GSM development kits: TI's and ours

[Mychaela Falconia]

Hello FC community,

I just performed an experiment I've been wanting to do for a while: I
powered up our D-Sample board again, but this time with both UARTs
connected, as opposed to just one.  TI's D-Sample and other boards in
that family bring their Calypso UARTs out on full DE9F ports at real
RS-232 levels, i.e., they were meant to be used with desktop PCs - the
people who worked on this stuff at TI back in Those Days had the
luxury of this work being their actual job...

If you are like me and use a laptop for development and hacking, with
only USB and no real RS-232 ports, one can use a USB-to-DE9 serial
cable with a built-in USB-serial adapter, but this approach works with
only one UART.  Of course one can use two identical USB-serial cables
of this kind, but I would go nuts trying to keep track of whether
/dev/ttyUSB0 is the cable I connected to the MODEM UART and /dev/ttyUSB1
is the one I connected to the IrDA UART, or if it's the other way
around.

To keep myself sane with the two UARTs in the picture, I bought a
USB-COM232-PLUS2 board from FTDI that provides two serial ports with
one USB device (it's an FT2232H chip), and brings each port out on a
DE9M at real RS-232 levels.  In the absence of other USB-serial
devices, port A on this board is always /dev/ttyUSB0 and port B is
always /dev/ttyUSB1.  Now the two RS-232 serial cables (identical
except for the sticky labels which differentiate them) with plain old
DE9 connectors go between the FTDI board and the D-Sample, with port A
going to the MODEM UART and port B going to the IrDA UART, and I
always know which /dev/ttyUSBx is which UART on the target.

Unfortunately we still don't have the requisite bits to be able to
build our own firmware images for this D-Sample, hence we are limited
to playing with the fw image it came flashed with, dated 20020917 -
quite old compared to Openmoko/Motorola/Pirelli/etc stuff.  But having
both UARTs connected allowed me to send AT commands to this ancient
firmware on the MODEM UART while watching the debug output on the IrDA
UART, and guess what I did: I inserted a real SIM into the SIM socket
on the D-Sample board, connected a real antenna to the GSM RF
connector, and commanded the ancient 20020917 firmware to connect to
the network!

And guess what: it succeeded!  Not only did it succeed in finding and
syncing to a local GSM cell, but the network allowed it to register
despite this MS (mobile station) having no valid IMEI.  This D-Sample
kit came with an almost blank FFS, and in particular, there is no
/pcm/IMEI file in this FFS, so it must be using whatever the pcmdata.c
compiled-in default was back in 2002.  I was expecting it to fail at
the registration step (refusal from the network) until I programmed
some IMEISV which the network would accept as valid, but whatever this
fw transmitted, the network must have swallowed it, as it was able to
both make and receive phone calls!

Zooming out to the bigger picture though, the only really special
potential practical value in the D-Sample kit is the handset part with
the 176x220 pixel LCD and a full set of buttons.  For pure modem
applications (no UI components) our work-in-progress FCDEV3B will
fully supercede TI's D-Sample: we'll have all of the same functionality
that comes with having two UARTs, audio circuits and debug interfaces,
but with a newer version of TI's chipset that is supported by the
TCS211 firmware which is being source-reconstructed as explained in my
previous post.  And we'll be able to produce our FCDEV3B kits in
whatever quantity we get demand for, whereas the D-Sample kit we got
may be the last one in the world.

It would still be nice though, purely for the fun of it, to be able to
compile TCS211 in the D-Sample configuration and see it run with the
full color, 176x220 pixel version of TI's demo/prototype UI, as opposed
to the C-Sample UI we brought up earlier on the C139 - so I'm going to
continue the search for the missing tpudrv10 code.

Hasta la Victoria, Siempre,
Mychaela