Following the attention that Apple had gotten over the past few weeks regarding the discovery of mechanisms that reduce CPU frequency on devices with aged batteries, Apple has now issued a more comprehensive statement and apology addressing the matter:

First and foremost, we have never — and would never — do anything to intentionally shorten the life of any Apple product, or degrade the user experience to drive customer upgrades. Our goal has always been to create products that our customers love, and making iPhones last as long as possible is an important part of that.

When power is pulled from a battery with a higher level of impedance, the battery’s voltage will drop to a greater degree. Electronic components require a minimum voltage to operate. This includes the device’s internal storage, power circuits, and the battery itself. The power management system determines the capability of the battery to supply this power, and manages the loads in order to maintain operations.

The statement doesn’t address any new information as to cause of the issue and confirms my initial technical explanation of the battery impedance causing the battery to no longer be able to supply a stable voltage supply during transient loads.

What we do have as new information is the various other effects that the throttling mechanism touches:

  • Longer app launch times
  • Lower frame rates while scrolling
  • Backlight dimming (which can be overridden in Control Center)
  • Lower speaker volume by up to -3dB
  • Gradual frame rate reductions in some apps
  • During the most extreme cases, the camera flash will be disabled as visible in the camera UI
  • Apps refreshing in background may require reloading upon launch

As it appears, CPU and GPU frequency reductions are not the only things done by iOS to prevent shutdowns of iPhone 6, 6S, SE and 7. The system also reduces backlight dimming (which can be overridden in settings), lowers speaker volume by up to -3 dB, disables camera flash cease app refreshing in background. All of the said performance-related features are important to the user, yet are not crucial when it comes to phone usage in general and in emergencies in particular. Apple stresses that while it reduces SoC frequency, it preserves cellular call quality, networking and GPS performance, location accuracy, captured photo and video quality, operation of sensors as well as Apple Pay. In fact, the FCC and other regulators have a set of emergency-related requirements and recommendations for wireless service providers and hardware manufacturers, there is also the Twenty-First Century Communications and Video Accessibility Act and the iCanConnect program to provide people with disabilities a viable way of communications (including video communications during emergencies). To participate in this and similar programs a vendor probably needs to guarantee that its hardware can make the aforementioned features (and therefore emergency services) available to users at all times.

Based on the large media attention and relatively negative feedback which prompted Apple to this second official response and statement, Apple promises three key points to address consumer’s concerns:

  • Apple is reducing the price of an out-of-warranty iPhone battery replacement by $50 — from $79 to $29 — for anyone with an iPhone 6 or later whose battery needs to be replaced, starting in late January and available worldwide through December 2018. Details will be provided soon on apple.com.
  • Early in 2018, we will issue an iOS software update with new features that give users more visibility into the health of their iPhone’s battery, so they can see for themselves if its condition is affecting performance.
  • As always, our team is working on ways to make the user experience even better, including improving how we manage performance and avoid unexpected shutdowns as batteries age.

Reduction of the cost of an official battery replacement from $79 to $29 is a much welcomed change that makes this a much more attractive option considering replacement batteries only cost $10-15 depending on model; Apple’s previous pricing at $79 was extremely extortionary given the critical aspect of this service. I would now recommend any users who hesitated on replacing their iPhone batteries on their own to make use of the official service as it will have very noticeable impact both on device battery life as well as device performance (due to the nature of this story). Meanwhile, the program has been announced for the U.S. and it remains to be seen how Apple handles it in other countries, including smaller European states that barely have official Apple service providers that can handle additional workload

The way that Apple has handled disclosure on the throttling mechanisms has also been heavily criticised as users felt their devices slowing down with iOS updates and not knowing the reason. Here Cupertino promises key changes in the way that iOS handles information sharing on battery health and reporting, as well as promised improvement on performance management under degraded battery conditions. The issued time-frame for when we can expect these updates are “early 2018”.

Overall the response from Apple was the only possibly correct one to the whole fiasco, and the only one which was to be realistically expected, though it took longer than it should have to implement changes such as drastically reducing the battery replacement cost.

