FreeCalypso progress update

Mychaela Falconia mychaela.falconia at
Fri Feb 9 21:28:30 UTC 2018

Hi DS!

> This is excellent news! The calibration report looks really good too.

Further update: I got an email from R&S saying that the unit has been
shipped, and tracking indicates that I should be getting it on Monday.

> Apparently your unit has a WCDMA option installed?

Yes, it does, and it is not a particularly rare option: most CMU200
units on ebay are configured similarly to mine, including GSM, UMTS
(WCDMA), a B11 or B12 OCXO and an AuxTx generator (either B95 or B96).
Some also have additional hw and sw for CDMA2000, although mine
doesn't have it.  I am not really interested in the Verizon/Sprint-
style Amerocentric, SIM-less, non-international CDMA2000 standard, but
UMTS is a different story - given that UMTS belongs to the same larger
family as GSM and given that TI did have a dual-mode GSM+UMTS protocol
stack which I would love to liberate some day, I certainly like having
UMTS/WCDMA RF testing capability.

> Although it's not
> really needed for Calypso or even LoCosto, perhaps it might come in
> handy if you'd like to venture into 3G

I would *love* to be able to venture out into 3G/UMTS land and build
dual-mode phones and modems with the user having absolute total
control over the network search preference order (prefer 2G over 3G or
vice-versa), but the biggest current obstacle in that direction is the
lack of a protocol stack implementation for UMTS.

Most people in the "I want a libre phone" scene keep lamenting about
low-level hardware specifications for chips being withheld, but they
keep missing the much bigger problem.  Even if Qualcomm or MTK or
whoever were to release the full register etc docs for one of their
chips, that full documentation would still be mostly useless without
the source code for their complete turnkey product implementation.
Cellular protocol stacks of end user product quality (as opposed to
toy quality like OsmocomBB) are among the most complex software works
in the world, and I consider it absolutely infeasible to even think of
trying to implement one from scratch, even if there is an underlying
hw platform with absolutely perfect and complete documentation.  A job
like this would require 1000 person-years minimum.

Instead I firmly believe that the only practically feasible way to
gain a libre mobile solution for 3G/UMTS that would be no worse than
what we currently have for GSM/2G with FreeCalypso would be to obtain
and liberate the source code for some *already existing* commercial
implementation of the necessary protocol stack, similarly to how our
current 2G solution came about.

But there is an even further complication.  Suppose we were to find
some firmware source from, say, MTK that supports 3G.  If the vendor
is someone other than TI, that firmware's internal architecture would
be completely different from TI's, all surrounding software tooling
would need to be completely different, meaning that absolutely all of
the work we have done in FreeCalypso would need to be thrown out, and
we would have to restart from scratch working our way around the other
vendor's software architecture and paradigm.

Thus based on the reasoning above, what I really want is a 3G/UMTS
protocol stack in full source form not just from anyone, but
specifically from TI.  As far as I know, TI's 3G implementation never
saw much widespread use (unlike their 2G stuff which used to be super-
popular), as TI killed that baby and threw everything in the trash
just as they were getting their 3G code into good shape.  Given how
little life that code saw outside of TI (if any at all), I'm guessing
that the only way to get a hold of it would be to find someone at TI
who remembers that they once were in that business and convince them
to sell it.

It needs to be kept in mind that back in 2008-2009 TI were actively
looking for someone to buy their unwanted Wireless Terminal Chipset
Business Unit - for someone to buy that entire division wholesale,
that is - but when they *failed* to find an interested buyer (both MTK
and Qualcomm already had their own stuff, and no one else saw any
point in trying to compete with them based on TI's lagging stuff),
they threw everything in trash.  Given that they currently make ZERO
money from the "IP" which they threw in the trash, I don't see why a
rational manager would refuse to sell it, but finding someone who
would think rationally would probably be the most difficult part...

Hasta la Victoria, Siempre,
Mychaela aka The Mother

More information about the Community mailing list