On 1/27/2017 9:51 PM, Jeff Liebermann wrote:
On Fri, 27 Jan 2017 15:59:18 -0500, rickman wrote:
On 1/27/2017 11:52 AM, Jeff Liebermann wrote:
https://www.ubnt.com/airmax/airmax-yagi-antenna/
https://dl.ubnt.com/datasheets/airmaxyagi/airMAX_900MHz_YAGI_Antenna.pdf
I got a whole new setup. The old unit was one piece with what must have
been a panel antenna as the case was flat and broad. You can see what a
monster this one is. The beam is nearly three feet... no, I just
checked the data sheet and it around 4 feet long! It's not hugely
heavy, but to move it around I have to unbolt the mounting bracket and
it's a PITA while on a ladder.
900 MHz panel (patch) antennas have about 8dBi gain but are fairly
small. The larger variety with 4 patches might squeeze out 12dBi
gain, but will be huge.
Remember, with antennas, the bigger and uglier they are, the better
they work.
But more importantly, I've read that the director elements are
*insulated* from the support beam while these are all welded.
Not sure if you understand me. This Yagi is *not* insulated. But
remember, it has 17 passive elements! It's not like they were just
fooling around. lol
Sorry. I misread your statement. 4ft long is not a big antenna. If
it requires a crane, I would call it a big antenna.
I mean I'm getting some 6-7 dB better signal. Regardless of the SNR,
the receiver provides quality indexes that show a lower bit error rate
and higher overall throughput. That's why they put this up. It hasn't
improved my throughput so much, but it lowers the retransmits and
improves the utilization of their network.
Retransmissions can be a sign of too much noise (low signal),
interference, or collisions with packets from other users. If the
system is heavily subscribed, and your noise level looks about the
same as before, I would suspect collisions. Got any other users
nearby?
I'm not certain whether SNR or signal strength is most important.
SNR is most important. Having a strong signal isn't very useful if
the noise level is as strong as the signal (ignoring spread spectrum
processing gain).
That
would depend on the noise factor of the receiver, no?
Noise figure is part of it. A noisy front end will do as much to bury
a signal with noise, as will interference from other stations.
However, todays receiver front ends are quite good and are not the
horrid noise generators I recall from the vacuum tube days. I would
say that interference is far more important than receiver noise
figure.
I know in lower
frequencies the environmental noise is high enough the receiver noise
nearly doesn't matter.
True. However the problem is a bit different at HF frequencies. The
noise level can be so high, that if the receiver had too much front
end gain, it would overload on the noise alone, producing zero dynamic
range. That's why many HF radios have a 20dB attenuator switch on the
front panel.
At higher frequencies I thought the limitation
was in the receiver front end.
As I mumbled a few paragraphs up, it's a system problem. The
demodulator doesn't care if the noise if thermal, shot noise, or
interference. I can make any of these be predominant by committing
some kind of design screwup, but if the radio is reasonably well
designed, it's interference and collisions with other users packets,
that limits the throughput.
So until the noise gets to be high
enough that it approaches the receiver noise, it won't matter.
If your receiver shows an increase in base line noise level when the
antenna is connected but there's no receive signal, the noise is
higher than the receiver noise.
I'm not going to mess with this setup. The antenna connects to the
receiver with some rubber booted connectors I'm not familiar with and
I'm leaving them alone. So this will have to remain a thought
experiment. But your point above that if the noise from the environment
were below the receiver noise, I wouldn't see it change... however, I
was comparing two units which may well be calibrated differently or
something... too many variables.
What would a spectrum analyzer show me that would be useful. No, I
don't have one, but I could get one... 
You'll be amazed at what you can "see" on the 902-928 band.
Most of this crap will be visible:
http://www.ccrane.com/AM-Antennas?by=Category
However, I think you'll see quite a bit of junk from the utility
Smartmeters. Around here, we had to move the 900MHz ham radio
repeater input frequencies to the bottom of the band, where PG&E has
gratiously left a few MHz unpolluted by their wireless metering
system.
This is what I'd like to see around 60 kHz. If I ever get the thing
built it should have a *lot* of DSP filtering to give it a very narrow
bandwidth. But the question remains if that will be good enough. Joerg
seems to feel that a 1 bit ADC can still be overloaded by out of band
noise. I think the signal will still show up and can be dug from the dirt.
BTW, I think your Nanostation M900 has a built in spectrum analyzer.
Not sure about your unspecified model replacement. I think that would
be the easiest test for intereference.
I listed the new receiver model somewhere, Rocket M900. Actually they
are sold together I believe. The data sheet covers them both anyway.
--
Rick C