Kontron has just released their first Cavium OCTEON® II AdvancedMC packet processor module, the AM2411. Kontron’s module features the Cavium Networks Octeon II CPU. It’s an incredibly powerful network processor and Kontron has really taken advantage of it to create an industry thumping feature set. This kind of product will serve telecomm companies very well in creating 4G networks using either LTE or WiMAX. The board designers clearly made the product durable as it has passed testing for Telcordia GR-1089, GR-63, IEC 60950-1, EN/CSA/ UL 60950-1. And this quote from their datasheet really tells it all regarding the 4211’s design point, quality: “This board is designed to meet NEBS Level 3, Earthquake Zone 4”
Board quality and specifications are not going to be a question with this product!
What is going to be a consideration is what this product is like in a chassis full of other boards operating at peak performance. There is no question Kontron has done a great job designing their board, but, their board now becomes a component in an integrated system. Such an integration of Commercial Off the Shelf Components (COTS)is almost always a challenge across a number of technical areas.
Trenton Technologies wrote a bit on this in their blog back in July of 2010 with their post, “Can You Really Simplify Integration of Industrial Computer Systems?” In that post they noted that 6 key factors to decided if a COTS solution could be used in an application, those six points are:
- System location requirements
- Projected service life of the system
- Number of option cards needed in the computer
- System expansion requirements
- System power and fail-over requirements
- System O/S and application software solution requirements
But the one piece they missed was the overall system thermal load. So let’s talk about that a bit. First, the key target temperature to keep in mind is the junction temperature. The junction temperature is the temperature of the silicon die within the package of the device when the device is powered. The junction temperature can also be referred to as the operating temperature. The goal of thermal management is to manage the heat to the junction temperature.
A successful thermal design requires that the junction temperatures of all critical devices in a system be sufficiently below their critical level in the worst-case ambient temperature. Otherwise, higher temperatures may result in data transmission bit errors and/or a decrease in a system’s life expectancy. Thermal coupling and non-uniform PCB layout make this determination a complicated process.
The list of thermal parameters that a complete system has to manage to help reduce the junction temperature on a device are pretty long:
- Thermal coupling within the system and surrounding equipment
- System standardization (e.g., the ATCA Standard)
- Constrained space
- EMI/EMC requirements for high frequency devices, boards and packaging
- Acoustic noise/li>
- Limited air flow
- Rigid performance standards (e.g., 72 hours operation at 55oC)
- Non-uniform power distribution and congested PCBs
- Service Level Agreement(e.g., system operational within 4 hours of failure)
The following table shows the approach, domain, anticipated results, and the needed tools to successfully obtain the junction temperature of a device in the system.
| Solution Order | Package Level | Expected Results | Requirements and Tools |
| First | System | System flow distribution and boundary conditions for the card racks, T, V, and Pressure | Measurement or |
| Second’ | Card Rack (chassis) | Boundary conditions for the PCB, and thermal coupling between the board and the rack | Measurement or Simulation (CFD) and solid modeling (temperature gradient in the solid) |
| Third | Board (PCB) | Boundary conditions for the component on heat transfer and fluid flow | Measurement of fluid flow distribution and board properties and board level solid modeling |
| Fourth | Component (device or module) | Boundary conditions for the Die (Tj) and its cooling solution | Fluid flow and temperature measurement equipment; Heat sink; Interface materials and their properties;Solid modeling;Package material properties |
Once the junction temperature is obtained by using two independent methods, it must be ensured that there is at least a 10% margin of safety in the design. Therefore, the following equation must be satisfied.
Where
Tj, calculated = junction temperature as the result of above calculation.
Tj, spec = critical junction temperature specified by the manufacturer.
Ta, reference = reference ambient or approach air temperature.
With a properly determined junction temperature in hand, a cooling method can be selected for maintaining the desired performance. Depending on the data obtained from the table, we can choose from these options:
- Natural convection
- Forced convection (air-mover)
- Active air cooling (not very common)
- Jet impingement (air or liquid)
- Advanced systems (liquid or refrigeration)
Although cooling by air is the most common and preferred system, liquid cooling has been used in unique circumstances. Regardless of the method selected, every cooling decision should consider the following parameters:
- Cooling capacity – does it satisfy the junction temperature requirements?
- Size – does it comply with the packaging requirements?
- Regulatory requirements – does it meet the system and site implementation requirements (e.g., NEBS) ?
- Reliability – does it meet the expected life requirements?
- Budget – does it comply with the cost constraints imposed on the system?
- Market availability – is it readily available? Can supply-chain requirements be met?
Obtaining a successful cooling solution is made possible by combining methodical thermal analysis (as delineated in our table) with the results from at least two independent approaches, and taking into account the above system parameters.