Analyzing the Ara Project Of Google's Motorola Subdivision
Comment Now
Follow Comments
Dan Zhang, Computer Engineering PhD student at UT Austin
Motorola’s newly announced Project Ara is an innovative project designed to introduce customization and modularity to the smartphone market. Rather than the current model of highly integrated phones that we have right now, Project Ara allows for component mix-and-match, so that you don’t need to buy a new phone when your screen breaks or if you want a larger internal Flash storage.
Right now, there isn’t a lot of information released by Motorola. However, by simply looking at the pictures, we can make a lot of educated guesses about how it works and how it’s different from the impractical but similar PhoneBloks (see Dan Zhang’s answer to Smartphones: How technologically feasible is the idea of Phonebloks?)
…
First of all, lets take a look at some of the factors that made PhoneBloks impractical and see how Motorola resolves it with Ara:
- PhoneBloks undoes years of progress through integration using SoC design by breaking the SoC into its individual components.
From my best guess, Motorola Ara solves this by avoiding the problem entirely. This is what Motorola describes as what a module can be:
A module can be anything, from a new application processor to a new display or keyboard, an extra battery, a pulse oximeter–or something not yet thought of! – See more at:Goodbye Sticky. Hello Ara.
It’s important to differentiate marketing spin from specifics. Note how they never mention anything about the components within an SoC or its main memory! My guess is that this SoC (e.g. an NVIDIA Tegra 4) must be upgraded in an all-or-nothing fashion, replacing the Tegra, VRMs, heatsinks, and DRAM within the same module. You can augment the features of the SoC through the application-specific processors and other peripherals.
- PhoneBloks’ modularity wastes space and increases the overall size of the device.
True, but amusingly this is starting to matter less and less. Unlike laptops, which have shrunk from the average 8lb 15.4″ form factor to svelte 3lb 13.3″ Ultrabooks, phones are growing in size. Right now, Apple has proven that we can make an extremely high performance phone fit within a slim form factor with a 4″ screen. However, clearly a market exists for phones with larger screens (the new Google Nexus 5 has a 4.95″ screen!), which increases the overall volume of the phone. Therefore, we can waste some of this extra volume on modularity.
So how big is the Motorola Ara prototype?
Using the following image, we can calculate the dimensions of the phone.
We know the dimensions of the Micro-USB Type B connector shown in the picture. Using Photoshop, I determined that it is approximately 39×10 pixels wide. I assumed that the dimensions did not include the metal surrounding the connector. The lengthwise dimension of the phone is 801×53.
We can now convert between pixels and millimeters by using the reference dimensions shown in the lower image. Note that the ratio in pixels above is 3.9 whereas the ratio below is around 3.72. Close enough … we were never going to get it exactly right anyways.
This means that each pixel is ~0.177mm in each dimension. Doing the simple math (stay in school, folks!), this implies that it’s roughly 141.8mm x 9.381mm in dimensions.
Note that this doesn’t tell us anything about the width of the phone. However, we can estimate it by comparing it with existing phones.
The massive Google Nexus 5 (with a 4.95″ screen) has dimensions of 69.17mm x 137.84mm x 8.59mm. This implies that the Project Ari prototype is even larger!
I estimate the Project Ari prototype to have dimensions of 71.15mm x 141.8mm x 9.38mm with a 5″ LCD screen.
This provides plenty of space to waste in the name of modularity! You can probably even give each module its own EMI shield and still have room to spare.
- A breadboard-type base doesn’t work; each device requires a different interface and the interface changes across generations. Plus, components such as DRAM especially need high bandwidth buses, and the performance of this bus is guaranteed to change every generation in the name of performance.
My guess is Motorola Ara solves this by either making performance-critical components non-upgradeable or packaged together within a single module. This enables high performance modules to scale with every process generation.
