EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: rs20 on August 10, 2015, 04:38:59 am
-
So, I just had this random idea, and I'm wondering if it is a) flawed, b) can be improved upon, or c) has already been done?
-- Get a highly directional 5GHz* parabolic antenna (e.g., this one (http://www.tp-link.com/Resources/document/TL-ANT5830B_V1_QIG_7106503929.pdf) with 30dBi gain, 6 degree horizontal halfpower beamwidth, 4 degree vertical beamwidth)
-- Place it somewhere with a good view of a city skyline, and scan it in the elevation and azimuth directions,
-- Have some sort of SA or detector recording the signal coming off the dish,
And for a 120 degree field of view, you'd end up with an image that is "sharp" at 20 pixels wide or so. Now, that's hardly a megapixel image, but it might produce a recognizable skyline, and you'd make a slightly blurry 100 pixel wide image at the very least. The image might even be amenable to a bit of classical image sharpening, since the blur is well characterized. By having the SA discriminate different Wi-Fi channels, the image could have colour.
5 GHz chosen because lots of Wi-Fi uses it, so I imagine office buildings are bathed in a 5 GHz glow (over a long enough exposure time). Although 2.4 GHz is even more popular, its longer wavelength makes larger dishes required for a given beamwidth. Can anyone think of higher frequencies that office buildings are likely to emit?
I'm sure someone has tried this before, but I can't think what keyword to search with?
-
It seems like it could be interesting :)
You might want to take a look at the details of astronomical interferometry so that you can resolve an image.
-
You might want to take a look at the details of astronomical interferometry so that you can resolve an image.
That's basically a phased array, right? That requires phase-coherent multi-channel receivers, which sounds expensive :(
-
Direct illumination will just give radar, which won't work great with sheer surfaces (of which there are many), and probably wouldn't look great. Ambient illumination might be cool, but where would it come from? If you have a sensitive enough receiver, you might use ambient white noise (which should primarily be from the Sun).
Tim
-
Direct illumination will just give radar, which won't work great with sheer surfaces (of which there are many), and probably wouldn't look great. Ambient illumination might be cool, but where would it come from? If you have a sensitive enough receiver, you might use ambient white noise (which should primarily be from the Sun).
My proposal was to have a completely passive receiver, and rely on all the 5 GHz Wi-Fi routers as sources?
-
Ah yeah, hmm... it'll be pretty small and diffuse though...
Consider: sunlight is about 1kW / m^2 at the surface. Supposing you had an antenna array large enough to receive that over a mere 32 x 32 mm square, you'd get 1W. Which is more than the total power outputs from a legal router.
Of course, sunlight is pretty frickin' bright!
A room is typically illuminated with a few hundred watts of incandescent, or under a hundred watts of fluorescent or LED. In either case, the total power output is around 1-10W, depending on room size, how bright you like it, etc.
We would reasonably expect that, if we could have a 5GHz camera the same size as a visible-light device, that a room so illuminated would look about as bright either way. Assuming, for some reason, an analogous intensity scale.
So...
During the day, offices are typically lit (artificially). Even when the sun is shining (yeah, we make a lot of sense, don't we?.. :P ). But you can't see it, because of contrast.
But at night, it's usually pretty easy to see windows glinting around. Perhaps this would be pretty representative, even during the day, for our 5GHz camera?
It's worth noting that IR-reflective windows may not be very good microwave windows: they may have a bandpass characteristic, being much more reflective at generally lower frequencies, not just IR. (It comes to mind, a very fine gold plating was (is?) used on space suits as a sunscreen. I don't know what normal window coatings actually use.) So the windows themselves might be pretty well "tinted" in the microwave range.
That said, even if they're tinted, well, intuitively we should still be able to see that against reasonable darkness, so it can't be too bad.
So it's looking pretty good, assuming we can resolve the windows, or at least a good population of windows on the side of a building if they're all blurred together by our poor resolution.
