Editorís note: Dan Harmon, Sensing Business Development Manager, Texas Instruments is our guest blogger for this 100th blog post from Texas Instruments on Signal Chain Basics
Managing system thermal performance is critical in todayís electronic systems if you are to maximize performance and the user experience. As systems grow more powerful, and in many cases smaller in size, managing the thermal profile has become an ever-increasing challenge. Monitoring the current provides a leading indicator to potential thermal issues.
Many implementers today use a single system-level fuse or over-current detector that monitors the main power into the system and makes system-wide decisions or implements system-wide actions when an event occurs. While this can be a low-cost implementation, it prevents the system from optimizing performance. The alternative is to implement distributed over-current detection at the sub-system or module level. This allows for making decisions more efficiently. This distributed monitoring addresses three main customer concerns:
- System utilization and efficiency
- Determinate fault Identification
- Offloading event detection from the system controller
System utilization and efficiency
Here the system integrator wants to maximize the system performance while minimizing the power consumption by monitoring the current at the sub-system level. To help explain this, letís consider a central office scenario.
Example of a server farm featuring multiple blade servers
By monitoring the load current of each individual blade in a server or channel in a in a communication systems, you can understand when usage in that module is increasing or decreasing. During off-peak hours, enabling only a single card minimizes system power consumption. As you monitor the current being consumed by that individual card, you can detect when the load is increasing and enable additional modules to increase the system capabilities while also increasing power consumption. By monitoring the load current at each individual sub-module, you can optimize the user access experience by bringing on the appropriate number of communication cards while minimizing the active power consumption.
Determinate fault identification
Using a system-wide fault identification method shuts down the entire system when any fault is seen as an increase in the total system current. While this could be important for critical system protection or user safety, it does not allow for deterministic module shutdown and debugging. An example of deterministic fault identification versus system level protections would be ground-fault circuit interrupter, or GFCI. In a GFCI system, the individual outlets have fault identification and protection built into them. This allows for only that specific outlet to be disabled rather than the whole circuit on that fuse or circuit breaker. This enables faster and easier fault identification and potentially shortens the debug cycle time as well as the downtime.
Example of how GFCI outlets enable distributed fault identification
Localized over-current detection enables the system management controller to know precisely where the fault is. The controller can decide to shut-down just that one module, a larger portion of the system, or in the worst-case, the entire system. This information also enables faster debug and minimizes downtime as the repairman will know where to look for the problem, versus going through each subsystem one-by-one. Additionally, this can allow portions of a system to continue running normally while only a portion is shutdown awaiting repair.
Showing a repair of one portion of a blade server while the balance of the system remains active
Offloading event detection
Simplified over-current detection devices, such as the Texas Instruments INA300, can be used as simple-to-implement interrupt generators for the system management controller. This frees the system management controller from having to continually poll the individual sub-systems for their current level and make decisions based on those levels. This lowers the overall processing requirements for the system management controller, allowing for a lower cost, lower power implementation.
Distributed over-current detection offers the system implementer improved system utilization and efficiency, faster and more accurate fault identification, and can offload event detection from the system controller. To minimize impact requires that the implementation be very simple to manage, and requires a very small footprint. Miniaturizing and simplifying implementation are key to enabling multiple measurement points without complicating or growing the system.
Please join us next time when we address how jitter in the signal controlling sampling rate reduces SAR-ADC ENOBs.
About the author
Dan Harmon is Sensing Business Development Manager for TIís sensing group. In his 29 year career at TI, he has supported a wide variety of technologies and products including interface products, imaging analog front-ends (AFEs), and charge-coupled device (CCD) sensors. He also has served as TIís USB-IF Representative and TIís USB 3.0 Promoterís Group Chair. Dan earned a BSEE from the University of Dayton and a MSEE from the University of Texas in Arlington.