mini-pcie.jpg
Here we see the WWAN mini-pcie slot in a CF-19 MK1 (CF-19CJBAXBM). We will henceforth refer to the board containing this slot as the WWAN board (not to be confused with the card you'll insert into this slot if you actually want WWAN, we'll call that a WWAN Card). Panasonic's service manuals irritatingly refer to this as the HSDPA board, despite the fact that even they offer EVDO as an option. You'll note that
  1. There are two coax feedlines entering the compartment connecting to two antenna connectors on the WWAN board. I believe these connectors are here to prevent the coax from flapping around in the case when no WWAN card is installed.
  2. There are four dip switches in the upper right corner, labelled upside down (enlarged in inset).

Well, I can't definitively answer whether the feedlines do what I expect, or what on earth those dip switches do. What I can tell you is that I bought a Novatel U720 USB evdo dongle on ebay for $30, cracked it open, removed the mini-pcie card inside it and installed it into this slot. I set the dip switches to 1, 0, 1, 0 (which is what the pictures in this post seem to indicate -- my WWAN board doesn't have any such silkscreening), pulled the coax off of the connectors on the WWAN board, hooked the black coax to the "main" and the brown to "aux", stole two screws out of an old laptop hard drive to keep it in place, and put it back together. And it works great. Also, I now have not only GPS, but tower assisted GPS, which panasonic's standard offering simply doesn't have.

I'm fairly confident that I haven't accidentally plugged my WWAN card into my bluetooth radio instead of an antenna, because that still works fine. The 802.11 is still working great also. So, I think with this $30 addition I have gotten away with effectively four integrated radios in one small and light (but still fully rugged) toughbook.

Ever since I put the screws in the thing, it seems to have started acting more predictably, so they may have an important grounding function for the WWAN Card. Or I may be imagining it. Or perhaps reinstalling the software helped. Just go find an old 6 gig laptop hard drive and salvage the screws, what'll it hurt? Another thing I may be imagining is that it matters that the black feedline goes to main and the brown to aux, at least as far as having GPS that fixes quickly.

The Sierra Wireless 595U is another USB dongle that contains a mini-pcie card, difference being that Sierra Wireless has documented how to enable the GPS in Linux. My experimentation with these units has shown that you can indeed enable the GPS in the unit without provisioning the modem to a working account, though the performance in this case suggests (as does basic logic) that the GPS is no longer tower assisted in this scenario. Sounds like a lot cheaper option to me than other GPSs, even if you don't want EVDO. Ever since I switched to Linux, seems everything works better, but I can't make an accurate judgement as I switched to linux, the 595U, and an SSD all at the same time.

I got a Sierra Wireless 595U, and I want the GPS automatically enabled at boot in Linux!

Me too, brother. Following is how I did it. This may not be the most efficient way (I could probably do a lot of what the script does in the filter), but it works and I'm not messing with it now. This even works if you flip the wireless switch off and back on.

/lib/udev/595u
#!/usr/bin/perl
 
use Fcntl;
require "flush.pl";
 
$dev = $ENV{'DEVNAME'};
 
if ($ENV{'ID_MODEL'} ne "Sierra_Wireless_AC595U_Device" || $ENV{'ACTION'} ne "add") {
  `/usr/bin/logger -t 595u "notified about wrong device: $dev!"`;
  exit(0);
  }
 
if ($ENV{'ID_PORT'} == 2) {
  `/usr/bin/logger -t 595u "Linked GPS ($dev) to /dev/nmea"`;
  `/bin/rm -f /dev/nmea`;
  `/bin/ln -s $dev /dev/nmea`;
  exit(0);
} elsif ($ENV{'ID_PORT'} != 0) {
  `/usr/bin/logger -t 595u "notified about wrong device: $dev!"`;
  exit(0);
  }
 
`/usr/bin/logger -t 595u "Notified about <$dev>!"`;
 
sysopen(MODEM, $dev, O_NONBLOCK|O_RDWR)
    or die "Can't open modem $dev: $!\n";
 
$BUFSIZ = 0x7fffffff;
$rv = sysread(MODEM, $buffer, $BUFSIZ);
 
# print("> $buffer \n");
`/usr/bin/logger -t 595u "$buffer"`;
 
print MODEM "AT!GPSTRACK=2,60,100,1000,5\r\n";
&flush(MODEM);
 
sleep 1;
$BUFSIZ = 0x7fffffff;
$rv = sysread(MODEM, $buffer, $BUFSIZ);
 
# print("> $buffer \n");
`/usr/bin/logger -t 595u "$buffer"`;
 
 

/lib/udev/rules.d/40-595u.rules
SUBSYSTEM=="tty", ID_MODEL=="Sierra_Wireless_AC595U_Device", ID_PORT=="0", ACTION=="add", RUN+="/lib/udev/595u"

By the way, for your viewing pleasure, here are the relevant antennas.

cf19-wwan-ants.JPG

I note that the EVDO-AUX antenna has a brown feedline, whereas the "US WWAN" antenna has a grey feedline (they're even labelled right on the PCB!). The coax entering the WWAN board hole are brown and black. I believe the WWAN main feedline makes a stop at the docking connector to allow connection to an external antenna. If my assumptions are correct, then it should be noted that the LAN MAIN coax, at its connection point to the antenna, is also black. Psychos!

Hold on a sec! That silkscreen there says WLAN3, doesn't it? I bet these antennas are aftermarket from a later revision with 802.11 N. I might have to try installing the relevant card in this thing! That, of course, will require more effort -- removing the whole underside of the case. The WLAN and BT cards are under the WWAN board, which obviously cannot be removed through this little hole.