Hello ladies and gentlemen,
A little under a year ago Mr. Sean Moss-Pultz released Openmoko's
original PCB design files for GTA02 revisions A1 through A6:
Fast-forwarding to the present, I am now working together with an
anonymous company in a non-Western, non-1st-world region, and we are
jointly producing a new Calypso modem board that is based on a direct
reuse of the modem section from GTA02_MB_A6.
The name of the project is FCDEV3B, which stands for FreeCalypso
development board, triband. Yes, it is triband like the GTA02 - while
I still desire to go to quadband eventually, the change was deemed to
be too risky for our first board, hence we are playing it safe and
keeping the proven-working GSM RF section completely intact from the
GTA02, which necessarily includes keeping the triband RFFE.
The board will be 90x50 mm in size, and functionally it will consist
of a bare GSM modem with a bunch of interfaces brought out to
connectors. A power input connector, an SMA connector for the GSM
antenna, headers bringing out both UARTs, JTAG and MCSI at 2.8V logic
levels. A SIM socket, PWON and RESET pushbuttons, and analogue
loudspeaker and microphone circuits following TI's reference schematics
will be on-board.
We have made one change to the design which we hope shall fix the
infamous bug #1024. Zero-ohm 0402 jumper-resistors at reference
designators R1003 and R1004 have been removed, and the power traces
have been connected straight through where these 0402 R footprints
used to be. TI's reference schematics (unfortunately TI's reference
layout appears to have been lost) prescribe star routing for the V-DBB
power net, and they have STAR_CONFIG "pseudo-component" symbols in the
two branches, but turning these "star points" into physical 0R jumper-
resistors must have been FIC/Openmoko's idea. I believe that the
latter change contributed to bug #1024; with our change that
effectively reverts FIC/Om's change, the layout we have now appears to
match TI's intent as expressed on their Leonardo reference schematics,
hence I hope that the change we've made will be an improvement.
We are also going to populate a 22 uF cap at C214 (Leonardo refdes
corresponding to Om's C1009) instead of the originally-called-for
10 uF. The hw rework for bug #1024 on Openmoko-made devices consisted
of likewise changing this cap to a higher value, or adding a second
10 uF cap in parallel, hence I hope that even if the bug does not get
fixed with our layout change, it will nonetheless be nipped in the bud
by our use of a larger capacitor.
But please feel free to look at our design files and judge for
The latest revision as of right now is the one with 20160220 in the
filenames. It still has a few outstanding issues which we are going
to fix, so it isn't 100% final yet, but the day when we declare our
design to be finished and send it out to PCB fab is approaching very
soon, hence I decided to give this list an opportunity to review our
work, particularly my thoughts regarding bug #1024 and my approach to
fixing it, before it is too late.
P.S. When we get this board built and working, we are going to offer
them for sale to anyone who wants one on a retail basis, but we also
freely publish all of our design files, hence anyone else is free to
run their own production based on our design.
hardware mailing list
|Free forum by Nabble||Edit this page|