Thanks, Charles and Dustin,
The challenge I'm working on is to determine if the degregation of
my test link, is caused by
A) Distortion on the transmitter, at full power? or
B) Overload or lack of acuracy of the receiver. or
C) Or Just plain interference creeping in. (tested at about -80db)
note: multipath unlikely, as LOS link, 10 miles, parabolic antenna,
o wall behind antennas, 100ft above other buildings.
In my Trango test case, w/ 2ft antennas, QAM16, at -55 db I got
worse signal Quality quality (packet loss) than at -65db. To me that
would infer case A or B was happening.
What was interesting, is my Mikrotik test link w/ range5s, actually
got peak rssi (full power) of -47db apposed to Altas's peak signal
of -55.
(note: path analisys calculated -55 db appropriate, so not a
negative for the Trango, but a Plus for the Range 5, exceeding
expectations).
With the Mikrotik, the higher the rssi radio power, the better the
speed results, and lower the packet loss. So Mikrotik did not seem
to be plagued with the same delimna. However, at a surprise, the
Mikrotik performed at a slower speed, and had more packet loss, in
its best link configuration, than Trango had. So the Trango at
-65db QAM16, outperformed the Mikrotik at -47db.
I attribute those results partially, to how the radios deal with
interference. One side of the link (AP/MU) had significant noise,
causing the Mikrotik to lower modulation more frequently. I proved
this, by repeating speed tests with Trango using 5.3Ghz, which
performed perfect links (no loss). However, the 10-11 miles was
pushing the maxrange of 5.3, and I felt 5.3 was to risky, based on
that. I actually had to turnup the Power a little over the legal
limit to get the perfect link, but still lower rssi than the 5.8G
link. But my point was, when noise wasn't there, the links worked
much better.
So the decission I am trying to decide on is,
a) increase the gain (dbi) of the antennas and lower the gain (dbm)
of the radio, to improve the link.
For example, upgrade from 2 ft dished to 3 or 4 ft dished. or
b) get a better 2 ft antenna with more isolation.
For example, upgrade Gabriel cheap 2 ft para to the high performance
2 ft Gabriel Drum style antennas?
Either one could have a possitive effect. Its likely that my noise
is comming from my colocated antennas at the same site. The Drum
style antenna will likely have much better isolation comming from
the sides. Better F/B ratio is not jsut about an antenna behind me,
but also beside me, and interference is not always cured by lowering
the beamwidth, if the interference is comming from the side. So
better isolation antenna could be the choice.
However, if the packet loss was from self generated noise, larger
antenna would keep my gain up, even after lowering power. However, I
actually would still have a gain improvement, because the antenna
increases gain in both directions, where as lowering he TX power
only does it in one direction. Because most of my interference is at
the AP/MU side my paln was possibly to....
Increase the antenna at the RU/Client, to a 3-4 ft dish. If packet
loss at -55db was due to transmitting to high power, and loss was at
MU/AP then it would be most importantto lower transmit power at the
RU/Client side. Increasing dish size at RU would help this.
Then on the MU/AP side, I would add the high performance 2ft
antenna, with better isolation, taking that most of teh interference
may be colocation interference. Increasing the antenna size may not
block interference comming from the side.
But then again, if interference comming from the front (I have
another site 20 deg off to the left), its possible the larger dish
and narrower beam may in fact also help isolate interference.
Now to make it complicated, what if the cause is not interference at
the radio receivers? But instead its all the RF in between and
reflections comming out of phase and distorting my signal before it
gets to my radios?
Now I could just add 4 ft high performance drum antennas on both
sides, and call the problem done, but then that would be $4000 just
in antennas :-( But also means upgrading mounting pole and ballast
hardware.
Which brings me back to my original post, is it just cheaper to buy
better radios, and which have better C/Is and SNR threshholds?
The orthogon has more power, but no matter how good their path anal
tool is, and how good the reputation, when a large part of the
problem is noise floor, I have a hard time believing that faster
speed can be pushed through a narrower spectrum width, while hard
setting speed in each direction.
So what I am learning is its not about whats the best radio, its
what tool do you need to solve each unique problem. The hard part of
this business is conclusively identifying what problem exists, to
know the most cost effective way to solve it.
Decissions, decissions, decissions.
Tom DeReggi
RapidDSL & Wireless, Inc
IntAirNet- Fixed Wireless Broadband
----- Original Message ----- From: "Charles Wu" <[EMAIL PROTECTED]>
To: "'WISPA General List'" <[email protected]>
Sent: Wednesday, February 08, 2006 9:29 AM
Subject: RE: [WISPA] BPSK QAM16 DSSS interference
As you start to walk up the modulation line you definitely need
more C/I,
but you also start to loose the ability to use full power out of the
radio.
A small bit of trivial regarding this issue
With higher order modulation schemes, the EVM (Error Vector
Magnitude) can
be so high that even on a perfect link (no noise) the receive chip is
incable of decoding the signal properly into the correct 64 "dots"
of the
QAM modulation plot.
This QAM constellation "interference" can be represented by a grid
of 8x8
dots that are being blurred by the transmitter not handling the
signals with
enough linearity (e.g., the radio power amp is turned to high).
When too
much blur occurs, the adjacent dots touch each other and the
receiver will
not be able to decipher the signal (it's blurred)
-Charles
-------------------------------------------
WiNOG Austin, TX
March 13-15, 2006
http://www.winog.com
