AdvancedMC Insider - November 2006

AdvancedMC™ Cooling Considerations

With such a wide range of potential application targets, AdvancedTCA® and AMC designers must anticipate the system level thermal environments in which their designs will ultimately operate. These diverse environments, combined with the wide range of module configurations, yield a multitude of potential thermal permutations. Therefore, it is important to establish a few reliable ground rules that will serve the designer well in most, if not all, new design situations.

AMC board requirements
AMC.0 defines a modular add-on or 'child' card that extends the functionality of a Carrier Board. Often referred to as mezzanines, these cards are called 'AMC modules' or 'modules' and they lie parallel to the Carrier Board. Carrier Boards may range from passive boards with minimal intelligence to high performance single board computers.

AMC modules are also integrated into MicroTCA (MTCA or µTCA) systems. That is they are inserted into a shelf, enclosure or subrack just like plug-in cards in modular computers. In this case, they are then referred to as ‘cards’ or ‘boards’ rather than modules.

Mechanical (AMC.0 R2.0 specification)
The latest revision to AMC.0 changes the naming conventions and adds a new front panel size. However, AdvancedMC modules are still available in Single and Double sizes (much like cPCI or VME boards which come in 3U and 6U sizes). Both sizes are exactly the same length from the front panel edge to the rear card-edge connector--181.5 mm. However, the width of the front panel is variable, which is a departure. Actual sizes are as follows:

Single	73.5 mm	also called Single-Width (SW)
Double	148.5 mm	also called Double-Width (DW)
Compact	13.88 mm	also called Half-Height (HH)
Mid-size	18.96 mm	new size added by R2.0
Full-size	28.95 mm	also called Full-Height (FH)

These size variations have an obvious influence on thermal conditions and requirements. Larger boards may dissipate more heat. Compact-, Mid- and Full-Size modules have different component height profiles and therefore develop different back pressures to airflow. There are defined keep-out and different component-height zones on the board surfaces. Any mechanical variations, including individual component heights, also influence airflow, back pressure and turbulence.

PICMG® Definitions
PICMG has specified thermal design rules and recommendations for AMC boards. It is still unusual in this industry that a standards body does this. However, these are just general guidelines or fringe values. Individual simulations and practical tests are needed to verify safe operation within defined thermal conditions.

PICMG specifications require that product AMC documentation shall include a table or graph for an airflow between (10 to 40) CFM (0.28 to 1.12) m³/min. and a drawing that identifies principal carrier airflow paths for each carrier/mezzanine or card/shelf combination. The system integrator shall ensure that the airflow through the carrier and all installed AMCs ensures the required operational temperatures of the installed components in the defined environment. In the MTCA specification an airflow graph showing a range between 10 CFM (0.28 m³/min.) and 50 CFM (1.4 m³/min.) is required.

In a running system air temperature rise from inlet to exhaust in a module/carrier combination shall not exceed 15 °C for the complete assembly. It should not exceed 10 °C. On an individual module it should not exceed 2.5 °C. The component which is of most thermal concern shall be monitored by a sensor. Modules shall provide at least two geographically dispersed airflow sensors on each Module. Carriers shall provide a sensor monitoring the inlet air temperature. There are still more requirements stated in the AMC, ATCA and MTCA specifications.

At any time during operation, the heat energy produced by the installed components has to be transferred to the air moved through the system. The thermal energy dissipated in a unit of time has to be removed within the same unit of time by a certain volume of air which is warmed up by the difference between the ingress air temperature and the egress temperature. The airflow is impeded by the resistance of the components, modules and boards. This resistance (back pressure) corresponds to a pressure drop, which is a function of the speed of the air movement.

There is no general purpose specification for module deheating or cooling in the AMC standard. Analysis of airflow through module and carrier boards is very complex and airflow vectors are three-dimensional. An AMC module of any size can be placed into any of the positions on a carrier or in a card cage (enclosure, subrack). The air entering this system may be laminar, turbulent or in transition between these states. There might also be eddies (circular flow like in a whirlpool) from modules which are located before the module in question in airflow direction.

