Can you help me interpret this spectrum analysis noise plot?

Danny D'Amico wrote in news:pan.2013.12.18.01.48.13 @is.invalid:

only guy who can do this is gone, Jeff Liebermann

--- news://freenews.netfront.net/ - complaints: snipped-for-privacy@netfront.net ---

Jeff is great. He was in Santa Cruz, and frequented alt.internet.wireless before moving mostly on to sci.electronics.repair.

What do you mean he's "gone"? Isn't he still around?

Q: What's the problem. A: It's my understanding that every dB of noise reduces the dB of signal by that amount. The less signal, the slower the Internet speeds.

Q: How much noise do you consider normal? A: I have no idea. That's why I'm asking! :) Googling, I find you "should" have about 20dB of headroom between noise & signal. My signal is about -52dBm and my noise is -88dBm, so I'm within that range, but, my signal to noise ratio is -52dBm - -88dBm = 36

Q: What's the bandwidth & noise figure of your receiver? A: Googling for the "Rocket M2 bandwidth gain specifications", I find this datasheet for a "RM2" receiver & "2G-24" 24dBi dish reflector:

Which says, on page 6: Rocket M2, Operating Frequency 2412-2462 MHz 2.4 GHz RX POWER SPECIFICATIONS llg = 1-24 Mbps => -97 dBm min +/- 2 dB 11g = 36 Mbps => -80 dBm +/- 2 dB 11g = 48 Mbps => -77 dBm +/- 2 dB 11g = 54 Mbps => -75 dBm +/- 2 dB 11n = MCS0 => -96 dBm +/- 2 dB 11n = MCS1 => -95 dBm +/- 2 dB 11n = MCS2 => -92 dBm +/- 2 dB 11n = MCS3 => -90 dBm +/- 2 dB 11n = MCS4 => -86 dBm +/- 2 dB 11n = MCS5 => -83 dBm +/- 2 dB 11n = MCS6 => -77 dBm +/- 2 dB 11n = MCS7 => -74 dBm +/- 2 dB 11n = MCS8 => -95 dBm +/- 2 dB 11n = MCS9 => -93 dBm +/- 2 dB 11n = MCS10 => -90 dBm +/- 2 dB -87 dBm +/- 2 dB 11n = MCS12 => -84 dBm +/- 2 dB 11n = MCS13 => -79 dBm +/- 2 dB 11n = MCS14 => -78 dBm +/- 2 dB 11n = MCS15 => -75 dBm +/- 2 dB

Given that my Rocket M2 is 11n MIMO, and on channel 10, I'd say the receiver sensitivity is from -88 to -92 dBm.

Hmmm... I just noticed, that this is the same (essentially) as my noise figure. But, I'm not sure what that tells me.

Rumors of my demise might be a bit premature. I'm buried in projects, work, Christmas Chrisis', ladyfriend, and trying to untrash my house, office, and vehicle. Incidentally, this is what I've been doing for "fun".

Hi Jeff,

Ooooh. Yours is bigger than mine!

That looks like a neat project.

Well, I'm here in the Santa Cruz mountains, just trying to figure out what my noise plot is trying to tell me (I'm on channel 10):

Hi, First do you understand what decibel means in the context of voltage, current or power? Can you calcualte receiver sensitivity of -92dbm comes out (?) volts? On what modulation mode are we talking about?

Googling, I can make a guess that my transient noise (blue line) is very high at channel 10 of -40dBm, but that my average noise (green line) is very very low at -90dBm.

The receiver sensitivity that someone asked me to look up is at about

So, the question I ask, without having the experience to know what "good" transient noise and average noise levels are, is whether or not these numbers are "good" or "average" or "bad"?

Also, I have no experience whether a peak instantaneous noise of -40dBm (which is admittedly high) has any detrimental effect on my radio performance (even as the average seems very low, at -90dBm).

I just don't know. Do you?

