I'm a member of the DCARA here in Denton,TX (North TX). We're setting up nodes around the county and I'm trying and still planning to link with our "Mesh Hub" node at 16km distant. Radio Mobile calculates a -76dBm RX level which seems marginal looking at the noise floor. So, I would like to understand the internal switching topology within the Rocket M2 of the chain ports. Do I understand correctly that the ports are power divided in diversity mode, therefore a -3.5dB loss? If so, appears I should use a separate 23dBi grid parabolic on each chain and configure one for RX and the other for TX, effectively dedicating 1 rocket for this long haul link instead of the 12dBi Omni on chain1 and the single 23dBi grid on chain1 aimed to the node target. Sounds like I could gain another 3+ dB. However, again without understanding the internal switching topology configuration within the Rocket, this is just an educated guess.

Any advice help here would be appreciated, as I can act as a long haul link to my closer neighbor K5WXR with which I have good link with by using another Rocket with sectionals.

Thanks in advance & 73,

Bob AF5RS

My Laptop tries to do an identify network when I connect to a new node.  It usually needs a minute or so to do this and it blocks the laptop from connecting to the node until it is finished.  I can tell it is finished by the icon in the lower right for the network changing.  Clearing out the old network identifications seemed to help.

The other thing I recommend when setting up a node is to turn off/disable the WIFI  connection on the Laptop.  They compete against each other and cause problems.

For UBNT devices:

Set a Static ip for the network port on your computer to an address in the 192.168.1.xxx space, where xxx does not equal 1, 20 or 255.  I normally use 100.  That seems to help with the network id. 

Plug in an Ethernet cable between your computer and the ubnt device.  Wait for the computer to id the network.  It may tell you that you don't have an connection to the internet.  That's fine.

Then log into the ubnt device, admin screen, update firmware.

Once the node has rebooted, I usually go back into the lan port and change it to get assignment automatically.  This normally is not necessary since the ubnt is in access mode and uses 192.168.1.xxx addresses.  Make sure you set your DNS to automatically assign as well.

Wait for the network to id and then do the localnode:8080 to set it up.  Some computers require a fully qualified domain name so if localnode:8080 doesn't work try localnode.local.mesh:8080.  You may want to try 192.168.1.1:8080 as well.

If that doesn't work then unplug the network cable, turn on the wifi connection and look for an SSID called meshnode.  Connect to it and then try the localnode:8080 or localnode.local.mesh:8080

I hate to say it but if all that fails your best shot is to use a computer with linux on it and follow the same steps.  The network id is not a problem in Linux.

I also believe you will find you want to be on the latest release.  Most people will be updating and 1.1.2 does not talk to the 1.0.1 devices.  Even you don't connect to anyone else the 1.1.2 fixes some security bugs that you want fixed.

Clint, AE5CA

I have seen this before.  What I came up with was both sides need to be hearing each other to get the handshaking that goes on within OLSR.  The far end node is not seeing you that well.

I had two Linksys boxes, separated by 8 miles.  Box 1 was a 24 db grid antenna, Box 2 was a 13 db Horizontal Polarized omni.  I could see the omni but it was not seeing me all that well.  I would just get the ip address to show up.  My experience is the receive side of a Linksys is not all that good.

I fixed this problem by changing out the Linksys attached to the omni with a Bullet M2HP, and then swapped the grid antenna/linksys for a Rocket M2 and a Sector antenna similar to yours.  Problem Solved.

We have another location with the other problem similar to yours.  We counted 31 ssid's on channel 1.  Our intention is to try and solve this by switching to 5.8 GHz.  Beam antennas may also help to eliminate some of the other strong signals.  If 3.4 GHz was available I would suggest that instead.

Our experience with amps is don't bother.  Signal path is the main problem.  If you have a good path you won't need the amp.

Clint, AE5CA

Ticket 48 was written to address this issue.  It looks like it is still present in 1.1.2. 

I can confirm that I have been able to forward port 80 to the wan using a Rocket M2 and a Netgear GS105E or GS108E Ethernet switch with the switch programmed to tag the WAN Connection as VLAN 1 before continuing on to the Rocket. 

My friend that had the original problem has since switched to a NanoStation M2 and he is also able to port forward to Port 80.

I believe this is just a matter of adding the right line in iptables but I am not 100% sure of the command. 

Clint, AE5CA

After reading through the information, it states that the whole section is for amateur satellite use, and if there are no satellite operators in the area to interfere with, then the bands can be used for amateur TV. It doesn't state that data equipment can be used.

I know that the 2.4GHz spectrum is licence free and channel 1 is within that allocation, but is it still legal if the system is used as the mesh instead of a home wireless point?

Jack.

Greetings All,

My name is Chris (K2CTC) and I just purchased a WRT54GL to configure as a HSMM-MESH node.  The device has not arrived yet, but I am excited to play with the technology.  I would appear that this forum is the place to be if you are in Utah and interested in mesh.  I will post an update once I get things up and running.

I am curious,

Why are you rebooting the node if internet is lost ?

Nodes can still be very useful even if they don't have internet, putting it in a reboot loop every 5 minutes would degrade the network, and if you have an entire area internet outage and this is on lots of nodes you could take your entire mesh network down at once as all the nodes going into constant reboot cycles (a cascade failure)

Nevermind, found a persistent storage in /jffs/usr/bin/ and storred the script there. if anyone is interested, the i used this script to get the job done:

#/bin/sh

count=$((`ping -c 3 8.8.8.8 | grep -c '64 bytes'`))

if [ $count -le 1 ]

then

  `echo "timeout" >> /tmp/network_watchdog` # add timeout to watchdog

else

  `> /tmp/network_watchdog` # reset timeouts

fi

count=$((`cat /tmp/network_watchdog | grep -c "timeout"`)) # count occurred timeouts

if [ $count -ge 5 ]

then '/sbin/reboot` #reboot if more than 5 concurrent timeouts

fi

That's a high noise level.  I would definitely try horizontal polarization.  On the multipath question, you need to be sure the Freznel Zone is clear (see http://en.wikipedia.org/wiki/Fresnel_zone).  You can estimate the required clearance using the formula: 36*sqrt(miles/GHz)=clearance in feet.  Using 15 miles and 2.4GHz, that results in a clearance requirement of about 90' above and below the beam at the midpoint between the two nodes:

36*sqrt(15/2.4)=

36*sqrt(6.25)=

36*2.5=90’

So check to be sure that area isn't obstructed in any way.  However, again I'll suggest Ubiquiti gear in a scenario such as this.  You will gain nearly 10dB in receiver sensitivity (which will help when you get the noise issue resolved) and nearly 10dB in higher power output.  I don't have any experience in circular polarization.

Andre, K6AH

Hi All,

I have tried to install a script to check internet connectivity, and reboot a remote node if the connectivity has been gone for more than five minutes, this script was supposed to be executed by cron every minute. Where do I store the script on the hsmm firmware image to persist between reboots?

73 and Best Regards

LA3QSA

Per Øyvind

Thank you both for your replies!

our remote site node is co-located with a 2 meter and 70 cm repeater.  but,  there are lots of commercial transmitters close by.  the router shows a noise floor of ~-70 dBm.

the antenna used on the remote site is a commercially built sector array (HBW=90 degrees, VBW=7.5 degrees, gain ~16 dBi)

we are wondering also if multi-path is a factor. on the node opposite the remote site (located in a park), when lots of people show up, the signal levels really start to vary.

i have read several articles where circular polarization was used to mitigate multi-path. has anyone experimented with circular polarization?

thanks for your help!

N7RTA