Robotics and Automation Expert
page 2 page 3 page 4 page 5 page 6

Non Volatile SRAM Improves Robotic System Safety (page 1)

Abstract ó Complicated multiple axis robot controls generate a lot of data and contain states that must be restored quickly in the case of partial or complete power failure. While technologies like flash and battery-backed SRAM have been used, a solution based on non-volatile SRAM (nvSRAM) offers significant advantages for the safe startup of robotic systems.

As the number of modern manufacturing facilities increases worldwide, the use of robotic machinery is now more the rule than the exception. Typical applications include mechanical assembly, material handling, dispensing, routing, electrical assembly, packaging, test, and inspection. Six-axis robots, conveyor control, and other feed mechanisms require substantial coordination. Multiple robots may also work together on the same project simultaneously. As the number of robots increase, the need for a technology that allows for a safe restart of these robots after a sudden shutdown.

With each passing day, factory automation requirements are demanding a greater number of axes on robotic equipment, as well as tighter tolerances on their movement and performance at the point of application. This coordinated subsystem interaction demands intelligent fault management and continuous improvement of distributed control programs. It becomes essential for robotic system machine safety  that data is captured in fault situations, including power spikes, brownouts, high noise, and power grid failures. This article examines different semiconductor memory solutions for data storage and retrieval after a power failure event has ended and system power has been restored.

Combine the constant flow of closed-loop positioning information fed back from the robotic sensors with the data logging of detailed arm movement history for quality control purposes and the result is ever increasing sizes of constantly written cache SRAM. Other memory technologies such as EEPROM and flash physically wear out with repeated use and exhibit extremely slow write speeds. Keeping position and velocity information properly recorded at a blistering rate of five million updates per second is essential for high-quality contour profiling. Flash memory write delays are 2,500 slower than SRAM limiting the closed-loop control rate to 2 KHz. Position, velocity, and acceleration can all change continuously during motion profiles.

The high update rate of Proportional, Integral, Derivative (PID) motion control makes real-time tracking of all positions next to impossible for the upstream control. Eliminating nonvolatile updating of the robotís state in real-time introduces the risk of errors during system operation restart. It is crucial for a robot to properly record its last known state under the most adverse conditions, or else the overall systemís operational costs could increase as a failure of some kind would inevitably occur.
Using low-power SRAMs and a large battery source for backup, some manufacturers have attempted to address system recovery after a power failure by making all of the semiconductor memory space non-volatile. Because SRAM is of the lower power variety, it is also slower and more expensive when used in large banks.

Research in the area of persistent storage systems, known as Solid-State Drives (SSDs), reveals that data corruption is a significant risk. First, electrical or electromagnetic noise can still upset memory contents because of the slow response of the battery switchover circuits. System faults frequently increase in industrial environments where large motors, electrical contactors, and electromechanical valves all create radiated and conducted emissions. Second, machine software programs can have bugs, bad pointers, stack overflows, or arrays addressed out of bounds that can cause unintended write operations to what is supposed to be secure data. Data stored on a flash drive or a hard disk drive is not part of the computerís address space. Therefore, other techniques in battery-backed main memories must be employed to truly secure valuable and fast-changing parameters from power loss or other risks.

© copyright SafeMachines, PLLC all rights reserved