Of course, doing it serially, you should probably expect exposure times on the order of hours, which may be annoying for consistency of signals (some will flit in and out, especially the radiation from portable devices). In principle, a phased array should be able to pull down a complete image in parallel (give or take exposure time, for averaging), but you'd have to do a live FFT to get an image out of it, and that has to be done in a phase-correct manner, before detection and averaging.
One could also construct a lens and use an antenna array scheme (without minding phase, just an antenna, LNA and diode detector per pixel, then read out the detectors like they're a CCD array). Of course, a lens has a variety of practical problems, so a curved reflector (a parabolic mirror, past the focus?) would be preferable.
Tim
-
I think this is similar to what you have in mind, just made with 2.4G spectrum: http://wificamera.propositions.org.uk/ (http://wificamera.propositions.org.uk/)
While not exactly what you proposed, I find this amazing: 3D mapping of signal strength in space: http://hackaday.com/2015/02/17/mapping-wifi-signals-in-3-dimensions/ (http://hackaday.com/2015/02/17/mapping-wifi-signals-in-3-dimensions/)
Edit: I think I saw several years ago a "WiFi camera" taking picture of a view from window, but I can't find the link.
-
I think you should try this idea. All sorts of interesting problems may come up, including nonlinearity/gamma to make things look better, let alone the sensitivity of the receiver in the first place. Keep wifi devices away!
Consider: sunlight is about 1kW / m^2 at the surface. Supposing you had an antenna array large enough to receive that over a mere 32 x 32 mm square, you'd get 1W. Which is more than the total power outputs from a legal router.
That 1kW/m^2 is spread across a broad bandwidth. A router's output would be (relatively more) narrowband. I can't guess at the exact ratios
Also, I believe 5Ghz does not survive the atmosphere well.
On the contrary, 5Ghz (~6cm) is let through. https://en.wikipedia.org/wiki/File:Atmospheric_electromagnetic_opacity.svg (https://en.wikipedia.org/wiki/File:Atmospheric_electromagnetic_opacity.svg) http://www.cv.nrao.edu/course/astr534/Introradastro.html (http://www.cv.nrao.edu/course/astr534/Introradastro.html)
-
Regarding interference from the sun, I was wondering whether the 1kW/m^2 figure was relevant to 5GHz when I thought that figure was more about IR/visible light in the hundreds of THz. You seem to be strongly correlating 5GHz and visible light with respect to office light as well, is there a reason for this? I found this:
(http://www.britastro.org/radio/RadioSources/images/sources-fig31.png)
(Edit: link & credit (http://www.britastro.org/radio/RadioSources/sun.html))
Which gives about 1M-2M "Janskys", a new unit I've only just learned about. Anyway, assuming an observation bandwidth of 100 MHz (which seems generous), and the 600mm x 900mm parabolic antenna I looked at, that's about 10^-12 watts, or -90dBm. (Google calculation (https://www.google.com.au/webhp?sourceid=chrome-instant&ion=1&espv=2&ie=UTF-8#safe=off&q=2e6+*+1e-26+W+per+meter+per+meter+per+Hz+*+.9+meter+*+.6+meter+*+100+MHz)).
Meanwhile, a Wi-Fi router, a 250mW isotropic radiator, at a distance of around 500 metres, captured by the same parabolic antenna, works out at 4 * 10^-8 watts, or -44dBm. (Google calculation (https://www.google.com.au/webhp?sourceid=chrome-instant&ion=1&espv=2&ie=UTF-8#q=250mW+%2F+(4+*+pi+*+500+m+*+500+m)+*+900+mm+*+600+mm)). Edit: these calculations assume a perfect efficiency of an antenna, I've since calculated that a perfect antenna of this size would be 32.75dBi, but it's actually 30dBi, so we need to penalise these figures by 2.75dBi or so.
Please feel free to question my calculations, but a Wi-Fi router at 500m distance appears to be 46dB stronger than the sun, even if the sun is directly in line of sight. And as for the sky in general, I don't believe much scattering occurs in the atmosphere? It's not like there's a "blue sky" of 5GHz radiation, I boldly claim.
