Energy-efficient memory for lasting battery-life

As issues of the environment move to the forefront of our collective consciousness, the electronics industry is starting to focus on low energy consumption in order to market its products. Developing trends include lowering products’ power consumption to make them more environmentally friendly, and operation with only one battery throughout their entire lifetime. Going a step further, these ultra-low-power capabilities are even creating a new breed of products that do not require a battery at all, and derive energy from the environment.

The key to many of these developments is the products’ ability to sleep at very low power. To achieve this, the device must essentially shut down, since conventional sleep modes consume too much power. Shutting down, however, only works if the product can remember its state when it reboots. To achieve this, designers need a non-volatile memory with low operating currents and high write endurance.

 

 

Comparing power consumption of non-volatile technologies

To find the best solution to these requirements, the power consumption of various non-volatile serial memory technologies should be considered. The technologies compared here include F-RAM, EEPROM and Flash, all of which are common memory architectures used to store configurations. Since Flash is only available with a Serial Peripheral Interface (SPI), it will be compared with SPI versions of EEPROM and F-RAM to create a fair test.

For the sake of this study, the amount of energy used by each memory technology will be calculated. Energy is a good way to make a comparison as it takes into account the duration of the task as well as the amount of power required to perform the task.

Energy (Joules) = Power (Watts) X Time (Seconds)

Substituting Power = Volts X Amps gives:

Energy = Volts X Amps X Time

F-RAM, EEPROM and Flash vary most during writing and erasing, therefore these are the key elements that will be considered in this energy-efficiency comparison.

Consumption during the reading process is roughly the same for each technology even if operating speed, voltage, and current consumption differ. One might assume that the speed of the serial interface would play an important role here, however, when the calculations are repeated for the same part at different SPI bus speeds, the total energy spent remains roughly the same. To eliminate any issue associated with calculating SPI bus overheads (i.e. issuing commands and setting-up the address), the comparison will be performed using a significant amount of data, specifically 64Kb. While F-RAM and EEPROM have no erase time, Flash requires a significant amount of time to erase sectors. As such, the test will compare how long it takes for 64Kb to be erased and written with new data.

Finally, it should be noted that manufacturers do not always use consistent standards to quote power consumption. Some devices may be specified to operate at Vcc of 1.8V, but their datasheets offer operating current figures at Vcc of 2.5V. The figures used here are the worst-case figures specified in the datasheet.

 

 

Conclusion

Table 1 shows that serial data Flash uses significant amounts of energy when erasing pages/sectors due to the considerable amount of time needed for these operations. The energy requirements of both EEPROM and Flash could be reduced by using polling techniques rather than waiting for the worst-case time to complete the operation.

The results clearly show that F-RAM has a significant advantage over the other non-volatile memory technologies when it comes to providing a low-power memory that enhances energy efficiency and enables products that require only one battery throughout their entire lifetime.

 

Ramtron F-RAM Design Note