> This can be implemented basically with a fixed value of resistance, but because of the way resistance interacts with voltage, it’s actually very gentle on the battery. As the battery voltage falls, the current falls accordingly, so the delivered power falls even further. For modern appliances, this type of loading is not common.
Dumb question - is it at all possible to design a modern device which operates this way? (As battery voltage falls, device performance also falls? It's been a really really long time since my EE classes - are transistors just unable to operate this way?)
Apple did this with iPhones, where the CPU throttles itself when the battery can’t keep up with the peak demand anymore. They got sued for it (mostly because they didn’t tell anyone and people bought new phones when the old ones started mysteriously slowing down).
Have you ever used an old flashlight in the 80s or 90s?
As AA charge fell, the light got dimmer and dimmer. Today, we have devices that do the exact _OPPOSITE_, pulling the last bits of electricity out of the cells through boost-converters or whatnot (boost converters existed back then, but weren't as efficient or cheap as today).
Consumers demanded consistent and reliable performance no matter if at full-battery charge or nearly empty. People preferred their devices to suddenly "shut off".
This behavior is still common in flashlights. Flashlights using three alkaline (or NiMH) batteries in series, or a single Li-ion cell can drive a white LED via a linear regulator (or occasionally just a transistor), and it will dim as the battery falls below the forward voltage of the LED at its maximum output. At higher price points, a single Li-ion cell and a regulated buck converter is common to see and much more efficient, but maximum brightness is still usually limited by battery voltage.
A flashlight using a single AA or AAA battery must use a boost converter because all white LEDs require about 3 volts. Even these often don't produce stable output as the battery drains, which is sometimes intentional because that behavior would produce terrible battery life with alkalines due to their high internal resistance. It's fine with NiMH.
Even Li-ion lights with boost converters don't always manage full output on a low battery because it's common to find overdriven components on a 20mm driver board (it needs to fit in a pocket) that's trying to push as much power as possible (lumens sell lights). Inability of the electronics to maintain full output isn't necessarily a significant limitation in the real world anyway; a 25x100mm aluminum tube pushing 40W gets hot fast, and there's almost always some sort of thermal-throttling mechanism. That said, full output on a low battery usually earns praise from reviewers.
> It's been a really really long time since my EE classes - are transistors just unable to operate this way?)
Yeah, most CPUs nowadays have dynamic frequency adjustment to maximize battery life. Lower frequencies also mean transistors can operate at lower voltages, so by reducing the operating frequency, you can reduce the voltage and therefore power draw on your battery.
Dumb question - is it at all possible to design a modern device which operates this way? (As battery voltage falls, device performance also falls? It's been a really really long time since my EE classes - are transistors just unable to operate this way?)