I have spent some time looking around and found that you can purchase a 24dbi 2.4Ghz grid antenna from Walmart online of all places (there's a first) They will deliver to the nearest store for free and the total cost here in AZ with our tax is $50.81. Not bad for new equipment with a warranty. I have since ordered the antenna and should be here in a few days. I thought letting everyone else know about this source may be helpful. The price is $46.57 before tax.

N7SKV Brent

 TP-Link TL-ANT2424B 2.4GHz 24DBi Grid Parabolic Antenna 

Ebay has some 25db yagis for about 25 plus s&h....from china so it might be a gamble. I saw a few grids too.

$35.45 b4 s/h and tax

http://www.provantage.com/scripts/cart.dll/x/0/continue/TPLK01Y

That one comes out a few cents more to AZ  with s/h so another good deal. Good find...May be cheaper to other states.

K7CQX was over here the other day with a Chinese Yagi we were testing. It didn't look bad but we were not getting very encouraging results. We were having a difficult time seeing the SSID of a Dunkin Donuts 2 miles down the road that the grid easily picks up even when horizontal. I think it was the length of the coax (they put about 6 feet on it) He ordered a grid and is gonna shorten the coax on the yagi. Maybe he will end up with 2 good antennas.

I ordered a grid last night. Now I need a good omni...any good ideas? I just got an email from the shipper. Out of stuck. They hope to ship within 30 days.

I have the Walmart Grid ant it is a TP-Link. Yes you can change the polarization of the Driven element (but it needs to be a certain way according to the dish) so you should actually change the mount point to make it go horizontal. The cable is about 18 inches exposed from the back of the grid. Looks to be LMR 400/600 range cable. I am picking up a standard Wifi (Dunkin Donuts)indoor Vertically polarized 2.2 miles away while I am horizontal. I live in an open area so this may be optimum conditions. My avatar shows the Walmart grid in set up for horizontal.

I’ve done a little testing with 2 inexpensive omnidirectional antennas purchased via eBay.  One is 37 inches long and is encased in white fiberglass.  It is advertised as 12dBi.  See http://www.ebay.com/itm/321100508704?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1497.l2649.  I connected it to the mesh node with an 18 inch pigtail made from RFC 195 cable.  The other antenna is 15 inches long and black; it looks like a thicker, longer version of the Linksys stock antennas.  It is advertised as 15dBi.  See http://www.ebay.com/itm/160524130348?ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1497.l2649.  It connects directly to the Linksys antenna jack just like the stock antennas.

I don’t have any fancy test equipment.  My test procedure was:

1. Attach the antenna to the right side antenna connector.  Turn the node on and wait for it to finish its startup sequence.
2. Set the mesh node to transmit and receive on the right antenna only.
3. Go to the WiFi Scan page and turn on Auto mode.
4. Watch for 2 minutes and record the best reading for each of several WiFi access points in the neighborhood.  There are no other mesh nodes within range of these antennas, so I could only look at signals from regular WiFi access points.

My results were mixed.  The white antenna provided the best results for some access points, and the black antenna provided the best results for others.  Neither was more than 3 db better than the stock antenna.  I suspect my test method was flawed.  Any suggestions on how I could perform some more meaningful tests without any expensive equipment?

Consumer wifi antenna gains are usually rated by the manufacturer with a Mark I Fantasometer.   i.e. the numbers are often fantasy, if not outright fraud.

Also propagation in a cluttered environment through walls, etc. is very variable.  Think of old style TV rabbit ears and how you had to just tinker.

If you can get another HSMM router and experiment with it in an open space somewhere, you might get much more meaningful results. 

Also, do some of the routers dynamically increase or decrease their radiated power based on the signal strength?

I finally found some time today to do some comparisons between directional antennas. We didn't have enough time to test my omnidirectionals, so I'll have to test those another day. There was a light rain/snow falling, so that may have affected the results. But the intensity of the precipitation remained pretty consistent through all my tests, so hopefully the results are valid in comparison to each other.

For the first round of tests the nodes were 2.9 miles apart in a mostly rural area. There were some houses between them but not nearly as many as in a suburban area. And both nodes were high enough to have a clear line of sight. One end used a cantenna made from a soup can. At the other end I tried several antennas and got these SNR ratios:

Cantenna: 5
MFJ-1800 yagi: 9
Hyperlink 19dbi parabolic grid: 12
Hyperlink 19dbi parabolic grid + 500mw amplifier: 25

We noticed that the MFJ-1800 was much more sensitive to being aimed in exactly the right direction than the other 2. But it had significantly less wind load than the parabolic grid.

We then drove to a spot that was 5.9 miles from the first node. There was now a small town between the nodes but most of the area was still rural. It was still half raining, half snowing lightly and the wind was blowing harder. We had an unobstructed line of sight back to the first node. My fingers were getting cold enough that I didn't stay long enough to try all the antennas. I measured these SNRs:

Hyperlink 19dbi parabolic grid: 10
Hyperlink 19dbi parabolic grid + 500mw amplifier: 21

I was pleased with these results given that one end of the connection was just using a soup can antenna and that we had light precipitation attenuating the signals. I was surprised that the 500mw amplifier on just one end of the connection made as big a difference as it did; I had expected it to add half as many db as it did.

I want to thank KC7OY for standing out in the snow with me and helping to conduct these tests. It would have taken much longer if I'd done them alone. I'm planning to do another round of tests in an area where I can have a 20 mile unobstructed view. But I've got some travel and other events coming up, so it may be a month or two before I get a chance to conduct those tests. I'll report back once I do.