external wireless antenna?

[bad_gui]

I got a guruplug server plus to replace a dead 802.11g wireless bridge and to use
as a NAS for some external USB drives.

I opened the case and am waiting for some heatsinks before setting up. 

I see (on the PCB and FCC site schematics) that J1 is a connector to the Azurewave chip.
I live in a crowded neighborhood and would like to use my Hawking directional antenna
to keep interference down.  My previous Linksys bridge kept dropping the connection
until I replaced the default omni directional antenna with a directional one.

My dead Linksys has a cable to connect an external RP-SMA jack to the PCB, but the guruplug
PCB connector is slightly different and won't accept the wire.  Here is a link with pictures
of what these look like.  The picture quality is low so I can't tell which one is in which device.
http://www.rosenberger.de/PDF/Publications/10_Communication/PCB_Flyer_2010.pdf


A couple of questions:  is the J1 connector for 802.11g?  what is the device labeled "ANT 1"
on the PCB (is it for Bluetooth?)  what kind of connector is J1 and can I get a cable to
connect my antenna?

I will move the power supply to accomodate the heatsinks and fan so an extra hole in the
case isn't a problem

[mjoconnell]

Did you have any success with an external antenna??

J1 looks like an MMCX connector. Can anyone confirm?

[mjoconnell]

Hmm, upon more inspection J1 could be a u.fl connector.

[bad_gui]

I agree it's u.fl but the cable from my dead wet54g doesn't connect with the guruplug
pcb connector.  Similar appearance but not exactly the same.

I'm not sure my soldering iron is fine enough to connect wires to the pcb pads.

I have been using my plug as a wep client and I get the occasional hangs reported
by others on the forum.  The current location is a bedroom above the AP.

Code:

guruplug:~# iwconfig wlan0
wlan0     IEEE 802.11b/g  ESSID:"home"  Nickname:"plug"
          Mode:Managed  Frequency:2.437 GHz  Access Point: XX-XX-XX-XX-XX-XX 
          Bit Rate:1 Mb/s   Tx-Power=13 dBm   
          Retry short limit:8   RTS thr=2347 B   Fragment thr=2346 B   
          Encryption key:XX-XX-XX-XX-XX   Security mode:open
          Power Management:off
          Link Quality=30/100  Signal level=-72 dBm  Noise level=-87 dBm
          Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0
          Tx excessive retries:66  Invalid misc:160   Missed beacon:0

Compare the plug stats (1Mb/s and 30/100 connection) with my desktop computer
right next to the plug that has an external antenna (that came with the PCI card).

Code:

root:# iwconfig wlan0
wlan0     IEEE 802.11g  ESSID:"home"  Nickname:"desktop"
          Mode:Managed  Frequency:2.437 GHz  Access Point: XX-XX-XX-XX-XX   
          Bit Rate=54 Mb/s   Tx-Power:20 dBm   Sensitivity=-109 dBm 
          RTS thr=2347 B   Fragment thr=2346 B   
          Encryption key:XX-XX-XX-XX-XX   Security mode:open
          Power Management:off
          Link Quality:71/100  Signal level:-50 dBm  Noise level:-96 dBm
          Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0
          Tx excessive retries:0  Invalid misc:0   Missed beacon:0

My plug has been modified.  I put copper heatsinks on the chips and removed the
power supply.  I have a 40mm 5V fan attached to the plug pulling air through a 35 mm
hole I cut on the opposite side from the ethernet/usb ports.  I am using a spare external
3A 5V Netgear powersupply.  I moved the pcb wires from the original internal plug
power supply to a jack that I attached to the plug.

I don't have a gigabit switch but the copper heatsinks are room temp.  I still get
system freezes due to the fact that it is in client mode.

[guidol]

I don't have a gigabit switch but the copper heatsinks are room temp.  I still get
system freezes due to the fact that it is in client mode.

Do you think thats the software issue (bad driver) of the client mode like in the following thread?:
http://www.newit.co.uk/forum/index.php/topic,455.msg3208.html#msg3208

[bad_gui]

I think the freezes are due to the wlan client mode.  I have had it freeze while downloading deb packages
although it has been stable for days at a time.  So far I only use it for downloading podcasts with a cron job.

I have the same kernel 2.6.32-00007-g56678ec #1 PREEMPT as the author of the other posting. 

What is the least painful way to cross compile the wireless driver for the plug? 
http://www.plugcomputer.org/plugwiki/index.php/Compiling_Linux_Kernel_for_the_Plug_Computer

My desktop linux box runs a slackware based distro with an older kernel 2.6.22.14 #1 SMP PREEMPT.
After the cross compiler, I download the sheeva/plug kernel source for 2.6.32 and build against it?