POST A COMMENT

98 Comments

View All Comments

  • willis936 - Friday, December 29, 2017 - link

    There is an element of truth that they try to use to mask their lie. The simple circuit model of a battery is a voltage sourcr in series with a current limiting resistor (referred to as internal resistance). The internal resistance goes up as a battery ages or gets cold and I’m sure many other physical effects. This lowers the peak current it can supply. However the lie is in saying that the phone’s ever pull close to a battery’s peak current, even when aged. If this was the case then every old smartphone would be shutting down or locking up randomly. Transient load spikes are filtered out by energy storage elements (capacitors) in the power supply.

    It’s okay to run damage control on a mistake. It’s not okay to lie. I’m pissed about their response.
    Reply
  • shadowjk - Friday, December 29, 2017 - link

    You are correct about the simple circuit model of a battery. And it's a close enough approximation when you're operating at a point far from the absolute maximum ratings of a battery, and with a new battery.

    As the battery ages, its internal resistance grows, as you quite correct stated.

    However, some patterns of use can significantly alter in what way the resistance changes. Whether those patterns be be time spent on charger near 100%, time spent empty, or time spent in high or low temperatures, rate of discharge, or combinations of all of the above, it's hard to tell. I've profiled the internal resistance of various batteries from various manfuacturers, subjected to various real life uses, and found that sometimes the internal resistance on a used battery will increase sharply towards empty, in some cases exponentially. Furthermore, the industry standard cycle tests are very "easy" for a battery, in that the internal resistance of a laboratory-aged battery stays relatively fixed regardless of the state of charge.

    So that's one factor to consider, batteries subjected to real life cycles behaves differently than batteries subjected to the controlled method and circumstance of industry standard tests. The engineer counting on getting exactly the performance indicated in the datasheet will be disappointed.

    Then the next factor, the internal resistance will also rise exponentially if the current draw approaches the absolute maximum capability of the battery. The maximum will go down as the battery ages, and the internal resistance goes up

    Normally only things like high performance RC helicopters and such operate near the battery's maximum.. Now if a RC Heli can empty the battery in 10 minutes, how can the power draw of a phone, which is tiny in comparison to a Heli, cause issues? Simply, because the batteries are built differently. Phone batteries are optimized for energy density, sacrificing their ability to provide power.

    I think it's the sum of 3 factors that has hit Apple's design, their batteries are heavily optimized for storing energy (sacrificing power), their SoC is (one of?) the highest performing in the industry (which means peak power use is the highest), and their battery capacities despite having energy optimized batteries is relatively low compared to other phones.

    So, small battery optimized for providing a lot of energy at low power levels, coupled with a SoC that uses alot of power (admittedly briefly, but that doesn't matter). Being Apple, they pushed the boundaries, and this time they pushed it a bit too hard.
    Reply
  • Pneumothorax - Monday, January 01, 2018 - link

    As a fellow RC Heli flyer I had the same hypothesis that this is what is happening. As Apple is getting away with the smallest mah rated batteries as they’re so power efficient, it’s now biting them in the rear as smaller capacity batteries have lower max amp output vs android phones which have much larger average batteries. Total capacity X C rating = max amp output. So as the capacity goes down the max current output goes down with it. If you’re trying the pull higher than this the voltage crash occurs.
    As android phones have more margin for battery degradation than what Apple has they don’t have this problem as common
    Reply
  • lmcd - Friday, December 29, 2017 - link

    Wait this comment is actually so incredibly wrong. Battery degradation is primarily about an inability to maintain the original voltage curve, which is vital to operation. Reply
  • willis936 - Saturday, December 30, 2017 - link

    No it is not wrong. Voltage curve? Voltage vs. what? Do you know what you’re talking about or are you blowing smoke? Reply
  • Dragonstongue - Friday, December 29, 2017 - link

    well said.