I think this is similar to what you have in mind, just made with 2.4G spectrum: http://wificamera.propositions.org.uk/ (http://wificamera.propositions.org.uk/)
While not exactly what you proposed, I find this amazing: 3D mapping of signal strength in space: http://hackaday.com/2015/02/17/mapping-wifi-signals-in-3-dimensions/ (http://hackaday.com/2015/02/17/mapping-wifi-signals-in-3-dimensions/)
Thanks for those links, both look cool. I don't know what to make of the first link, lots of expensive-looking technology, and then tiny tin can antennas that can't possibly have the phase discrimination ability to get any sort of directivity? And I can't make anything out from any of the pictures... I thought "wow, you can see the window frames!", but that turns out to be a optical sensor on the side!
-
The best graph I could find looks to be a black-body curve, where 5Ghz looks to be a tiny (off-graph) proportion relative to total sun output (not visible/IR output):
(http://csep10.phys.utk.edu/astr162/lect/light/spic-sun-ant.gif)
http://csep10.phys.utk.edu/astr162/lect/sun/spectrum.html (http://csep10.phys.utk.edu/astr162/lect/sun/spectrum.html)
6cm = 6e^8 angstroms
-
The best graph I could find looks to be a black-body curve, where 5Ghz looks to be a tiny (off-graph) proportion relative to total sun output (not visible/IR output):
(http://csep10.phys.utk.edu/astr162/lect/light/spic-sun-ant.gif)
http://csep10.phys.utk.edu/astr162/lect/sun/spectrum.html (http://csep10.phys.utk.edu/astr162/lect/sun/spectrum.html)
6cm = 6e^8 angstroms
I always take these graphs with a grain of salt, because the curve shape depends entirely on whether the vertical axis is given in per-wavelength or per-frequency units. The peaks are in different locations and everything. In particular, this graph is probably per-angstrom, but a 100 MHz bandwidth centered on 5 GHz (4.95GHz-5.05GHz) is 6.06e^8 to 5.94e^8, a range of 1.2e7 angstroms. So even though the curve is obviously quite low down there, you have to integrate over 600 times the width of the graph shown there to get the actual power.
-
I think this is similar to what you have in mind, just made with 2.4G spectrum: http://wificamera.propositions.org.uk/ (http://wificamera.propositions.org.uk/)
While not exactly what you proposed, I find this amazing: 3D mapping of signal strength in space: http://hackaday.com/2015/02/17/mapping-wifi-signals-in-3-dimensions/ (http://hackaday.com/2015/02/17/mapping-wifi-signals-in-3-dimensions/)
Thanks for those links, both look cool. I don't know what to make of the first link, lots of expensive-looking technology, and then tiny tin can antennas that can't possibly have the phase discrimination ability to get any sort of directivity? And I can't make anything out from any of the pictures... I thought "wow, you can see the window frames!", but that turns out to be a optical sensor on the side!
The many cantennas take care of vertical sweep of the sensing beam. In one of the links on the page you can find a "camera" built with just one cantenna. The antennas don't have very good resolution, but I would not expect much from 2.4G signal. It's probably more of proof-of-concept design, good enough to make an interesting art installation, but shows that mapping ambient radiation is possible.
With an array of phased antennas, and boatload of advanced math you could try to greatly improve resolution by exploiting correlations between signals received by antennas, building something resembling synthetic aperture radar. This is what virtual radiotelescopes do, but IMHO it might be a topic even for a PhD.
-
Regarding interference from the sun, I was wondering whether the 1kW/m^2 figure was relevant to 5GHz when I thought that figure was more about IR/visible light in the hundreds of THz. You seem to be strongly correlating 5GHz and visible light with respect to office light as well, is there a reason for this?
Yes:
- Because a router puts out a bit under a watt, typically;
- You know intuitively what 1W of radiation looks and feels like (over a 32mm square, it feels warm; over the visible spectrum, it's about a flashlight's worth),
- The contrast might be wrong (perhaps wildly so), but the intuitive sense of an image of that intensity should still be just fine.