Modeling and simulating this is computationally demanding. Component placement is critical. Some components which may not need cooling (capacitors) will nevertheless impede airflow and mask components located downstream. Large heatsinks impede and mask airflow and additionally heat the passing air. If the cooling fins of a heatsink are spaced too close then the air will flow around the heatsink rather than through it. In that case an air tunnel might force the air to pass through between the fins rather than around the heatsink.

With these complex variables in mind, the table below shows AMC heat dissipation limits (approximate values) if airflow is uniform and hot components are placed strategically.

Table: Approximate power consumption for proper forced-air cooling

Width	Height	Watt
Single	Compact, Half	20
Single	Mid-size	30
Single	Full-size	40
Double	Compact, Half	40
Double	Mid-size	60
Double	Full-size	80

 

- back to top -

AMC.0 R2.0

An interview with Jeff Marden

Jeff, why did PICMG® feel that it was necessary to make changes to the AMC.0 spec?

PICMG did not actually make the decision. Companies that were involved in AMC product and solution activities submitted two Engineering Change Requests (ECRs) to PICMG in order to modify AMC.0. PICMG member companies reviewed these ECRs and voted to accept the ECRs and create a new Sub-Committee to address them. The two ECRs include the following:

• AMC.0 ECR001 – An Engineering Change Request to add additional, new AMC connector types and vendors to the AMC.0 specification.
• AMC.0 ECR002 – A more general Engineering Change Request to review and update the AMC.0 specification, add features and correct errors in the specification.

What’s the official name for this new revision now that it has been ratified?

The output of the AMC.0 ECN Sub-Committee work will be AMC.0 Revision 2.0, reflecting both significant change and closure to the ECR/ECN process for the specification.

There seems to be a new nomenclature in addition to a new card size. Can you describe the new names—Compact-, Mid- and Full-Sized cards, as well as the height/width vs. single/double naming conventions?

The new naming convention was created as a result of a new AMC form-factor and a desire to better define the AMC and ATCA Carrier height and width characteristics in the AMC.0 specification. The following chart identifies this new nomenclature:

New	Old
Compact Module	Half-Height Module – 3HP
Mid-Size Module	New Form-Factor – 4HP
Full-Size Module	Full-Height and Extended Full-Height Module – 6HP
Single	Single-Wide Module
Double	Double-Wide Module
Conventional Carrier	same
Cutaway Carrier	same

Did the intense level of interest in MicroTCA™ contribute to the need for a revision?

Interestingly, no not really. In fact, the MicroTCA.0 Sub-Committee adopted the charter to not require changes to the AMC Module as part of MicroTCA concept and specification development.

Doesn’t the addition of a new card size just serve to add more complexity and confusion?

While an increase in the complexity of the AMC.0 specification is a by-product of any ECN work, one major goal of the Sub-Committee was to review and clarify text in the AMC.0 specification, reducing complexity. One example of this was the re-write of the mechanical section of the specification. This re-write yielded common, unified details for the mechanical aspects of the AdvancedMC Module and should result in a better understanding of the Module concept as folks review the specification.

Were the existing AdvancedMC™ cards sizes really that much of a problem, or was it a case giving people more options?

The goal of the Mid-Size AdvancedMC Module’s creation was to provide for a Module with sufficient side-1 and side-2 component height to address most mezzanine application while solving ATCA carrier panel space constraints when Full-Height AdvancedMC Modules were installed in all bays of the Carrier. Use of a Mid-Size Module allows space under the AMC bay for ATCA panel area and facilitates labeling and LED requirements on the ATCA/AMC Carrier. So no, the existing modules sizes were not the problem, and in-fact all AMC.0 R1.0 form-factors have been retained in AMC.0 R2.0

It appears that the Mid-Size card has the same wattage as the Full-Size card. However, since more Mid-Size cards can fit into a MicroTCA chassis, won’t this cause cooling issues?