    Then there comes the constant when a new model is released of course they "update" the software to slow the older models down to upsell the new ones.

    from default they should have had a battery service life tester or for that matter use properly thick phones so could use a larger capacity battery..the built in tester needs to be accurate but not constantly paging the battery every ms either (conserve battery life, if the software is smart enough, they KNOW how much an app is consuming power wise is a simple 1+1=2 calculation based on X % of "life" remaining)

    Also, all phone makers BY LAW should have to make batteries user changeable/replacable (easy to buy OEM or better battery as soon as that model hits the market) instead of having to bring to a "authorized repair person" or buy fancy tools/kits to replace the battery..the phone itself is proprietary, ok but the ability to replace batteries should not and never be ^.^

    Apple, Microsoft, Nvidia, Intel and many others seem to be "smartly" engineering these things to fail as covertly as possible, instead of building them for the quality we are paying for.
    Reply
  • sonny73n - Friday, December 29, 2017 - link

    I replace iPhone battery for $15 and it'll be done in 5 minutes. You should stop defending crooks. Reply
  • Andrei Frumusanu - Friday, December 29, 2017 - link

    > You seem to play the analyst fairly frequently, but that comment makes zero sense. First of all, this isn’t a design flaw.

    If the issue at hand is simply transient load spikes caused by in-rush currents then it's *absolutely* a design flaw in the power delivery system of the SoC as it wasn't designed to be robust enough for the *inevitable* voltage drop that comes with ageing batteries.

    DVFS is a thermal and energy efficiency and energy preserving mechanism, not a system for designed for mitigation of power delivery issues. In this case it's a patchwork solution for a hardware design issue.

    > Yes the battery might only cost $8, but the labor isn’t free and there are other plenty of other costs as well.

    It takes 15 minutes tops to change an iPhone battery, experienced people can it in probably 5 minutes. $29 is reasonable for labour costs. $79 is not.

    > Why is Apple liable for any of that when no other smartphone OEM is?

    Because no other smartphone OEM with sufficiently large user-base currently suffers from this. Snapdragon 810 devices suffered from this for likely the same reasons, and in my opinion they and Qualcomm should have been liable for class-action lawsuits, but I guess that never happened.
    Reply
  • shadowjk - Friday, December 29, 2017 - link

    I think the maximum discharge current for the battery is dropping as it ages, and they're encountering the exponential part of it.

    That is, their batteries are performing like this (example, fictional, but perfectly plausible numbers):

    0.1W Discharge: 5000mAh battery capacity
    0.2W Discharge: 4950mAh battery capacity
    0.5W Discharge: 4900mAh battery capacity
    1Watt Discharge: 4800mAh battery capactiy
    2Watt discharge: 4000mAh battery capacity
    4Watt discharge: 2000mAh battery capacity
    5Watt discharge: 100mAh battery capacity
    5.5Watt discharge: 0mAh capacity

    So in this hypothetical scenario, if they limit the maximum draw by 2Watt they increase battery capacity by a factor of 30.

    I'm reminded of days gone by when phones had NiMH batteries, and once you had a really old battery, there was enough power for SMS and Standby, but when someone called and you answered the phone just died because the battery couldn't sustain the power demand of the radio transmitting continuously.

    Back to the apple situation, I think the more interesting side of it is, how do they determine what the present maximum is? Their battery fuel gauge chip can potentially track internal resistance, or maybe they just record unexpected shutdowns, and bump power limit down one notch after each unexpected shutdown?

    In the future, will we have SoCs that can shift powerstates quickly enough to counteract falling voltage in real time instead of Apple's present method of setting static performance limits?

    That's atleast what the engineer in me finds fascinating, the technical details of it.

    If I put my user hat on, I'd just ask "Why can't they just add an inch of thickness to the phone so they can fit a properly big battery in there and we wont have these issues and these discussions?". :-)
    Reply
  • Andrei Frumusanu - Friday, December 29, 2017 - link

    Those figures are not plausible at all. Generally batteries shouldn't be continuously discharged at more than 1C, which means 1.8A or about 7W which anyway is well above a phone's thermal envelope anyway. However if the power delivery system isn't up to par and relies on a current rush to come from the battery then that can lead to failure.

    > In the future, will we have SoCs that can shift powerstates quickly enough to counteract falling voltage in real time instead of Apple's present method of setting static performance limits?

    That only exasperates the issue as instantaneous current rush will be even higher, again, the solution to this is either just have a bigger battery or have a proper decoupling circuit which can handle the current spikes.
    Reply

Log in

Don't have an account? Sign up now