I added several "weasel words" to allow for these differences, which might be missed in a casual read. :box: :-DD
Which gives about 1M-2M "Janskys", a new unit I've only just learned about. Anyway, assuming an observation bandwidth of 100 MHz (which seems generous), and the 600mm x 900mm parabolic antenna I looked at, that's about 10^-12 watts, or -90dBm. (Google calculation (https://www.google.com.au/webhp?sourceid=chrome-instant&ion=1&espv=2&ie=UTF-8#safe=off&q=2e6+*+1e-26+W+per+meter+per+meter+per+Hz+*+.9+meter+*+.6+meter+*+100+MHz)).
Meanwhile, a Wi-Fi router, a 250mW isotropic radiator, at a distance of around 500 metres, captured by the same parabolic antenna, works out at 4 * 10^-8 watts, or -44dBm. (Google calculation (https://www.google.com.au/webhp?sourceid=chrome-instant&ion=1&espv=2&ie=UTF-8#q=250mW+%2F+(4+*+pi+*+500+m+*+500+m)+*+900+mm+*+600+mm)).
Please feel free to question my calculations, but a Wi-Fi router at 500m distance appears to be 46dB stronger than the sun, even if the sun is directly in line of sight. And as for the sky in general, I don't believe much scattering occurs in the atmosphere? It's not like there's a "blue sky" of 5GHz radiation, I boldly claim.
So that should be pretty good, and my analogy of lit buildings seen at night (except that this applies, for this frequency, even during the day) is probably pretty accurate -- the light from a skyline is plainly visible, over a (whatever it is, 0.01 lux or less?) dark background. Probably, the ~40dB figure is pretty comparable!
Thanks for those links, both look cool. I don't know what to make of the first link, lots of expensive-looking technology, and then tiny tin can antennas that can't possibly have the phase discrimination ability to get any sort of directivity? And I can't make anything out from any of the pictures... I thought "wow, you can see the window frames!", but that turns out to be a optical sensor on the side!
Yeah, that's probably just a swept array of detectors (or, simultaneously orders of magnitude both smarter and dumber: an array of wifi radios read by RSSI only, I'd be willing to bet).
Like I said, a phased array would be cool, but rather involved. The radio part wouldn't be terrifically bad: each antenna gets a tuner, LNA, mixer (driven with an LO isolation buffer), coherently downconverting all simultaneously to a more manageable frequency, like 100MHz, or 10MHz. Problem: you need a center frequency at least as high (or wide, since you can direct convert to DC as the center frequency) as the desired bandwidth. If you want to watch the entire band at once, that needs to be over 100MHz. If you're okay with slices, you can get by with less (and vary which slice you're watching by sweeping the LO -- a handy enough way to do time division multispectral). Finally, you can't just make snapshots of this, or send all the channels into detectors: that throws away all the phase coherency and doesn't net any spacial correlation*. You have to record the v(t) or V, Phi with an equal array of ADCs, so you can do the DSP later. Which understandably will take gigs of RAM and more than a few bucks worth of DAQ!
*You still get some (which is even helpful for some things, like measuring the diameter of a star (http://mysite.du.edu/~jcalvert/astro/starsiz.htm), perhaps), but mostly, it will be the difference between a camera with no lens whatsoever, and a million lenses in every direction at once.
Tim
-
Um... You might wish to take a look at some Ham Radio microwave pages on "Noise Temperature". Then look at professional devices called "Dicke Switches" or "Dicke Switch Radiometer" You will need a very low Noise Figure front end to do this, and I'm not sure you will see much. Whether you see WIFI nodes will be very much a factor of how close you are to the node, but if you are at any distance at all they will just be a very tiny noise below the thermal noise. Simply because they do not have highly directive antennas.
We do aim Ham systems at the sun, sky, clumps of trees, local earth, and moon as noise figure tests, but we really don't expect resolution enough to map cities.
