There has been a great disturbance in the Android charging world lately but it is a tempest in a teapot. Let SemiAccurate walk you through the details and hopefully clear a lot up.
This whole affair started out with Android Police noticing a bit in Google’s latest Android N compatibility definition guide. They took this to mean, “Quickcharging bad unless it is USB-PD based“. Three bullet points in the actual source text in Section 7.7.1 says:
*Type-C devices also supporting USB host mode are STRONGLY RECOMMENDED to support Power Delivery for data and power role swapping.
*Type-C devices SHOULD support Power Delivery for high-voltage charging and
support for Alternate Modes such as display out
*Type-C devices are STRONGLY RECOMMENDED to not support proprietary charging methods that modify Vbus voltage beyond default levels, or alter sink/source roles as such may result in interoperability issues with the chargers or devices that support the standard USB Power Delivery methods. While this is called out as “STRONGLY RECOMMENDED”, in future Android versions we might REQUIRE all type-C devices to support full interoperability with standard type-C chargers.
This all sounds pretty dire for Quickcharge and other fast charging technologies, doesn’t it? Before we tell you why it isn’t, if you aren’t familiar with the details of Qualcomm’s Quickcharge 3 and USB-PD, read the links, we will assume you understand this stuff in detail from here on out.
So why do we think this isn’t a problem? The main thing is we are in a hardware transition period from USB-A/B to USB-C at the moment. OEMs who are not on the higher end of the market fight for razor-thin margins on phones and related consumer devices, if you think the 15% or so that a premium PC rakes in is tight, the mid-market phones tend to get a lot less and the low-end, well it is low single digits on a good day. Cost isn’t a problem, it is THE problem. While the high-end $300+ phones all went to USB-C overnight, if you look at the lower end of the market, they are still on micro-B or related ports.
Why? Cost. USB-C ports are much more expensive than A/B variants. Their cables are much more expensive too, and the MUXes and other hardware is yet more. If you go to USB-PD functionality, you effectively replace a small analog chip in the case of Qualcomm’s Quickcharge, functionality which is often built into the PMIC, with a larger digital one, authentication hardware, muxes, and a lot more. Someone SemiAccurate talked to about this article pointed out that you are making a chip for a spec with, “low tens of pages” with one of, “400+ pages” plus adding a lot of external supporting hardware. True the USB-PD spec does a lot more, but in this case you only need a small subset of its features to get the job done here but the hardware needs to support it all.
In short this additional hardware does not add up to tens of dollars like most people think, it is an adder of single digit dollars. Unfortunately this difference is more than the margins of most phone sales, counting by units not dollars. In the world of low to mid-range devices, this cost adder is right out. Why? Because if you implement it and your competitor doesn’t, they can hit a theoretical price point of $99 with razor slim profits and you can’t. Game over in the relevant markets which are the vast majority of phone sales. This is why Google starts Section 7.7.1 of the Android N guide with:
*The port MUST be connectable to a USB host that has a standard type-A or type-C USB
*The port SHOULD use micro-B, micro-AB or Type-C USB form factor. Existing and new
Android devices are STRONGLY RECOMMENDED to meet these requirements so they
will be able to upgrade to the future platform releases.
So Google isn’t mandating USB-C which is the only place that USB-PD can work due to cabling requirements. If they aren’t mandating Type-C, how can they threaten USB-PD as becoming mandatory for quickcharging? The answer is easy enough, the non-C devices can’t use quickcharging technologies under the new order, something that will not please most users of non-C bearing phones with big batteries. If you want a device that consumes power faster than it can be charged, your future outlook is bright.
Luckily for the majority of us, this interpretation is not the case, Google isn’t really saying what most people are reading this document as. If you want to be pedantic, they technically are saying it that way, but the message they are sending to the relevant device makers is very different. What you are seeing is a gentle nudge to push the transition from USB-B to USB-C and USB-PD along. They are standing in front of the OEMs and screaming, “HINT!!!” in their face. If you have been in the hardware world for long enough, you will know that this is how transitions are made, gentle encouragement followed by less gentle encouragements, then the proverbial baseball bat to the side of the head in the form of an eventual hard cutoff. This is not a process for months, it is one that takes years if it ever actually happens. Anyone remember BTX PC motherboards?
USB-C has been around for years now and the high-end phones are just picking it up this year. The reason for this is purely cost, a $300+ phone has the margins to bear it, and their target market sees USB-C as a selling point feature, so in it goes. Below this it is flaming death for devices because of cost and margins. The gentle nudge approach will see it move down the stack and hardware volume will increase in proportion. This also means hardware cost will decrease in proportion, and eventually it will make its way into low-end phones. Once again, years not months.
So rather than the, “OMG the sky is falling” view that the press has, amplified by the echo chamber of additive ignorance that accompanies each copy of the original, this is just normal business as usual in the hardware world. Google wants everyone to move to USB-C and USB-PD for a standard and eventually low-cost way of quickcharging. The new small Pixel phone uses standard USB-C charging because it is under 15W but the larger one uses USB-PD. One look at the costs of the devices says that they can bear a few extra BoM dollars. What did their last device, the Nexus 6, use? Qualcomm’s Quickcharge.
So where does all this leave the current quickcharge technologies like Qualcomm’s Quickcarge 2/3? Exactly where they are now, business as usual. Qualcomm’s works fine, other may or may not, the author doesn’t have enough specs on other ways to comment on each and every competing quickcharging spec. The changes that sounded so dire to Android Police and others are the first gentle nudge to move toward a new spec over the objections of those deciding on the BoM, nothing more other than they are now public. These specs are given to the hardware makers months or years in advance. It is very likely they already know when and how the hard cutoff will occur, if it ever will. Given the BoM cost advantages of USB-B + Quickcharge3 or even slow charging, it is unlikely that those technologies can be forced out for a long time to come.
Hardware takes years to build, test, and produce. PMICs and USB controllers may be a lot simpler than CPUs but they are still much more than a year from clean sheet to silicon out the door. Google knows this, the silicon vendors know this, the OEMs know this, and all are in the loop and know the timetables. Anything else would see the industry just not work. Luckily the way things actually happen are not the way they are read by those not involved. How do we know? At a talk about Quickcharge next at Qualcomm’s 4G5G Summit a few weeks ago which SemiAccurate attended, it was strongly intoned that the next version would be based on, or at least compatible with, wait for it, USB-PD. See, the system works, nothing to see here, tempest in a teapot, move along people.S|A
Latest posts by Charlie Demerjian (see all)
- Intel shows off 10nm 112Gbps SerDes - Mar 12, 2019
- Intel releases Compute Express Link spec - Mar 11, 2019
- Qualcomm rolls out a second gen 5G modem called X55 - Feb 19, 2019
- What is Intel’s Foveros tech and what isn’t it? - Feb 11, 2019
- Why SemiAccurate called 10nm wrong - Jan 25, 2019