Free board offer to OBB developer with a CMU200

Mychaela Falconia mychaela.falconia at gmail.com
Sun Apr 7 05:39:25 UTC 2019


Craig wrote:

> I have been working with osmocom-bb since around 2010,

So you have been at it for at least 3 y longer than I have, and you
still haven't offered much to show for it..

> dabbling in porting layer1 to nutt-x

What is the deal with that NuttX thing?  What makes all of you think
that it is such a good idea?  Just because someone said so?  None of
the mainstream, commercially successful GSM modem firmwares from any
reputable vendor use NuttX, they all use Nucleus, ThreadX or L4
instead.  (Both MTK and the stuff I inherited from TI use Nucleus, and
I seem to recall hearing of some other vendors using ThreadX and L4.)
So I would be extremely wary of that NuttX stuff.

> and starting 2-3 years ago working more earnestly on
> understanding layer1 and porting it to fernvale.

2-3 y ago is the same time when I went from running FreeCalypso purely
as a software project to producing my current FCDEV3B hardware.  So far
my approach has worked out a lot better: FCDEV3B is now a practically
usable commercial-quality GSM+GPRS modem that is fully fit for
operation on live GSM networks, which cannot be said for Fernvale.

> I have also worked on getting
> fernly (a small system for experimenting with similar commands like osmoload
> has) working on other mediatek chips like mtk6261, 6735 and 6737.

In other words, operating those MTK chips as general-purpose
microprocessors without any cellular radio functionality, right?  What
you are doing here would indeed be a necessary foundational step *if*
you had some *realistic* plan for how you are going to bring up
cellular radio functionality, but it does not sound like you have any
such realistic plan.

> So at this point I have some work that others did to transmit a burst on a
> specific ARFCN.

What you are describing sounds like a low-level RF test mode,
transmitting a burst by itself, without any synchronization with a BTS,
a test mode provided in most GSM modem implementations (FreeCalypso
has it too) for exercising the RF Tx hardware by itself, without a GSM
network connection - this test mode is typically used in factory
production line processes when doing Tx calibration.

> As far as I know there is no direct knowledge/sources for MTK's DSP but I
> haven't quite gotten to the point where I needed to know just yet.

This is the part that shows most clearly how out of touch you are with
reality when it comes to your idea of OBB on MTK.  The DSP is the most
essential part, and if you don't have that, then everything else is
useless, plain and simple - so if you are investing work in other parts
of the system without having the DSP firmly under control first, then
you are simply spinning your wheels and doing completely wasted work.

> This is a very part-time endeavor for me so I do what I can.

I am part-time too, I am NOT independently wealthy to where I could
work on FreeCalypso full-time.  Yet having a limited time budget is
not an excuse for making poor judgments as to what is realistically
doable and what isn't, deluding yourself and misleading others into a
land of false hopes.

If you really wish to have a shot at working GSM functionality on an
MTK platform, the only way you might be able to achieve something
functional would be if you start with MTK's official firmware:

ftp://ftp.freecalypso.org/pub/GSM/MTK/

Please note that I am acting only as an archivist and nothing more by
hosting that ware on my FTP site.  The contributor who sent it to me
back in 2015 said that it was supposedly MTK's reference fw for the
MT6260 chip, but that is all I know - I don't even know if the
contributor's impression was correct or not - but it is the *only*
available starting point for those chips, to the best of my knowledge.
You would have to spend oodles of time understanding the architecture
of that firmware, how to compile it in different configurations (for
example, how to select whether you need an AT-command-controlled modem
or a self-contained phone handset), how to customize it for different
board-level hardware (board-specific GPIO and multifunction pin
assignments, and if you are doing a handset fw build, how do you select
the screen size and keypad layout and all those details), and then you
would need to try getting it to run on your hardware of interest, i.e.,
SIM800 modules or Fernvale or whatever.

You won't have any realistic chance of bringing up any GSM functionality
on any MTK platform until you get MTK's official fw running on that
platform in a configuration which you compile yourself from semi-src
and thoroughly study that MTK official fw and its architecture, just
like how the people who did OBB on Calypso in 2010 made critical use of
Openmoko's semi-src as a working reference for the Calypso+Iota+Rita
chipset.

Meanwhile, for anyone who needs a working GSM+GPRS modem with full
source code *right now*, my FreeCalypso offering is the only working
solution that is available *today*.  In another few years I will
hopefully also have a self-contained "dumbphone" handset product to
complement the already-available modem offering, whereas with the
approach taken by Craig, it is highly unlikely that his approach will
result in any kind of usable product (be it a phone or a modem) any
sooner than 2035 at the earliest, and would probably be more like
year 2050.

M~


More information about the baseband-devel mailing list