Does competence matter when proposing a new platform?
We did recently found about the ‘Project Ara’ of Goggle – about a platform for creating modular smartphones.
First of all we are glad that Goggle is realizing the importance of connecting hardware outside of the devices based on their platform and finally opening this area for development.
Apple, on the other hand is a hopeless case. The most popular way of connecting devices on iOS devices has been through the audio port – ridiculous from the perspective of a hardware and system software developer.
There is a separate web site and documents -
Project Ara , Module Developers Kit (MDK)
Release 0.11 (alpha) , May 28, 2014
which bear the Google’s copyright.
The first look at the document shows it has been created by people skilled in mechanical design.
However most likely they:
- did not hold a soldering iron in their hand
- and, did not read any technical reference manual of a chip
- and, did not write any device driver
- and, did not set a foot in Shenzhen, China
So, what specifically is objectionable in the ‘project ara’ doc?
1. Regarding mechanical design
Analysis:
While it is innovative, I am not sure putting connectors and modules on the back of a smartphone is a good idea.
It will definitely increase the thickness of the device , and the bulge in my pocket.
Making a connector out of individual pins, using magnets, laser cut holders spells – exotic and expensive!!!
Proposing:
Using standard (cheap!!!) connectors, places on the edges of the device.
2. Regarding electrical design
Analysis:
The proposal defines pins using LVDS pairs – table 3.13 . I did not know yet of and would be very surprised to find an ARM SOC having LVDS level signals.
The voltage level on the power pins is not firmly defined – hooked directly to the battery.
Proposing:
Define standard voltage level – for ex. 3.3V as the voltage on power pins.
The ARM SOC has a combination of the following interfaces on pins – GPIO, I2C, SPI, SDIO, USB, PCI Express (specific chips).
Place on separate connectors and wire to them directly or through buffers the following interfaces -
- I2C and/or SPI
- USB 2.0 host
- GPIO, and/or parallel bus implementation
- PCI Express (if available )
3. Regarding system design
Analysis:
The doc defines the system as a network architecture around MIPI M-PHY and UniPro protocols – allowing the modules to communicate with each other.
The modules have to have a bridge based on LVDS level signals from MIPI M-PHY and UniPro protocols to – I2C, GPIO, USB …
Proposing:
The above system design over-complicates the system design. The ARM SOC already natively have a number of interfaces. It is easy to wire directly I2C, GPIO, USB … to connectors.
There is already kernel level support for some of the interfaces, and additional support would be easyer without the need to write software for GPIO to MIPI M-PHY/UniPro control and then from MIPI M-PHY/UniPro to I2C, SPI, GPIO, SD … emulation.
4. Regarding manufacturing
Prototype hardware for ‘Project Ara’ is available not directly from Google but from a Chinese OEM/ODM – AQS / aqs-inc.com .
Regardless of all assurances, when putting licensees in the hands of OEM/ODMs Google has to explain to the licensee that they will not completely own the design.
A company owns the design when they select components , design the pc board and enclosure.
PCB Manufacturing and assembly in China is fine as long as a company knows what are they doing.
We do not want to criticize only.
We can actually write and offer a counter proposal incorporating the above points.
How about creating and having an alternative – ‘project ARS‘?!