Cooling issues are always a concern, and some additional attention to the AdvancedMC Module spacing and required air-flow for a MicroTCA systems with Mid-Size Modules installed is needed. However, as long as proper consideration of chassis form-factor, Cooling Unit specifications and thermal load is applied, this should not represent any additional thermal issues in my opinion.

Some people are saying that the single Mid-Size AdvancedMC™ will become the most dominant form factor. Do you agree?

The Mid-Size Module represents a good compromise between the Compact and Full-Size Module form-factors, with component height, power and panel space usage in-between the two. If it is true that ATCA/AMC carrier vendors like GE Fanuc Embedded Systems will engineer their products to support the Mid-Size Module form-factor, and dual use of AdvancedMC Modules between ATCA and MicroTCA is likely to be the goal of Module developers, then the Mid-Size form-factor should become quite popular. One caveat though. Users of AdvancedMC Modules often have connector density or “cost per channel” ideas in-mind when they consider requirements for their systems. Since the Mid-Size module has most of the front panel size constraints of Compact Modules, connector/channel counts beyond four ports will only be achieved with a Full-Size panel. For Example, our Telum 628 AdvancedMC Octal T1/E1 Module has 8 x RJ48 connectors and requires a Full-Size panel to host these eight connectors.

Assuming that the Mid-Size card will find traction in the marketplace, is GE Fanuc Embedded Systems making Mid-Size cards at this time?

GE Fanuc Embedded Systems is engaged in a two phase program to support AMC.0 R2.0 in general and the Mid-Size AdvancedMC form-factor in specific:

• Phase 1 calls for the Engineering of Mid-Size Front Panels for the existing family of Telum AdvancedMC Modules. Since a 1st generation Mid-Size panel can be mounted on our unmodified Full-Height AMC.0 R1.0 Modules, this strategy allows GE Fanuc Embedded Systems to provide early Mid-Size form-factor AdvancedMC Modules for customer use.
• Phase 2 implements full compliance to AMC.0 R2.0 for members of the Telum AdvancedMC Module family, including LED position changes, Mid-Size form-factor and other required changes for AMC.0 R2.0 compliance.

AdvancedMC™ cards were originally designed as mezzanines for AdvancedTCA® boards. How does this affect module fit on an AdvancedTCA carrier?

Along with the AdvancedMC Module, AMC.0 R2.0 defines updates to the ATCA Carrier as well, not the least of which is panel mechanical changes to support Mid-Size Modules. Details of mechanical spacing, placement and dimensioning are updated, along with electrical, thermal, management and regulatory aspects of ATCA/AMC carrier design.

Apart from the card sizes, what were some of the other changes incorporated into AMC.0 R2.0?

All sections of the AMC.0 specification were reviewed and updated as a result of the AMC.0 R2.0 Committee work. These updates included the following:

• Definition of alternate AMC connector options for Module carrier products. This was handled as part of AMC.0 R1.0 ECN001, and now allows a variety of new connector implementations to be used in Carrier design.
• AMC.0 R1.0 ECN002 covered the following:
  • The addition of the Mid-Size form-factor, and changes to definition and naming of the other Module and Carrier form-factors (as discussed earlier).
  • Maximum Payload Power for AdvancedMC Modules was increased to 80W.
  • Maximum Management Power for AdvancedMC Modules was increased to 150mA.
  • New AdvancedMC panel LED positions and definitions.
  • Additional Telecom clocks were added. AMC.0 Port 16 was reserved for two of the new telecom clocks. A separate Fabric Clock was defined as well.
  • Mechanical dimensioning was updated.
  • Telecom Clock E-Keying was added to the management section of specification.
  • Some alternative Latch/Handle designs were defined and included in appendix to the specification.
  • Changes & updates to all sections of the specification during review in the ECN002 process occurred.

There has been some talk of another AdvancedMC™ revision which would address the needs of more rugged applications. Any news on that front?

