I recently finished a board with a sensitive optical detector. The front-end is a trasconductance amplifier detecting currents in the hundreds of femtoamps. The optical detector uses a Peltier cooler that requires around 0.5V at 0.5A. 5V with plenty of current available comes into this board from elsewhere in the system. I could simply use a power resistor to drop 4.5V. This would dissipate 2.25W, which is a small fraction of system power and would not present a thermal issue. A switching power supply, even an inefficient one, would waste much less power. Because the front-end relies on tiny currents flowing into large impedances, though, I worried about inductive coupling.
This problem made me think of a coverage I read four years ago of Al Gore’s speech at the Embedded System Conference (ESC) in which he called on electronics engineers to do their part in reducing anthropogenic climate change. The small decisions that engineers make in cases like this add up. 2W saved on a device that is on most the time works out to a kWh over a year, which results in a little over a pound of CO2 emitted into the environment. When you multiply that by the number of systems deployed, which we always hope is as high as the EAUs we provide to vendors when requesting samples, it works out to a small but non-trivial reduction.
So I went forward with the switching supply. To reduce noise coupling into the amplifier and detector, I put the switcher on the opposite end of the board. I also made a simple linear regulator with an op-amp and transistor to filter the switching noise. This might not filter high-frequency noise, but high-frequency noise is less important because of the bandwidth limitations of detector amplifier.
Critics of the CO2 motivation for reducing power consumption will rightly point out that while we know CO2 emissions play a significant role in climate change, we are not sure to what extent. We cannot be sure that reducing CO2 output is the most efficient way to control anthropogenic climate change. Setting aside the CO2 issue, there is the question of energy scarcity. Any effort in reducing energy usage will ease the transition from oil to other sources of energy.
In addition to high-minded concerns, we never know when a product will be used in an environment where efficiency matters. I worked on a communications product for industrial communications that I thought would always be used in factory environments where its power consumption would be insignificant compared to the machinery it controlled. It found application, however, in remote pipeline locations where users needed to run it on solar power.
In line-powered systems many projects start with claims that power budget / real estate / BOM cost is not an issue only to find by the end of the project that one or all of these has become a major concern. It is nice to have a little headroom beyond initial product requirements.
Many engineers have a natural predilection for efficiency. We have a hard time disregarding efficiency even when it is not important to users. We should embrace this tendency when we can.