"A Nice Next Step is to Figure Out Whether it is Even Possible."

There have been a lot of posts about Google's Project Ara and I just wanted to put down why I think it is a bad idea and just a headline grabber by them.

I have posted it here rather than in Googleplex as its really a general mobile thing and I'd get accused of trolling there.


via www.gadgetreview.com

The reason I think it will fail is just doesn’t make economic sense.

The cost of engineering the component to be interchangeable by a consumer are ridiculous

Take the list below as an example. This is a breakdown for a typical smartphone (SGS4) and it might be assumed these would be the Legoblocks you would make.




Display assy















Sensor interfaces






Power management









Box contents





Taking them one by one:

Memory (RAM)

A cheap, commodity product, normally surface mounted on the same motherboard as the rest of the chips. To make it into a "block" it needs a different form factor with a pluggable connection. As it is static sensitive , this connection needs to be totally enclosed before mating (or have isolation components added). The memory is connected directly to the SOC and this length needs to be as short as possible. Inserting distance and a make-break connector will almost certainly cause interference and reliability issues. The memory need to be matched to the processor so would almost certainly have to be upgraded if you upgraded the SOC module. So maybe you should just make it part of the same module as the SOC…

Display assembly

The single biggest, most costly component. If you want to upgrade it, what is the upgrade going to be? Bigger screen? So the Frame structure needs to change as well. Higher resolution? Then maybe the graphics chip will need to change. But it is part of the SOC so that will need to change too. The screen is connected to both the SOC and the sensor interfaces, again need to be short, need to be static protected.


Probably the biggest example of integration and miniaturisation. CPU, GPU, local memory and caches all in one tiny block. Splitting them out would just be plain ridiculous. All the comments about static and protection apply.


In theory one of the simpler ones. A self-contained module of CCD and lenses. But would you need to change interface hardware when you change the modules? Probably

Wireless and wlan/BT.

Normally on the same PCB as the rest of the chips but also need to be connected to all the antennae. So if you are going to create a module if needs to also include these. If you are going to upgrade these how would it be? 5G? New BT standard? New WiFi? Sounds reasonable but you can bet there will be an impact elsewhere

User interfaces/Sensors

Would you break out sensors individually? Such a low cost it is hard to believe it would be worth it. Again you need a multiway, static safe connector to the main processor

Power management

Supplies all the power for the other systems so needs to connect to them and be sized appropriately. Charges the battery so needs to be sized for that. But which battery? Do you size it for the guy who wants a tiny phone with a 1200mah battery or the 3000mah Hummer phone? So maybe you move the charge to be part of the battery module?


One of the easier ones to do as effectively it exists at the moment and the connection is not difficult but it is also the biggest item so any change will have a significant effect on the size of the device.


My understanding is there will be a frame/backbone that holds this all together. That is needs to be expandable to allow for bigger screens and batteries. Not impossible but needlessly complex. They will also be limited in that certain modules have to be in certain paces (antenna needs to be at an edge as do audio and battery connections for example). Lastly you want this top go together and have a solid feel. multiple plugged modules is not going to do this.

So, in summary:

  • Increased size due to the need to have each module in its own self-contained block
  • Increased cost for these enclosures and the complex connection systems to fit them together
  • Decreased reliability due to the vastly increased number of plugged connection between highly sensitive components
  • Limited upgrade options due to interdependency of components

Other considerations:

  • Warranty - who fikes your phone if it is made of dozens of bits of variosu ages
  • The "Newness" factor - consumers like new things!!

To me, it’s a pointless, headline grabbing idea. It is not impossible but the costs massively outweigh the benefits. They’d be much better pushing to integrate more and drive component costs down further. This is creating a solution for a problem that doesn't really exist. People who want to buy new phones do so and people who don't buy them secondhand.