I wouldn't mind testing a binary driver compiled (by someone else) for the default kernel on the plug.

Off topic:  Why did Globalscale/Marvell provide a third party site for the Libertas driver?  It is some web design
firm that does very impressive work http://www.inovo.ca/portfolio/index.php  Scroll through some of
the pages they designed for clients.  Hopefully their kernel drivers are as good as their artwork.

[bad_gui]

Ooops.  The author of the post posted the driver sources on his own site.  Got it.
Does this mean this is a beta pre-release?   Should I wait and see if they roll-out
a fully tested and approved version? 

[mjoconnell]

I agree it's u.fl but the cable from my dead wet54g doesn't connect with the guruplug
pcb connector.  Similar appearance but not exactly the same.

Yeah, I had a similar experience from a spare u.fl antenna cable from a dead router.

I have some new parts in route from digi-key (apparently u.fl is only rated at 30 assembly/disassembly cycles). I'll post my results.

[mjoconnell]

I was able to get a u.fl pigtail coax connected to J1, and have an external antenna mounted.

I'm not seeing the signal gain I would expect.

I've tried configuring the antenna using uapctl

eg:

uapctl sys_cfg_antenna_ctl 0 1 #TX on antenna B
uapctl sys_cfg_antenna_ctl 1 1 #RX on antenna B

also tried boosting TX power

uapctl sys_cfg_tx_power 20 # range in uapctl is 0-20

The commands are accepted, but I don't really see any difference from the default settings, and these settings.

It would be nice to have more documentation.

[bad_gui]

Chip spec say external antenna is supported http://www.marvell.com/products/wireless/8688.pdf

This post confirms that the internal antenna is weak http://www.newit.co.uk/forum/index.php?topic=347.0

The Antenna connector was a query as I guess the wavelength of the 2.4Ghz band is somewhere around 13cm so the internal would be about quarter wavelength at best and I don't suppose it is a folded dipole. With a full wavelength dipole, the ERP from the chipset would probably be sufficient to cover my premises (though whether the heating effect might be a problem in such a confined space). Even if were a runner!

what is the part number from Digikey for the cable?  Is it 590-1014-ND  ADAPTER CABL RPSMA TO U.FL 100MM ?

I have a 15 dBi Hawking directional antenna that made a difference for the Linksys wireless bridge over the default omni whip.

[mjoconnell]

The digi-key part for the pigtail I bought is:

JF1R6-CR3-4I

http://search.digikey.com/scripts/DkSearch/dksus.dll?vendor=0&keywords=jf1r6-cr3-4i

But I've found that that assembly is a bit too short.

[mjoconnell]

On close inspection and comparison to a known u.fl connector, and given the lack of signal I'm experiencing, I'm having doubts that J1 is u.fl.

It would make the most sense for 802.11 that it is, but a known u.fl connector on a PCIe card I have shows a center conductor that extends out in the barrel of the connector. The piece at J1 has a center conductor that appears flat, and it looks like there is a small black inverted cone inside the barrel (looks like the cone on a speaker).

Not sure what J1 is.

[mjoconnell]

Reading the Globalscale Technologies Operational Description documentation filed with the FCC:

3.   2F-J1 is the connector for extern instrument connection and 2F-ANT1 is the Antenna

[flipflip]

I've shown the PCB to the RF guys where I work and they say that the connector is a test point (to inject test signals) and cannot (or: should not) be used to connect an antenna. We then desoldered the chip antenna and soldered a "real" antenna (from an aold WRG-54) instead. It didn't help anything. They suggested (here in my non-RF-guy words) that possibly the trace on the PCB acts as an antenna and not much of the signal actually goes into the connected (soldered) antenna. And also one would have to match the circuit to an other antenna carefully (by adding the right capacitors to the empty spaces along the antenna trace on the PCB). Unfortunately we didn't have time to investigate this any further.

I've previously also used an increased tx power setting using the config utility. That indeed improved the wireless connection for some devices (e.g. a wifi-enabled mobile phone) but didn't for others (media player with a "real" antenna). The problem with it was that the plug frequently (every other day or so) reset. That problem has gone since I reverted to the default setting. Or at least I think that the increased tx power was the problem.

[bad_gui]

Thanks for the information.  I found this on the web and it seems there is no easy solution to
improve wireless signal for the plug

http://colinkarpfinger.com/blog/2010/the-dropouts-guide-to-antenna-design/