I (think I) do understand decibels, at least at a rudimentary level. Here's my simple summary of how I organize my thoughts around decibels:

1) Every 3 decibels is a doubling (or halving) of power. 2) A halving of power would be if it went from -87dBm to -90dBm. 3) My radio is transmits at 27 decibels (compared to the mW reference). 4) When I want to convert dBm to Watts, I google "dbm to watts". 5) The first hit is always the best dbm-to-Watts converter. 6) 7) So, 27 dBm is 1/2 Watt (Note: 30 dBm would be 1 Watt, & so on). 8) The receiver is sensitive to -90dBm at 802.11n channel 10. 9) That means it can pick up a signal strength of 1 picowatt. 10) The signal to noise headroom needs to be around 20dBm. 11) That means I need signal to be 1/10 Watt greater than noise. 13) So, the transmitter (without antenna) is 6/10ths of a Watt. 14) However, the Rocketdish reflector & antenna add another 24dBi. 15) A dBi is relative to a fictional spherical-radiation pattern. 16) So, my effective isotropic radiated power is 28+24=52dBm! 17) An EIRP of 52 dBm is a whopping 158 Watts! 18) The FCC only allows me an effective power of 4 Watts 19) Googling for "watts to dbm", the first link is the best. 20) 21) That 4 Watts is 36dBm 22) Even though the equipment is capable of 52dBm, it's toned down to that legal limit of 36dBm.

So, given all that, my average noise is rather low, on channel 10, of about -90dBm or 1 picoWatt (which is the green part of the bottom graph below).

However, my instantaneous noise is rather high, at -40dBm or

10 microWatts (which is the blue line in the bottom graph above).

So, I guess my question is how much will my radio be adversely affected by 10 microWatts of instantaneous noise, when the average noise is only

1 picoWatt?

I have no experience with what noise levels are good, and which ones are bad - so I have no background to interpret the spectrum analysis.

REFERENCE: Transceiver datasheet (Ubiquiti Rocket M2):

Antenna (Ubiquiti Rocketdish RD-2G-24):

I never deal with volts when I'm working with the radio. I always deal in power. So, I don't know what the volts are, but, I do know this about -92dBm:

0) First, I always google "dbm to watts". 1) The first hit is always a great time-saving calculator. 2) That's . 3) So, -92dBm is about 631 femtoWatts (which is pretty small). 4) I don't know how to convert that to volts though.

I'm not sure if I understand the question, but, the radio operates in the 802.11n MIMO channel 10 (with vertical & horizontally polarized

NOTE: The narrow beamwidth is how the antenna gets all that gain in the first place. You can't create or destroy power, so, I'd have a wider beamwidth with a lower-gain antenna. This is a pretty high-gain antenna, so, the beam width is pretty narrow, but, since it's pointed at the WISP access point a few miles away, it doesn't have to be broad.

I didn't explain that one all too well.

Here's another try at my thought process:

1. If I put the legal limit of 4 Watts into an antenna with a wide radiation pattern, it goes only so far.
2. If I then change the antenna pattern to be more narrow, the radiated signal goes farther in the direction that it is pointed.

So, my antenna & dish reflector, having a gain of 24dBi, is pretty

Breaking out my trig (SOH, CAH, TOA), I see that I can create a right

Since I have the angle and the Adjacent, and I want the Opposite, it looks like the tangent will tell me the how large of a circle is painted on the WISP antenna 3 miles away.

1. TOA means Tangent is equal to the Opposite over the Adjacent.
2. So the Adjacent times the tangent is the Opposite.
3. Googling for "tangent calculator", again I take the first hit.
4. That's
   formatting link
   .
5. So 3 miles x 0.04 is about 633 feet.
6. So the beam paints a pattern twice that, at about 1200 feet.

is cut in half (i.e., by 3dB).

Here is a picture of that pattern from the Rocketdish RD-2G-24:

Notice that this radio is pretty directional, but even so, I have a catcher's mitt about 1200 feet wide to hit my access point.

