There is nothing like a free lunch. The manufacturing cost of smaller features on the chips is high, so while you save silicon area, you spend more on the tools and processes. Smaller features also mean not only faster transistors, but also lower breakdown voltages, lower latchup (-> ESD) immunity, higher static leakage, more temperature-, stress- impurity- and other environment-induced problems. Smaller feature FLASH means lower retention and endurance, higher crosstalk; not to say that some of the FLASH area is spared by making it programmable in huge pages (show me an ARM with a bytewise-programmable EEPROM). Using modern, yet-to-mature technologies also means introducing more unknowns as far as long term reliability is concerned - oh yes, that mobile phones and other BTT (bored-teens-toys(TM)) are not supposed to last longer than the legal 2 years warranty, are they?.

There is no free lunch. With a big step forward in one direction you get several small steps back in other directions. At the end of the day there will be some progress made, but less than the glossy manager stuff would like to make you believe. Keep your skeptical engineer hat on.


You are not really supposed to run the USB peripheral directly out of the USB power supply. At least the data lines shall be driven from 3.3V (so you can't really omit the voltage regulator), and also note that the "USB 5V" may be out of the common tolerance assumed of 5V power supplies.

Nevertheless, there is a slew of applications making use of wide power voltage, and also those making use of the higher voltage swing, for whatever reason. The manufacturers WANT to ignore this to save silicon and so produce more cheaply (assuming that will give them more points to sell). However, the developers are waking up gradually to the fact that they can't fool the physics and its consequences.

JW