It’s been a long week of power problems with my consumer electronics. This one’s been building for a while: the battery life on my iPhone 6S sucks. Don’t tell me to turn down my screen brightness, or turn off wifi (BS), or any of that crap. I’m an electrical engineer. This isn’t my first consumer electronic device. There’s a quantifiable problem with this thing (quantified below!)
The battery life on my 6S has never been great. I bought it used to replace a 6 that fell out of my pocket skiing. It was found 4 days later, after I’d lost hope and ordered its replacement.
Owing to its not-great-ness, it’s lived most of its current life in a battery case. At first a cheap one from Amazon, then when the crappy plastic lightning port on that one died, an Apple-official one. I had the battery replaced a few months ago. But through all of this, and with a size-doubling case on it the whole time, it’s still only ever lasted barely-a-day of reasonable use.
It’s also always seemed to get quite hot, especially lately. Like, the screen never operates at full brightness because it’s ALWAYS in thermal protect mode, where the screen dims to reduce strain on the power electronics. In a car with sunlight hitting the screen? Forget about it – it’ll just shut off.
Yesterday it got too much to tolerate. While listening to podcasts with the screen off and no other activity, the phone got hot to the touch and seemed to be draining much too fast.
I began diagnosis by backing up, then DFU-mode restoring both my older 6 with nothing on it and the newer 6S. I restored that backup to the 6 to get my day to day life on something that could handle a full day, then decided to do some A/B testing.
First Test – Topless (but no pictures)
I wanted to see exactly WHAT was getting hot, since it’s not unheard of for shorts on the motherboard to cause high draw. If the battery connector was heating up, or the charging IC, or the memory, that’d be informative. So I opened the screen to have a look at the main board with a thermal camera while I ran the phone. Unfortunately, the thermal images got lost in the backup/restore shuffle :(. All I have from this test is this: Nothing gets unreasonably hot, except the heat spreader over the main CPU (which has the PMIC directly opposite it on the other side of the PCB). The CPU shield ran at a continuous 50C during my podcast stream test, and peaked at 70C while I was measuring. Keep in mind, that’s the heat spreader, not the package itself. I don’t actually know how unreasonable that is – the chip will certainly get much hotter than the package around it, than the heat spreader around that, and than the big aluminum heatsink of a case around that. But a part RIGHT behind the screen seems like it should never get too hot to touch, if only for the sake of display life reduction.
Unfortunately, the location of the PMIC right behind the CPU means I can’t really know which is more likely at fault without a much more invasive teardown. But it does suggest that there’s not “some short somewhere” or a software issue hammering the storage, for instance. Of course, it’s not likely there’s a software problem, anyway: this test was run after a DFU restore and BEFORE restoring my backup.
Second Test – Apple to (Prior-Gen) Apple
While not too illuminating, the first test did at least seem to suggest off-nominal behavior, so I decided to get a better idea of what “nominal” should be.
One would expect that the same exact task running on both phones would consume roughly equal power and generate roughly equal heat. If there WERE a difference, one would further expect that the newer, more efficient, possibly more optimized-for-whatever-task 6S would consume less power and generate less heat. Given the almost identical construction of the two phones, one would further expect that exterior temperature should be a perfect proxy for heat output. In fact, the only construction difference between the two phones is 6061 aluminum for the iPhone 6 and 7075 for the 6S – 6061 has a specific heat of about .9 J/g*C, while 7075 has about .96 J/g*C. That is, the 7075 can hold more heat, which means its temperature will rise slower for the same thermal power into it. So in every facet relating to temperature and power dissipation, the 6S should measure cooler under the same load.
Another factor to consider when measuring “battery capacity consumed” as a proxy for “energy consumed” is what the real battery capacity is. My 6S has a newer battery, which shows 96% health in settings. The 6 shows 89%. That is, if the same task uses the same number of Joules on both phones, the battery percentage should drop MORE on the less-healthy and older 6 than it does on the newer battery of the 6S. All of this to say, greater drain on the older 6 would be possibly inconclusive, where greater drain on the 6S is almost certainly definitive.
To begin the test, I covered both phone backs in masking tape to provide a matte surface with emissivity ~=1. Trying to measure the aluminum directly with a thermal camera provides a poor result at best, and at worst, can actually show a reflection a different object entirely, as if the metal were a mirror. Think about it like trying to look at a very shiny object at a low angle – the glare in infrared can overwhelm the image without you realizing it. Notice in my starting image above that the phone on the left, the iPhone 6, is actually about 1C hotter than the 6S. I had queued up youtube to the same video on both phones, but apparently the app crashed or something because it wasn’t up when I went to begin. I had to use the phone for a minute to get it back to the same starting point, after having let them rest and equalize a while.
The first test was playing the same YouTube video on both phones at full brightness and volume. I recorded the temperature and battery % on both phones:
|After 12m of YouTube:|
|iPhone 6||iPhone 6S|
|97% 29.5 °C||94% 31 °C|
The 6S seemed to be draining much faster (~2x), BUT the temperature was about the same – only 1°C higher on the 6S. I had seen it scorching hot before, so I wondered if maybe YouTube and video playback was actually optimized on the phone in such a way as to avoid running into whatever issue I’m having with disproportionate heat generation.
Carrying on, I decided to switch back to the podcast test, which I knew to cause heat. I measured at a few more time intervals after that, with NPR One playing on both phones (NO screen)
|iPhone 6||iPhone 6S|
|At 20m total (8m NPR)|
|94% 30 °C||57% 45 °C|
|At 32m total (20m NPR)|
|At 36m total (24m NPR)|
|88% 34 °C||54% 45 °C|
So under a load where the phone should be able to chug for hours, the 6S runs about 15°C hotter than the 6, and consumes battery 3.85 times as fast (77%/hr vs 20%/hr). With a healthier battery.
If the numbers aren’t making the point on their own, I’ll say that 45°C is jarringly HOT for a phone. Like, left-it-in-a-sunny-spot hot. Not dangerously hot, and not uncomfortable like a hot pan handle hot, but very much annoying-to-hold-for-minutes hot.
I’m annoyed as hell lately with electronics. My 2018 MacBook Pro keeps crashing while it sleeps, the Surface Pro seems to be waking up all the time, and my iPhone gets hot and dies.
At this rate, I’ll probably have to suffer the slightly slower 6 until I can’t take it anymore and upgrade. Unfortunately, not even Louis Rossmann does board-level repairs on iPhones like this likely requires – too difficult to diagnose and successfully rework by hand, I guess. My best guess at this point is that there’s some kind of load-dependent power burn, like a high junction resistance on a regulator input pin – there are a few things that I suspect could cause just that. I’ll probably have a go at taking the shields off and reflowing it myself, but there’s a good chance I destroy the board in the process, and it’s definitely a project for a day with more time.
The REAL conclusion here is this: If you have an older iPhone that seems to be getting way hotter than normal, and has crap battery life even with a brand new cell in it, you might not be going crazy, after all. And maybe buy a Pixel this time, or something.