Managing the successful adoption of new low-voltage memory ICs

August 08, 2017 // By Conrado Canio, William Chen and K. C. Shekar

Today, the circuitry on the board in mainstream industrial and consumer products operates from a wide range of supply voltages: the power rails are most commonly at 5V, 3V, 2.5V, 1.8V and various lower voltages. To ensure compatibility between devices from different manufacturers, and to avoid unnecessarily complicating board-level power system design, merchant semiconductor manufacturers typically design their standard products to run from one or more of these standard power rails. But there is a strong force resisting this general preference for stability and compatibility: mobility.

In mobile products, the size and weight of the battery is one of the most difficult and challenging design constraints: a smaller battery allows the designer to shrink the end product design, or to use the saved space to add valuable features. But at the same time, users pay close attention to battery run-time, and show a strong preference for products that require less frequent charging.

Every milliwatt saved from the power budget is important to product designers. And for them, the industry’s use of power rails at various standard voltages, often at 1.8V or higher, is a problem, not an advantage: that is because many components – particularly those operating in the digital domain – would with some modification be quite capable of operating from a power rail at a voltage lower than 1.8V, resulting in valuable savings in active and stand-by power consumption.

Clear direction of travel

Today, system designers typically have to provide multiple power rails in order to accommodate components operating from different supply voltages. Analogue devices such as sensors commonly have a 3V or – in industrial applications – even a 5V supply. Legacy digital components might have a 3.3V, 2.5V or 1.8V supply. At the low end of the voltage range, the latest applications processors or systems-on-chip built on advanced process nodes, such as 28nm or smaller, might have a core operating voltage as low as 1.0V.

Yet in power system design, as in any field of design engineering, complexity is inherently undesirable, and in general, the fewer the power rails the better. This means that the industry will eventually want to settle on standard voltages at a level below 1.8V. What will these voltages be?  

Figure 1 shows how DRAM technologies have led the memory IC industry beyond 1.8V. Standard DDR2 DRAM was the last to use a 1.8V supply. After that, successive generations of DDR DRAM operated at 1.5V (DDR3), then 1.37V (DDR3L) before reaching today’s level, 1.2V (DDR4).


Fig. 1: range of operating voltages supported by common memory device types. In yellow, DDR DRAM; in green, Winbond NOR Flash (Image credit: Winbond)

Fig. 1: range of operating voltages supported by common memory device types. In yellow, DDR DRAM; in green, Winbond NOR Flash (Image credit: Winbond)

So for DRAM, there are three intermediate voltages below 1.8V. Figure 1 also shows in green the supply requirements of successive families of NOR Flash ICs from Winbond, operating at the standard 3V, 2.5V and 1.8V levels. Now the latest NOR Flash families offer two voltage ranges: one at 1.2V,

Pressure is building on semiconductor manufacturers, then, to respond to their customers’ demand to support operation at lower voltages. So how is the electronics component industry likely to accomplish any move to a lower voltage standard? And how should system designers be preparing now to take advantage of a new generation of lower-voltage component offerings?
Company: 
Low voltage memory