Here are the Internet speeds I get by hitting that catcher's mitt:

All I'm trying to do is *improve* on those Internet speeds, by understanding first, and then lowering my noise (or raising my signal-to-noise levels) within legal limits of 36 decibels EIRP.

REFERENCE:

First things first. I'm guessing your WISP picked channel 10 and you can't change that. Is that the case?

A waterfall is just time dependent sniffing. If your WISP wasn't spraying you with wifi on channel 10 (again, my guess), you would examine the waterfall display and find the area with the least activity. Now the waterfall is useful if someone burps wifi at you, as in a telemetry application. If the band was crowded, you would pick the channel with the occasional belch of wifi rather than one that is busy all the time. [If you were doing SIGINT, you would look for patterns in the occasional wifi belch. This is knows as traffic analysis.]

My next guess is you WISP provider has a customer on channel 2 at the same location that feeds you. [It could be another WISP from a different vendor.] The WISP provider has a beam antennas at the transmitter site. One beam for you on channel 10, another beam for somebody else on channel 2. That person may live near you since you are seeing the signal, but there are no red blobs in the waterfall, so the antenna isn't pointed directly at you. [And why would it be?] Red means a strong signal. The WISP on channel 2 is only 7 db less than your signal, but channel 2 and channel 10 have no common frequencies, so nothing to worry about.

You have two neighbors on channel 3, so that would be a bad channel for you, as would any channel that overlaps channel 3. Probably the WISP installer already knew that from when the site survey was done.

I'm not really sure how they determine the noise floor. At any one time, there is somebody on a wifi channel. It might be really low RF level, but not zero. Probably the receiver makes a determination that if it can't sniff a signal, it must be noise. Not a good assumption.

At -50dBm over 3 miles, I'd call it a day and go looking for something else to fix.

Hi, When you talk about noise there are many different kinda noise.

In the case here, it's all 2.4GHz signals which are not coming from the access point that I'm connecting to, yet, which the rooftop antenna sees.

Yes. The WISP has two antennas that I'm aimed at (because they're on the same tower).

I'm pretty sure both humps in the graphs are the WISP's antennas, one of which I'm connected to.

He's on Channel 10 and on Channel 2, and notice you see two humps, one at channel 10, and the other at channel 2.

I can only connect to channel 10 (because Channel 2 is a different password).

Wow. You're good! The WISP is on both channels, 10 and 2.

10 is for me. I don't know who is on 2, but, it's his channel too.

You got all that from the waterfall? You're good, because I only know that from other means.

Yes.

It's rare to see time on the Y axis, so I see how it's time-dependent sniffing. I guess it also looks like a waterfall, since it's columnar but in layers. Blue seems to be where I'd want to be, if I wasn't constrained to be on channel 10, which is my access point channel.

You are correct. The same WISP is feeding two different neighborhoods with two antennas, both at the same mast, one on channel 2 and the other on channel 10.

I can now see the waterfall is yellow'er on those two channels, as are the power levels bluer, and the real-time view greener.

You figured out a lot from that waterfall graph that I hadn't mentioned (because I didn't realize it might be relevant). Yes, the WISP is on both channel 2 and 10, and both antennas are on the same tower; but only one (channel 10) is meant for me to connect to.

The point is duly noted to stay away from channel 3. h

I think anything the antenna sees which is not a connection signal, is considered noise.

You have a point that the -50dBm isn't bad for a distance of 3 miles. I was more worried about the -88dBm of noise, but, now, after looking further, I think that the noise level is just about at the receiver

The main figure that worries me is the instantaneous noise of -40dBm. Do you know what effect this instantaneous noise might have on the radio?

Would wrapping the feed in tinfoil give a measure of internally generated noise?

They do sell RF Armor for the Rocketdish, but, it's prohibitively expensive:

I guess I could fabricate the RF Armor out of steel sheathing.

That would make it a neat home-repair project!

One receiver's signal is another receiver's noise.