Many peripherals do not require high bandwidth or low latency. For these components, we can use some sort of standardized bus to connect them. There are many examples of such interfaces, each with their own strengths and weaknesses. Potential interfaces include PCI-Express, USB, and Intel Light Peak (also known as Lightning). USB is probably the most likely choice due to its ubiquity, power restrictions and cost, or they may choose to create their own low-power proprietary connector. (Motorola, if you’re reading this, please choose USB!)
- All functionality found in these different blocks currently require custom software drivers, with custom interfaces to communicate with each other. This may be possible to standardize but the barrier is extremely high.
This is true. However, we’re already assuming that all performance-critical components are non-upgradeable or all found within one module, so the driver issue for those components is non-existent. Right now, each functional block has its own proprietary bus. If modules connect using a standardized interface such as USB, module manufacturers can write the standard USB drivers in a same way that this is done on desktop computers right now.
- Even if size doesn’t matter, extra packaging for each module will greatly increase the overall weight of the phone (thanks Rupert Baines).
Ironically, customers actually like weight in their phone. Plenty of massive, high-end smartphone have been criticized for being lightweight, which makes it feel “cheap.” Example: http://arstechnica.com/gadgets/2…
By using the same lightweight materials that have been criticized for feeling cheap (i.e. plastic rather than metals) for the extra packaging, we can create a phone that feels good in a customer’s hand with plenty of heft.
…
Now, lets consider what this Motorola Ara project might look like and its implications on the consumer and the overall market.
What are some example modules that can be found? For example, we have have low to high resolution display support (480p TN, 720p IPS, 1080p IPS). We may have cameras of various qualities. We may be able to upgrade the amount of internal flash memory. We can replace the SoC from processor generation to generation without having to throw away the entire phone. The most technically interesting part though is the ability for specialization.
For example, a major problem with accelerators right now is that what’s popular right now won’t be popular a year in the future. If we just put a “popular program accelerator” in a module, then this module can be upgradeable every year or so based on what the user plays. Think of it as a yearly flu shot! Of course, accelerators may need a high performance bus in order to be practical. Nowhere is it stated though that each module slot has the same interface! For example, a Motorola Ara phone may have 4x USB2.0 and 1x PCI-E3.0 2x module slots in it.
Also, if this is an open standard, then manufacturers looking in can start making modules too. For example, Xilinx may choose to make an FPGA module to replace that “popular program accelerator” that I’m talking about. Now you’ll only need to reprogram the FPGA to retarget the popular programs. Currently, the barrier of entry is too high for Xilinx since the risk of failure for something so radical is fairly high. This puts the burden of failure on Xilinx, not the phone company, so more aggressive designs can be built.
Now, lets discuss the marketing details for this phone. Currently, most cellular providers subsidize the true cost of the phone, such that people only pay $100-200 out of the total $650 cost of the iPhone or Samsung Galaxy. However, there’s no such thing as a free lunch: rest assured, you are paying for the full $650 amount as the difference is cost is simply applied to your monthly bill. However, customers are not exposed to the true cost of the phone and may experience sticker shock when, for example, they see that upgrading their Ara to a new SoC costs them $250, well above the cost of most subsidized phones.
However, there do exist a few cellular providers that do not subsidize or offer the option to not subsidize the cost of the phone. For example, T-Mobile recently made headlines when it dropped phone subsidies and contracts: T-Mobile Ditches Contracts, Phone Subsidies
For companies such as T-Mobile, this phone is simply a match made in heaven. As T-Mobile is not providing any subsidies, T-Mobile customers are less likely to want to upgrade phones. Therefore, incremental upgrades of $250 for a new SoC module in your Motorola Ara is much more attractive than paying $650 for the latest base model iPhone.
…
In conclusion, Motorola’s Project Ara, while similar to PhoneBloks at a high level, is actually likely to be quite different in order to solve technical challenges posed by such modular design. This design has the potential to be quite practical for certain larger form factors, which industry has been tending towards anyways. There are still plenty of challenges left to be solved, but the overall idea seems promising enough to be worth trying.
