原创 The "anything machine" is here again

The traditional desktop PC of the 1980s and 1990s was built around a standard bus architecture - first IBM's PC ATbus and then Intel's Periperial Compoent Interconnect (PCI). As a result of this strong feature, even in the average consumer's hands, it was an "anything machine."

Unlike the many of the current generation of PCs and ALL of the mobile smartphones where "what you see is what you get" with no options for individual options, traditional PCs were malleable and reconfigurable for a variety of applications and user needs, depending on which boards you inserted into the common bus.

I was not a developer, but it was possible for me to buy a basic PC and then add the features I wanted,- and sometimes remove those I didn't - by swapping out boards until I got the features and performance I desired.

It has been one of the frustrations of my life - and one of many reasons I have not bought one - that smartphones, including Google's Androids, are basically hand held PCs with everything that they think you want crammed into it, most of which I find useless or not worth the wasted memory space. .

While you can shop around and get something close to what you want, almost every smart phone I have looked at comes with a truckload of features and capabilities that I do not want. They take up memory and screen space from the things that I do want and there is no way to get rid of them and add in something I want.Sure there are thousands of apps you can download, but to rework it to give you the performance you want and any number of hardware related options, you are out of luck.

Now, according to officiialls from Google's Ara Project team,they have developed a modular developer kit for an open source build-your-own mobile phone. It was on display this week at Google's first Ara Developer Conference at the Computer History Musuem, home of a replica of one of the first computer do-it-yourself projects: Charles Babbage's 1837 Analytical Engine

The customisable Ara can be built using a set of 20x20 mm slip-and-click magnetic modules, stackable in various configurations for 4x7 inch, 3x6 inch, and 2x5 inch products (Figure 1 below). The prototype comes with Texas Instruments’ OMAP 4460 mobile processor, built on a dual-core ARM Cortex-A9, with a Melexis thermal image sensor and TI pulse oximeter.  Should the Project Ara team develop another prototype for a different size, it may come with another type of processor.

Currently the prototype they showed off is a proof-of-concept setup and the team is hoping that eventually there will be as many as 50 different modules for many of the processors available as well as for a variety of peripheral functions.

The current MDK provides developers with reference implementations, model templates, and sample code. For those who want to get on the hoped-for bandwagon of companies who want to add their own modules all such modules must support exchange no more than five watts of power, the limit of Ara’s aluminum endoskeleton. Other spects are support for 10 Gbit/s of data transfer. All modules must operate at either 3.0 or 5.5 volts.

The Ara prototype the team showed off in San Jose, Ca. has multiple batteries connected in parallel, configured for very high bursts of data communication. But I can also be configured to operate in a low-power hybrid mode when one battery is low. While it isn't configured for wireless inductive power charging, the kit does allow for various battery configurations - one or several batteries for the complete system or in a one module-one battery configuration.

It also allows for inclusion of several chargers within the same phone design. While the kit can be customised with SIM cards and wireless for foreign markets, the Google Ara team has not had time to build a work-around for the US carrier-based market. The team is considering development of an SoC for use in a cellular communications module which will incorporate 3D-print antennas as a set of conductive layers by early 2015.

There are a lot of questions that so far the company - nor the development team - have answered relating to the cost of the individual chip sets or the modules. This is an important consideration since if the components were cheap enough such a do it yourself platform could threaten existing manufacturers of mobile phones.

Then there is the question of who will be interested in such a modular approach,: the geek or engineer who can - and wants to - jump into the details of its construction OR an average non-techie who would need a much more tinker-toy approach.

That was the power of the PC and its common AT bus. It was tinker-toy enough to allow average users who were not satisfied with what they had to simply buy a board, plug it in and install the software to get the additional function needed. 

Ara is a good step in that direction, but Google has an uphill battle.

When the PC burst on the world, other than Apple a few years later, there was not existing norm to replace. With no preconceptions to get in the way, via the use of the common AT bus allowed even average users customised their purchases.  It only evolved later into a much more inflexible platform with only nominal ways of reconfiguring it.

Now Goggle and Ara has come into the market where the norm is one-size-fits all in terms of basic features and capabilities. Sure, you can do some shopping around and eventually find almost what you want, but seldom at the price you want and usually with something you do not want. Despite the flexibility that the Ara Project implies, I just do not think the majority of the market will buy into the concept. Nor is it clear to me that even the DIYers amongst them will either.