plate can be seen attached to the cage body in Figure 6b.
The second component identified for improvement is the latch plate, which is the component that separates the bottom port from the upper port. This component also houses the light pipes that transmit light from LEDs mounted on the PCB to the front face of the cage assembly. Industry standards dictate the design of the latching mechanism, so this was left unchanged, but within the latch plate a secondary component was added to improve EMI performance and prevent leakage.
The optimization of the gasket retention plate and the addition of the component to the latch plate yielded a significant improvement in EMI performance in the 10-15GHz range (Figure 7).As data rates and packaging densities continue to increase, more attention must be given to containing EMI and addressing thermal considerations.