It didn’t get much attention, but The Wall Street Journal recently reported that Google has killed its "modular" smartphone project. The effort, code-named Project Ara, has been underway for about three years. According to the article, it had been "a rocky journey for Ara, which began in 2013 under phone maker Motorola, then owned by Google. It later moved to a Google research lab called the Advanced Technology and Projects group." These modular phones would be upgradable, flexible, and longer-lived than our present units—at least, that was part of the hope.

Ironically, Wired had a long, gushing, almost hagiographic piece on Ara just a few months before, "Project Ara Lives: Google’s Modular Phone Is Ready for You Now." As they say, "Oh, well, never mind."

Project Ara 01 Figure 1: Google's Project Ara modular smartphone seemed like a good idea, but its design reality was more challenging than expected, versus the opportunities it potentially offered.

Frankly, I didn't see much chance of success when it started. Much as I appreciate and even like the concept of a phone with replaceable or upgradable parts, and which allows for easy add-ons of peripheral devices such as glucose meters, I know that reality argues against success. It's one thing to have a rack-type instrument with slots for adding in various modules, but doing so in a small, battery-powered, tightly constrained handheld unit is a tough call.

Stop and think about the challenges. There are issues of electrical compatibility, connectors, software integration, mechanical fit, and overall integrity, plus one low-glamour factor often not mentioned: power.

I don’t just mean drain on limited available battery power, although the more add-ons you have, the less run time you get, obviously. Instead, it's more than that: what if the modular add-ons need special voltages, for example? It's nice to think you can standardise on just a few rails for the add-ons, but some may need unique voltages due to the physics of their function. These rails would have to be developed by DC/DC converters within the add-on, adding to issues of efficiency, heat, power-rail transients, and more.

Experienced engineers on both hardware and software sides know that good set of power rails is like a good foundation for a structure: if it is not rock-solid, strange things can happen. Worst of all, they may be intermittent in their occurrence and hard to pin down. This is especially true in the hands of non-technical users, where meaningful data and details about what went wrong and when is sketchy and unreliable.

Today's smartphones are complex devices, with sophisticated power management as an inherent part of their design. If you start adding modular functions to the phone, the power-management strategy will have to adapt and then adopt new tactics and algorithms. Those are difficult mandates.

What about those various plug-in accessories you can get for today's smartphones which interface via the available I/O connector? Those are fine, but those don't make a phone modular. By using such an accessory, the end-user assumes responsibility to any deviations in the phone’s functionality and performance. Also, the basic phone architecture and its physical parts are not changed by the presence of this peripheral device.

Don’t get me wrong: I have nothing against "modular" products and, in fact, I like them in many cases, such as test and measurement instrumentation. I also know that every engineering design is a balance of trade-offs, and a modular product brings some new demands that are very difficult to meet at so many levels in a small product with limited space and power.

In a way, it's like the occasional griping you hear: "Why aren’t today's consumer products repairable? Why do we have to throw them away when they break?" The answer is easy: if you want a low-cost, high-volume small product, it gets designed one way. If you want a product that is accessible, repairable, and modular in hardware without small size, then the basic framework of the design form factor will be very different, often with multiple PC boards, socketed components and even test points and connectors.

We went through this modular phase in the early to middle years of the desktop personal computer. Many users would open the box and change RAM, hard drives, and other internal pieces of hardware, or install plug-in boards. These PCs often came with a list of their internal power rails (voltage and current), supported interfaces, and connectors so users would know what they could potentially add. Pretty soon, no two PCs were the same, and there were compatibility issues, crashes that were impossible to duplicate, and more.

Editor's Note: We've been tracking Project Ara for quite some time now. Here's some of that coverage that shows how the project progressed:

What's your view on the viability of a modular smartphone, or its desirability? Let us know on our Facebook page.