GTM900 support update

Mychaela Falconia mychaela.falconia at gmail.com
Thu Jun 13 18:33:07 UTC 2019


Hi DS!

> With regards to price, Songbosi sends those module by shipping with
> DHL, which costs $25 per shipment. So the overhead of buying just one
> $15 piece is a bit high.
>
> Maybe it could make sense, at some point in the future, for you to
> buy a bunch of them in bulk, and then resell with the new breakout
> board you're developing.

Shipping from USA to international destinations also costs around $15-25
per shipment for small shipments containing items of no value (meaning
no big loss if postal services mishandle and lose the package), or
about $50 for larger items and/or valuable ones where I would be using
the more expensive service with tracking and insurance, hence it would
not be any cheaper than having end users buy directly from Songbosi.

It also needs to be remembered that unmodified GTM900 modems can only
be useful to people outside of North America, as they only support
EGSM and DCS bands, no PCS or GSM850.  As discussed here previously,
a hardware mod to change the bands is very doable, but the cost of my
time to do the necessary experiments to find the correct L & C values
and to develop appropriate calibration profiles would bring the cost
to about the same as FCDEV3B.  (And it isn't just the cost of *my*
time either - the fine folks at Technotronix would also need to be
paid to do the physical hw reworks, and they would need to be paid at
USA fair labor rates, not 3rd-world rates.)

> This breakout board could in fact include
> additional hardware such as a FT2232C to connect to both UARTs or
> one UART and one JTAG. What do you think?

I don't plan on putting any USB stuff directly onto the breakout board
itself, instead I am going to bring both UARTs out on a 10-pin header
exactly like on the FCDEV3B, with the same pinout.  This way the
solution remains usable in both USB and non-USB environments - a prime
example of the latter would be connecting to some off-the-shelf
single-board computer with 3.3V UARTs.

I also have a desire to produce a custom USB to dual UART adapter
board that will have an FT2232D chip plus a level shifter to produce
2.8V I/O, specifically for connecting to FreeCalypso dual UART
targets: FCDEV3B, GTM900+breakout, other hw in the future.  Right now
the available options are to either use an off-the-shelf FT2232x
generic breakout board (puts out 3.3V I/O rather than 2.8V), or use
Harald's mv-uart board (CP2105 chip not as good as FT2232x in terms of
support for non-standard baud rates), and neither option is ideal.  My
desired DUART28 adapter board would provide a custom solution
specifically for FreeCalypso.

Connecting Calypso's 2.8V I/O pins to external 3.3V systems: it is
generally OK and TI's cal000_a.pdf document even says that the maximum
allowed voltage on inputs is VDDS+0.5V (I am guessing that there are
clamping diodes inside the chip and that 0.5 V is their forward drop
voltage, which is why up to 0.5 V above VDDS is OK), but at least on
the FCDEV3B we get another unpleasant issue.  Our board features 100k
pull-up resistors to the V-IO rail on RX_IRDA and RX_MODEM inputs so
they won't float if one of the UARTs is unused and unconnected in a
given application, but if an external system feeds 3.3V into either or
both of those UART lines, current flows through that pull-up resistor
in the opposite direction, pulling the V-IO rail itself up from its
nominal 2.8V toward the external 3.3V.  Everything works fine in
practice, but in the long term this issue needs to be rectified, and
making a custom USB-to-dual-UART adapter with 2.8V I/O seems like the
simplest solution to me.

M~


More information about the Community mailing list