On 11 Jun 2005 04:32:44 -0700,
(E-Mail Removed) (Dusan Sukovic)
wrote:
>The things i've read about amplifiers maked general impression on me
>that they are introducing more problems then they solve. Is that
>correct? How about using amplifier in the town with a lots of concrete
>buildings together with the sector antenna (equipment would be a 3
>hyperlinktech sector antenna HG2417P-120° 17 Dbi + 3 Cisco
>AIR-AP352E2R Access Points)? Is the amplifier only good to really long
>distance point-to-point links, and as compensation for cable losses?
>
>url for the antenna:
>
>http://www.hyperlinktech.com/web/hg2417p_120.php
Amplifiers are generally a waste of time and effort for such a system.
The Cisco 350 series xmits 100mw (+20dB) which should be enough.
The only place where an amplifier is necessary is to eliminate losses
created by long coax cable runs. I was involved with a WISP system
that had two Teletronics 1 watt amplifiers on a mast to cover about a
1/2 mile radius. The problem was that the radios were on PCI cards on
the ground floor. There were two floors and the mast was 30ft high.
About 80ft of LMR-400 and LMR-600 coax cable. To compensate for the
coax loss, I used an amplifier. I selected one that had AGC
(automatic gain control) because I had no easy way to insure that the
drive levels to the power amplifier input was within the correct
range. A fixed gain amplifier would not have worked.
We immediately ran into problems. You could hear the access points
anywhere in the area, but trying to get a reliable connection was a
problem. I could improve the situation by tinkering with preamble
length, flow control, fragmentation, and such. When I dragged out the
test equipment, the problem was obvious. The switching time between
transmit and receive on the amplifiers was excessive. It was fine for
1 and 2 Mbits/sec with a long preamble, but too slow for 11Mbits/sec
with a short preamble. The receiver was also comatose for a few msec
when going from transmit to receive, which clobbered part of the short
preamble of a returned signal. Xmit linearity at 1 watt didn't look
all that great resulting in the amplitude portion of the 5.5 and
11Mbit CCK modulation being distorted. 1 and 2 Mbits/sec was all FM
with no AM component so those worked just fine. I decided to take the
short cut, tweaked the amplifier a bit for lower power and better
linearity, and fixed the speed at 2Mbits/sec.
My opinion of the amplifiers went further downhill from there. Water
is the sworn enemy of 2.4GHz. All of the outdoor amplifier do their
best to be waterproof. All I've looked at fail or should fail. (I
used to design marine radios for Intech Inc). Constant failures of
the amplifiers immediately after winter rains was traced to water
incursion. I just drilled a small hole in the bottom of the case,
added a power resistor heater to raise the dew point, and sprayed the
boards with Humiseal (polyurethane waterproofing). End of water
problems.
Enough with the horror stories and on to theory vs practice. The
problem with tx amplfiers is that they are asymmetrical in a link. If
you had a point to point link, and installed an ampflier at only one
end, then your OVERALL range would not improve. Such arrangements
create an alligator, an animal with a big mouth and small ears.
Everyone can hear the amplified central access point, but the access
point can't hear any replies from the clients because they are still
running with the usual +15dBm xmit power.
Worse, since the transmitter now carries well beyond the useable
coverage area, it effively creates a jammer capeable of interfering
with neighboring systems in areas that it cannot effectively
communicate. Such things seem to be common in mesh networks (i.e.
Tropos) that use 1 watt radios at all nodes. If you want to improve
range with tx amplification, then it must be done at both ends of a
link. That also applies to point to multipoint type of WISP systems.
There are WISP's that supply amplifiers to their long distance
clients. For such systems, a similar amplfier at the central access
point would work just fine. But not at one end only.
Incidentally, all new FCC approved communications schemes authorized
in the last 10 years have required automatic transmitter power
adjustment. The idea is to only use as much tx power as required to
maintain a useable S/N ratio at the receiving end. Cell phones work
this way. FCC 15.247 appeared before the FCC recognized the problem
with overpowered xmitters.
I don't want to get into determining if your selection of antenna and
access points are proper for the intended purpose. I would need to
see the topology, layout, intended coverage area, building
construction, folliage, weather, and local aesthetics restrictions, to
determine the appropriate selection. Way too many factors to pass
judgement here. However, I can tell you what to watch for.
Mention of concrete buildings implies that you're either trying to go
through the concrete, deal with reflections from the concrete
buildings, or attempt to do NLOS (non-line of sight) communcations.
All will fail. 2.4GHz does not go through anything with water inside
and concrete is about 30% water. Reflections are a serious problem.
At best they will cause deep fades that tend to come and go. NLOS is
in my opinion science fiction. You can always establish a link via a
marginal path, through whatever obstructions happen to be in the way.
What you cannot do is maintain that link in a reliable manner.
Commerical microwave links and dual antenna access points use spacial
diversity to overcome such path problems. If the path to one antenna
craps out, chances are good that a nearby antenna will still have a
useable path.
Incidentally, it's very easy to tell if you're going to have a
problem. First calculate the fade margin. Absolute minimum should be
10dB of margin. Then, measure it by shoving a 10dB attenuator in line
with the client antenna. The system should still function. If it
craps out, then try smaller levels of attenuation. Given the measured
fade margin, I can estimate the number of support phone calls you'll
get per year from that client.
The general solution to the amplfier (problem) is to locate the radio
next to the antenna. That's the right way to do it, but has a
problem. It increases the number of certified tower climber service
calls that are required. However, on towers over perhaps 100ft high,
long coax runs with amplifiers become impractical and you're forced to
put electronics on the tower. Intermod, overload, blocking, and spurs
from other users on the tower are a major problem. Same with
lightning damage. If you have any electronics on the tower, you
should have at least one working spare.
Selection of antennas also tend to create problems. I'm glad you
selected sector antenna instead of an omni. Omnis are evil. The
vertical radiation pattern of your proposed sector is 13 degrees.
That's perfect for covering a large town area from a central tower.
However, a similar size omni will have perhaps 4 degrees of vertical
beamwidth, with possibly only a little downtilt. You could easily
send the bulk of your signal over everyone's heads. It gets really
bad if you're in hilly or mountainous terrain, where the signal has to
follow the terrain. Omnis can't do that. Panels and sectors can be
adjusted accordingly. For sectors, I use:
http://www.superpass.com/2400-2483M.html
Anyway, good luck with whatever you're plotting. Do a site survey
with a spectrum analyzer to see what interference you'll be getting.
There's always some junk around. For better answers in WISP service,
try the mailing list at:
http://isp-wireless.com
>I've deserved your sarcasm, I should write that question better. Hope
>this one is better? Personally I appreciate your postings in this
>newsgroup.
Yes, your revised question is much better.
--
Jeff Liebermann
(E-Mail Removed)
150 Felker St #D
http://www.LearnByDestroying.com
Santa Cruz CA 95060 AE6KS 831-336-2558
Similar Threads
Linear Mode
