A hardware design (was Re: open cell phone hardware questions)

Harald Welte laforge at gnumonks.org
Fri Jan 14 19:37:06 UTC 2011

Dear Scott,

On Fri, Jan 14, 2011 at 03:39:23PM +0200, Scott Weisman wrote:
> I was wondering if anyone knows if the Calypso and other chips in
> these old phones are still available for new designs? 

Not from TI, it is end of life for a long time.  However, you can certainly buy
the Calypso/Iota/Rita from the 'grey market', i.e. the various surplus
semiconductor traders in Asia.

At least one year ago when I last inquired, I could have easily bought
something like 10,000 of those old chipsets.

The biggest challenge is to find all the other required parts, like SAW filters
with the right impedance and mechanical footprint, and find a stable source of
them where you don't have to change your PCB layout every time you produce
another batch of the phones.

We were considering doing something like this (custom hardware) using the
hardware design of the GSM modem of the Freerunner as a basis.  However, when
we found that you could still buy the Motorola/Compal phones in batches of
thousands one year ago, we decided not to pursue down that road.

Doing a hardware design of this complexity, and building it, including
production testing and calibration is not something you are likely to do as
a one or two-man show in your spare time (which OsmocomBB was a year ago).

Also, if you actually ship a device consisting of hardware + software, you
easily run into regulatory issues like the RT&TTE directive in the EU, since
suddenly you're now selling an end-user consumer device and thus have to comply
with regulatory requirements.

Just distributing software is not falling under that directive, and thus
no certification / regulatory requirement is needed.  Same goes for measurement/lab
equipment like the USRP.

> How much interest would there be, say, in an open, but VERY SIMPLE, actual
> phone? 

Definitely by far not enough interest to ever justify the many man-weeks and
months spent in doing a hardware design, writing production testing software,
getting the tooling done for the plastics, etc.

What I find much more valuable is to put effort into porting OsmocomBB to the
Mediatek MTK62xx chipsets.  There are hundreds of millions of phones in the
market already, and close to 100 million new ones are shipping per quarter.

This way, once again the focus is where it should be: creating and improving
the Free Software GSM protocol stack, creating a simple UI on top.

No need to design, test, produce and verify a hardware design from scratch,
worry about mechanical issues, etc.   Plus, you benefit from the fact that
the devices are available at a price to which you will never get (large scale
production), and the hardware design has been verified.

> The Osmocom software would then be very easily portable to such a device.
> Given the seemingly widespread interest and enthusiasm for the Osmocom,
> OpenBTS, and OpenBSC projects, 

I think you may need to re-calibrate your perception.  Please look at the
actual number of people who have contributed to the source code of all those
projects together.  You will find it is very few, and the same names pop
up everywhere.

Growing the user base without growing the developer base is not something I
think we should put in such an enormous effort as running our own hardware

- Harald Welte <laforge at gnumonks.org>           http://laforge.gnumonks.org/
"Privacy in residential applications is a desirable marketing option."
                                                  (ETSI EN 300 175-7 Ch. A6)

More information about the baseband-devel mailing list