Making low noise preamps is an art.. It takes really good instrumentation... You might want to buy one from DownEast Microwave or similar suppliers.
Just get a used IF Noise Figure Meter or PANFI and hook that to a really clean downconverter, after a really low Nf preamp. Sensitive Noise Figure hardware is rather complicated. However surplus rigs from microwave communications links are out there for sale...
This is one case you probably do not want a an active phased array, antenna noise figures add... :-)
You want a fairly wide bandwidth, probably 20 Mhz or more, and the equations are out there to calculate what you need. Narrow is not good for your application. We use very narrow bandwidths to avoid noise when working on very long distance communications. You want the opposite, but not so large as to lose spatial resolution.
Let me clarify that last statement, you really do not want to look at a band that is so wide that you are seeing system noise outside the bandwidth of your dish or yagi...
I'd also look at the quite bands reserved for Astronomy and Earth science besides 5 Ghz... So I could differentiate out what is the city and what is possibly the WIFI..
A few guys are doing OK with a carefully selected RTL-SDR dongle for noise figure. I'd interpolate between pixels by linearly translating the rotating antenna after a scan, a la inverse synthetic aperture radar.
You will learn an awful lot about physics trying however...
Steve
-
Um... You might wish to take a look at some Ham Radio microwave pages on "Noise Temperature". Then look at professional devices called "Dicke Switches" or "Dicke Switch Radiometer" You will need a very low Noise Figure front end to do this, and I'm not sure you will see much. Whether you see WIFI nodes will be very much a factor of how close you are to the node, but if you are at any distance at all they will just be a very tiny noise below the thermal noise. Simply because they do not have highly directive antennas.
I'm only about 500 to 1000 metres away from these routers, which works out to -47dBm to -53dBm? Please double-check my calculation in my earlier post, but that seems like easily enough power for even a half-decent SA to pick up?
A few guys are doing OK with a carefully selected RTL-SDR dongle for noise figure. I'd interpolate between pixels by linearly translating the rotating antenna after a scan, a la inverse synthetic aperture radar.
That only works with active radar with a receiver coherent to the transmitter, right?
-
No, the RTL will easily digitize noise above its self noise in a given tuning. It's not known for its flatness across its bandwidth however.
Cantennas are horrible beasts, avoid!
-
No, the RTL will easily digitize noise above its self noise in a given tuning. It's not known for its flatness across its bandwidth however.
?? I don't understand what you're saying here?
Cantennas are horrible beasts, avoid!
Agreed, was never intending to use them!
-
Short version, the RTL/SDR Dongle can easily be used to see noise. As long as the noise is greater then its own internal sources of noise, ie thermal, 1/F, harmonics etc... Ie there is a minimum detectable signal that is device dependent...
Steve
-
Ah, confusing terminology. "Noise" is generally something unwanted. Now that we do want to read these signals we're still calling them "noise" because that's what we used to calling them.
Of course the rtlsdr can digitise signals above it's noise floor. Whether these signals are called noise or not is down to definitions/choices/conventions.
-
Microwave engineers usually preface this term with something, ie Thermal Noise, System Noise, Sky Noise, Sun Noise, Moon Noise...
So we can refer to it as City Noise or Skyline Noise in this case.
Noise measurements and Noise Figure are significant tools in the small signal engineer's tool kit at RF.
Steve
-
Regarding interference from the sun, I was wondering whether the 1kW/m^2 figure was relevant to 5GHz when I thought that figure was more about IR/visible light in the hundreds of THz. You seem to be strongly correlating 5GHz and visible light with respect to office light as well, is there a reason for this?
Yes:
- Because a router puts out a bit under a watt, typically;
- You know intuitively what 1W of radiation looks and feels like (over a 32mm square, it feels warm; over the visible spectrum, it's about a flashlight's worth),
Most of the radiation would be IR though. The amount of visible radiation from a 1W incandescent lamp is probably less than 10mW, around an order of magnitude less than a router.
