Gah, someone should have called me to this thread
Perhaps an iPhone never ‘runs on the charger’.
The power management system is centered on the battery. The battery is the most trustworthy power source. It cannot be suddenly unplugged or decide (without warning) to reduce power delivery because another device was plugged into the next USB port on the charger.
Power from the USB charger would be used opportunistically to charge the battery and then supplement the current demand of the iPhone electronics. But that demand jumps up and down rapidly as screen comes on and cell radio powers up, plus the CPU and GPU demands, amplifier for the speaker, GPS, chip, etc.
The charger, even back to phones I worked on, could never supply enough power for current peaks to the CPU, GPU, DDR, NAND, camera (AF voice coil), flash, etc. Phone chargers - especially the ones people use (random things from china) are widely variable, and generally you're not getting more than 15W from them best case. We had to do all sorts of dynamic tricks to deal with them browning out so people would still get a reliable charge.
Come the iPhone 4S we had to up the battery protection trip point, because we could draw 4A out of the battery (at nominally 3.7v) - 15W - in a peak power scenario. Today's phones are likely 2x that. With ~150mohm battery impedance when new, that's 0.6v headroom gone right there.
Right from the iPhone 4S, Apple moved to multiphase bucks - like you see in PCs - in order to supply the current needs of the CPU, because a single phase would have meant the inductors were ridiculously sized. Instead, they ran either two or three phase allowing use of multiple smaller inductors. This was a first in the phone industry, I remember looking at Qualcomm teardowns in the years after I left and they were still single phase - multiphase is hard, especially when you want to add and drop phases to maximize efficiency as loads change.
As I remember, the peak currents - back then - were middling single digit amps just for the core. NAND could draw a couple of amps if 8 die were busy simultaneously too, I think my name is on a patent about scheduling these to try to reduce worst case draw...
Summary: average powers are dominated by the amount of heat the device can radiate, but are almost unrelated to peak powers.
ISTR a black body radiator the size of the 3.5" screen phones could dissipate 1W per 10C rise, so sustained you were at 2W max (so operation at 35C, the top end of the spec, didn't put the metal parts of the phone beyond 55C, which is the UL safety spec for metal/human tissue contact).
Previous Topic
Index