At this time, no additional revisions of the AMC.0 specification are planned. PICMG’s work on update and change to the AMC.0 specification, resulting in the AMC.0 R2.0 document, covered as many open issues as could be defined, and none were left for another revision cycle. Stability of the AMC.0 specification was a goal of PICMG during the exercise.

That said, issues of ruggedization of systems that use AdvancedMC modules has been discussed informally as military, aerospace and communications solution providers consider the value of AdvancedMC Modules and systems. In design of these ruggedized systems, use of AdvancedMC Modules unchanged (meaning design of the system to accommodate the standard AdvancedMC Module) is the initial goal. Further work is needed to establish the feasibility of this goal. If it is concluded that modifications to the AdvancedMC Module are needed to meet the temperature, shock, vibration or other operating requirements of a ruggedized application, it is likely that a subsidiary specification (AMC.5?) would be defined and developed to meet this need.

You have an insider’s view of the specification process. Do you see any other new developments on the horizon that you can talk about?

With all of the features of the AdvancedMC Module, its use in next generation applications of all types seems likely.  Here are a few areas that may be of interest:

• Customer-specific AdvancedMC carrier technology to allow the use of these Modules in proprietary systems (like the IBM®  Blade Center™ family of systems, enabled by our BCT4-AMC carrier product).
• Definition of new AMC Bus interface technology, perhaps not even publicly defined yet.
• Solutions to the ruggedization issues discussed earlier.
• Expansion of the target application for AdvancedMC™ beyond the Communications market. Military, Commercial and Medical applications come to mind, especially if you factor in the MicroTCA system for AdvancedMC Modules as the host system.
• There are probably more ideas to consider, and I challenge the audience to think about this and let us know.

- back to top -

PICMG in China

by Howard Glassman

GE Fanuc Embedded Systems was pleased to be invited to present our AdvancedMC™ and MicroTCA™ products and technical directions at the recently held PICMG China 5th Annual Tech-Forum in Shanghai, China on October 25 & 26th. The main goal of the forum is to advance ATCA/AMC/mTCA and to improve international communication and cooperation.

Twelve sponsoring vendors participated, offering detailed presentations and product displays as well as seminars. I delivered technical product presentations including; Building Modular Communications Blades using AdvancedMCs and a MicroTCA systems and solutions overview to the audience of 200+ interested attendees.

China’s Broadband Telecommunications market is projected to more than double over the next few years with total broadband subscribers expected to grow from 50 million users today to 113 million over the next three years. Leading Chinese Service Providers will be offering IMS-ready 3G Wireless and IP/TV “Triple Play” platforms as a part of their NGN solutions. This will present a significant market opportunity for Open COTS Technologies like AdvancedTCA, AdvancedMC and MicroTCA.

PICMG China, in cooperation with the Government-sponsored Committee of Science and Technology of Shanghai, successfully launched the Shanghai Open Communications Technology Platform (OCTP) project in December 2005. OCTP is the first professional and technical service organization chartered with providing ATCA and MicroTCA public information in China.

B-Star (Shanghai's Engineering Research for Broadband Technologies and Applications) was the event co-organizer and has already developed an ATCA/IPMC carrier management module and provides ATCA interoperability testing services at their Shanghai facility. During the PICMG event, they demonstrated their new ATCA Node processor blade which has up to 4 PMC sites for third-party modules and can be used for a wide variety of control and packet processing applications including: WAN access, 3G Wireless, IPTV and NGN.

Liu Hui, the Deputy General Secretary of PICMG China, candidly mentioned that ATCA was gaining considerable traction with China Telecom Equipment Manufacturers and that MicroTCA’s high-performance capabilities, compact size and attractive pricing should be quickly adopted by these vendors for many different Telecom applications. Based on the level of interest at the PICMG event, China is a newly emerging market which has great potential for open, modular telecom platforms like ATCA/AMC/mTCA.

 

- back to top -