Wires. We don't think of them a lot of the time, except perhaps when we are dealing with high-speed communications, but wires are an essential part of every system created. They get power to a device; they are used to connect sensors to processors and processors to displays. In a typical car today, wires represent the third heaviest and costliest component after the engine and frame. The average weight of wires in a car is now 150 kilograms, according to the IEEE Institute.
Much of this wiring has come about because of the slow but steady increase in electronics in the car. As each system was added, new wiring was added to the harness. Also, the need to have centralized fuses for wire harness protection makes wires longer than they need to be. In planes the situation is even worse. A 747 transport aircraft has 140 miles of wire in it, weighing 3,500 pounds. A plane fitted for passengers would take even more.
So, why am I talking about this? There are many systems where the wiring is a constraint, and wires are the big limiter for the progression of the Internet of Things (IoT). The IoT is dependent on having very small, cheap sensors that can basically be placed everywhere and can communicate to whoever needs their data. Given that these networks are ad-hoc and transitory, they cannot be wired networks.
There are many standards looking into ultra-low-power wireless communications, some that consume as little as 200μJ per hour. But what about the power connections? When these sensors get distributed into every light switch, into your clothes, into signs along the road — nobody wants to deal with changing the batteries or wiring up to a power source. They have to scavenge the power they need from their environment.
Solar, vibration, heat, and RF are the primary ways power can be derived from the environment, but the amounts they can reliably generate have to be matched to the power draw of the electronics. Not only that, but the power they provide is erratic, where both the voltage and amount of power available can vary over time. Small batteries or super capacitors can be used to hold charge, but this is already adding to the size, weight, and cost of the solution.
All of this places a large burden on the analog parts of a system, but at the same time, the analog components are vital to the success of such a system. A large amount of pioneering work has been done in the medical field where long battery life has always been a priority in the design; but more is required, and often this requires better models in order to design better circuits. Most analog design has been using BSIM3 and BSIM4 models that are based on threshold voltages.
A new BSIM6 model is a charge-based core derived from Poisson's solution and includes physical effects such as mobility degradation, velocity saturation, high frequency models, etc. The key advantage is that currents and derivatives are symmetric at Vds =0 and capacitances and derivatives are also symmetric around Vds =0. In addition the solution is continuous in all regions of operation. These models are necessary for FinFET modeling, which provides increased gain and reduced leakage compared to planar transistors. BSIM6 was released in May 2013 and is already incorporated into many commercial EDA tools.
How are the demands of ultra-low-power going to affect you? Will more accurate modeling help?