Order Tray | Contact Us | Home | SIG Lists

[Ham-80211] Followup on 4.4 mile link problem posted May 2004

w3sz w3sz at comcast.net
Wed Jan 19 17:53:08 UTC 2005

Hello, All,

This is just a follow up to a post I made in May 2004, to relate the  
solution of the problem.  I am hoping that it will help someone like those  
of you on the list helped me!  I apologize if it it too long.

Last May I had posted to the list with a query and got very many helpful  
The gist of things was that I had just at that time set up a wireless link  
two properties that are 4.4 [4 and 4 tenths] miles apart.  Both are on  
and there is line of sight between the properties visually and confirmed  
by profiling the
direct path between the properties using Topo USA,  except for for  
possibly a large
oak tree that I can't remove at one end.

I used D-Link DWL 2100 AP's in Point to Point Bridge mode at both ends.
Each unit was mounted in a weatherproof box at the back of a barbeque grill
24 dBi parabolic antenna set up for horizontal polarization.

I found that after optimizing things I could only get a thruput of 400Kbps  
or so, and I could only do that by blasting things in at 36 Mbps or so and  
getting retry rates in the 97% range or so.  My link budget said that I  
should be OK [17.9 dB thermal fade margin], but obviously I wasn't OK.

At that time many helpful suggestions were offered:

go from auto to fixed speed;  that helped a lot [I was at 36 Kbps before I  
did that one, and this gave me an increase of 10x to about 400 Kbps].

do a site survey;  I couldn't do one on the tower, where the Bridges were,  
but with a site survey done down below there was no interference.  I  
realize that doesn't mean a lot :)

make sure it's really line of site by climbing up the tower [but its a  
crankup so that is not recommended] or putting a camera up there  
[basically to see how much the tree obstructs things].  Rather than do  
either of those, I decided to just wait and see if there was any change  
 from full leaf growth on a wet summer day to dry autumn leaves to winter  
with no leaves.  I kept written records of a lot of thruput checks, and  
there was NO change at all in thruput, so I don't think the tree was  
obstructing [its an oak with skimpy branches, but LOTS of leaves in the  

increase power.  I did this, but not in a scienfically 'clean' fashion.  I  
did it more cheaply, by getting close-out bridges to replace the D-Links  
that gave me an increase in power from 32 to 200 mw [see below].

increase height to avoid fresnel zones [moving the height of the crankup  
up and down made no difference once I was above about 40-50 feet, which is  
where I had started my tests].

use the most recent firmware [I did].

rotate the antennas to try vertical polarity [couldn't do this easily so I  

I tried all of the channels, and the performance numbers given are for the  
the best one.

More than one response suggested that 'D-Link are $#@)!...get another  
radio.  I did.  More on that below.

Align the antennas [did this and checked and rechecked and took thruput  
readings at various azimuth angles before posting to the list initially].   
Obviously an important thing to do.  Much easier when there is line of  
sight ;)

Use a helical antenna at each end to cut down on multipath.  I got two  
 from Directive Systems.  They look great and I was planning to use them,  
but don't need to, it turns out.  I will play with them at some point.

SO Here is what I actually did.

1.  I waited for winter.  No in thruput difference with leaves off of  
tree.  I decided that tree not likely to be the problem.

2.  Today I had braver, younger souls go up on tower and move network box  
and barbeque grill antenna in closer to the tower so I could work on them  
without climbing out on boom.  They didn't mind the snow too much ;)

3.  Then put Senao Wireless Multiclient Bridge at each end of link [to  
replace the D-Link units], running in Point-to-point bridge mode. set to  
11 Mbps fixed speed.

With this I get consistent 3.5-4 Mbps thruput with the tower all the way  
down! [checked using QCheck].

I haven't climbed it yet to see if I have line of sight [it is snowing  
pretty heavily now], but with the tower down I am already getting 10 times  
the thruput with the Senao that I got with the D-Link.  I suspect that  
when summer comes I will need to put the tower up.  I will do an  
experiment when it stops snowing to see if I can get the thruput higher by  
raising the tower.

The D-Link transmitter put out 15 dBm [32mw], while the Senao puts out 23  
dBm [200mw].  The Senao was a white box closeout that actually cost less  
than the D-Link.  I think it is ths NL-2511 CB3-PLUS [I didn't write down  
the number before I put the units on the tower, and it turns out that none  
of the documentation gives the number.  I am sure that it is 2511 and NOT  
2611, but am less sure of the CB3-PLUS part.  When it stops snowing I will  
climb the tower and check this].  With the D-Link I calculated a thermal  
fade margin of 17.9 dB.  With the Senao it is 25.9 dB.  So the improvement  
could just be the power increase, and not an inherent superiority of the  
Senao, I guess [comments welcome on this point].

Interestingly, the D-Link and the Senao wouldn't communicate at all in  
bridge mode even when set to 'identical' parameters.  I had read somewhere  
that some of the manufacturers used proprietary datastreams for bridge  
mode, so this didn't surprise me too much.  But the knowledge may help  
someone else on the list.

SO to summarize,

1.  the tree was not the problem

2.  there were no other path obstructions

3.  exchanging the 32 mw D-Link for the 200 mw Senao increased thruput by  

While I'd like even more thruput for some of my projects, I will be happy  
with this for now.

My next step would probably be to add bidirectional amps at both ends, but  
I will wait for the ARRL and the HSMM group to get things clarified with  
the FCC regarding encryption, controlled access, automatic power control,  
etc before I do that, and I will only do that if I NEED the extra thruput.

Thanks again to all,


Roger Rehr

Roger Rehr

More information about the ham-80211 mailing list