-
Most of the radiation would be IR though. The amount of visible radiation from a 1W incandescent lamp is probably less than 10mW, around an order of magnitude less than a router.
Who the hell uses incandescent flashlights anymore? I meant LED. Likely one of the bigger ones with more than a few watts of emitter.
If you had taken the time to read my first post, instead of skimming or ignoring it, you would've seen a much more accurate estimate, incandescent emitters included.
Tim
-
So what you're basically looking for is a radio telescope, in crude terms, except operating in an unusual frequency band (but one where devices such as LNAs are easily available) and it's pointed at the city, not the sky.
But a lot of the same physics and engineering factors apply here as apply to radio astronomy, so they're pretty well understood, documented things and you have a lot of prior art and literature on the subject you could look up to get started.
-
Hi
Sounds like you're venturing into the realm of Radio Astronomy. As you may know,anything above absolute zero (kelvin) radiates em energy, how much so at 5 GHz will depend on the underlying physics associated with black-bodies (check out the Rayleigh-Jeans law). There are all kinds of web sites that feature projects in armature radio astronomy, you don't need a giant dish like Greenbank (West Virginia) or, for that matter, an interferometer array like the VLA. An old television satellite dish and the LNB will do nicely to detect the solar temperature at that frequency. I used to get students to walk into the beam pattern of this type of ultra simple radio telescope to demonstrate that they also are radiating electromagnetic energy.
If any of you are interested you might check out a publication by William Lonc, Radio Astronomy Projects, 2nd. Ed., 2003.
Brad
-
You can pull some useful numbers out of here:
http://www.swpc.noaa.gov/ (http://www.swpc.noaa.gov/)
Specifically: http://www.swpc.noaa.gov/phenomena/f107-cm-radio-emissions (http://www.swpc.noaa.gov/phenomena/f107-cm-radio-emissions)
and here:
http://www.ham-radio.com/sbms/presentations/AF6NA/practical_sun_vs_sky.pdf (http://www.ham-radio.com/sbms/presentations/AF6NA/practical_sun_vs_sky.pdf)
Pulling some data out of the above presentation...
Cold sky temperature is about 30 Kelvin for Microwave purposes.
Earth / Building Temperature is about 290 Kelvin for Microwave purposes.
Sun Noise is ~ 350' Kelvin ( @ 10 Ghz) for Microwave purposes.
You need to factor in system and waveguide noise, but the math is straightforward for this task.
From those numbers, you can deduce your contrast ratios and amount of target emission at Microwave...
Hint, it might be easier to form a good picture at 11 or 24 Ghz where off the shelf LNAs cost 30$ and have very good noise figures.
Steve
-
You can pull some useful numbers out of here:
http://www.swpc.noaa.gov/ (http://www.swpc.noaa.gov/)
Specifically: http://www.swpc.noaa.gov/phenomena/f107-cm-radio-emissions (http://www.swpc.noaa.gov/phenomena/f107-cm-radio-emissions)
and here:
http://www.ham-radio.com/sbms/presentations/AF6NA/practical_sun_vs_sky.pdf (http://www.ham-radio.com/sbms/presentations/AF6NA/practical_sun_vs_sky.pdf)
Pulling some data out of the above presentation...
Cold sky temperature is about 30 Kelvin for Microwave purposes.
Earth / Building Temperature is about 290 Kelvin for Microwave purposes.
Sun Noise is ~ 350' Kelvin ( @ 10 Ghz) for Microwave purposes.
You need to factor in system and waveguide noise, but the math is straightforward for this task.
From those numbers, you can deduce your contrast ratios and amount of target emission at Microwave...
Hint, it might be easier to form a good picture at 11 or 24 Ghz where off the shelf LNAs cost 30$ and have very good noise figures.
Thanks! Where do I find these $30 LNAs? Are you talking blocks with SMA (or whatever) connectors, or are we talking PCB components?
-
Go to EBAY, type in "LNB" or "LNA"
Usually have a ~ 1 Ghz broadband output.
Steve