Special Pages

Sunday, March 6, 2022

A Ham Radio Series 31 – Phased Arrays, or A Dish Without A Dish

In Ham Radio Series number 31, I went into details on the parabolic dish antennas, especially their gain and radiation beamwidth.  Given the tradeoffs in antenna construction, we always ask if there’s a better way to do things.  For example, consider satellite communications.  The satellite TV world generally uses parabolic dishes;  they may not be 12 to 15 foot diameter, circular dishes like those used in the 1980s (when I was in the design side of that world), but they’re permanently mounted on a house’s roof, or on a mast in the backyard or in some other way so that they can always see the geosynchronous equatorial orbit (GEO).  The satellites use that orbit so that the antennas don’t have to track satellites across the sky – they’re stationary.  Being in the GEO imposes costs on the satellites: they need more power to get a good signal to a smaller dish from 22,236 miles out in space.   Bigger antennas on the satellites (or the ground) can ameliorate that.  All of that goes into the system design trades and you’ve got to know the system designers have chosen a combination of antenna gains on both ends of the link and transmitter power they thought was the best trade.

Now consider situations like the Starlink internet satellites that aren’t in GEO.  They don’t want the satellites to be that far out because it adds time for the signals to travel to and from the satellites, which adds latency to the internet link.  Just the limitation of the speed of light, which is rarely a factor in daily life, adds time.  The time for a satellite signal to reach the Earth’s surface from the GEO is around 1/8 of a second (119 milliseconds)  - double that for two way communications and add any time added by the hardware and system.  Starlink has said that a they are working towards a latency goal of 20–50 milliseconds and think that 10 msec is possible.  The satellite altitudes seem to only take up around 2 msec due to the speed of light.

Because the Starlink satellites are much lower in orbit than the GEO, they’re going to be moving across the sky and the antenna is going to need to track them.  Just based on what we know of the frequencies the Starlink satellites use and their altitudes, we can’t say much about the required antenna gain or beamwidth, but with different technology than the parabolic dish, both may be adjustable. 

Consider this side by side comparison between a Dish Network antenna that would be used to receive TV from GEO to a Starlink antenna and you’ll see a striking difference.  Please let me point out that I have no reason to believe that these antennas have the same gain, just that they have the same basic purpose: receiving signals from satellites.  While the Starlink uses several frequency bands, some are the same as the DishHD antenna uses.. 

The Dish Network antenna is a design that allows the user to put three different Low Noise Block downconverters (usually just called an LNB), and each of the LNBs is at a different focal point of the small dish, allowing it to get signals from three different satellites.   Although not very noticeable in this product photo, the triple feed arrangement on the boom sticking out from the dish is curved.  

The Starlink by contrast, does not have a boom at all and is a sealed container.  I don’t know if this is official in any way, but the antenna system is referred to as “Dishy McFlatface” and the antenna is flat.  It’s not a parabolic reflector and instead of an LNB sitting above the flat face, there are hundreds of LNBs integrated into the antenna assembly.  The rectangular version of Dishy is newer; earlier versions were more circular but with portions of the circle cut off, resulting in a rectangle with rounded corners.  Another important difference is that the Starlink antenna also transmits uplink signals from the users to the satellites, so it transmits and receives.  

So how does one make a flat antenna perform essentially equal to a parabolic dish?  Back in the first article I wrote on gain in antennas, I noted the basic idea was, “stick more metal in the air,” as long as we all agree that it can’t be randomly sized and placed chunks of metal.  In this case, the antenna is an array of quarter, half, or one wave long antenna elements fed by circuits called phase shifters that create delays in the signals to the radiating elements needed to change the wavefront to the desired shape and direction.  As a result, the antenna elements can combine received signals in phase creating a much higher amplitude signal or the phase shifters can create delays between the antenna elements during transmit, creating focused beams of transmitter power. 

Don’t think of phased arrays as something new and exotic you’ve never seen.  All of the commonly used directional antennas are phased arrays; a Yagi is nothing less than a phased array of half-wave elements which, when optimized, will be slightly longer (in the reflector elements), or slightly shorter (director elements) than exactly ½ wavelength.  The Yagi is phased by the mechanical spacing, not by electronic circuitry. 

Without getting too deeply into what can be very deep, a Yagi works by interaction of currents in the elements.  Consider a three element Yagi like the one in the first post on directive antennas (post 28 and the right half of this figure) 

The driven element, like any half-wave dipole, radiates mostly perpendicular to its long axis equally forward and backward; that means into and out of the screen in this view.  When that wave gets to the rear and front elements, the wave induces a current in them and they re-radiate it with a phase shift.  Some of the re-radiated signal goes back to the driven element, and some goes the other way.  Because of the (nearly) quarter wave distances between each element, and while this is happening the driven element continues to advance in its waveform, so that by the time those reflections get back to the driven element the current on it has advanced almost a half wave.  The timing (phasing) is set up so that signals coming from the reflector forward add to the next wave at the driven element while the signals from the director are out of phase when they reach the reflector and cancel.  The phasing of the signals by the physical spacing determines the forward gain (along with other factors, for sure).  It has one driven element and two parasitic elements, but it's phasing that makes it work.

You might be saying, “OK, but this is just getting some improvement in one direction, how can a phased array get better in multiple directions?”

The biggest advantage of a phased array antenna, like the ones used in “Dishy McFlatface,” is that there can be scores of elements, not just three (to 12 or 20, like a Yagi).  The second advantage is that they can be arranged in two dimensional arrays of antennas which can allow the beam of the antenna to be swept in space, electrically steering the way the antenna is pointed.

A simple phased array can be made from power dividers, programmable attenuators and phase shifters.  Power dividers can be obtained in many division ratios, but the lowest loss versions tend to be powers of 2.  Passive power dividers tend to be reciprocal; that is they’ll split one input to two outputs or they’ll combine two inputs into one output, just by connecting them as you want.   That makes a simple eight element phased array look like this schematically (the circles with the Greek letter “Sigma” in them are power splitters/combiners – Sigma from the common math notation of “summing”):

If connected in a line, as this implies, they can point the antenna’s beam where desired, but only moving in one plane (up or down in this picture).  Connecting another eight elements in a row at right angles to this one, if fed by twice as much power and another power splitter, the beam can be pointed in two axes; both up and down and in and out of the screen.  (Duplicate everything to the right of that first power splitter and cut its input; add a splitter to feed that one with one of its outputs and use the other output to drive the new schematic page, but you do need a bigger PA – or a distributed network of smaller PAs).  

As shown, the bottom antenna has the least phase shift (delay) and fires slightly sooner than the one above it, which fires slightly sooner than the one above that, and so on with the top element delayed the most.  

Note that this is just for the transmitter side because most people find it easier to visualize combining the eight antennas to produce a beam pointed in different directions, but antennas are reciprocal, too.  The same phase shifts that will transmit in a preferred direction will receive better in that direction, too.  Circuits like this are big business, especially with Multiple Input/Multiple Output (MIMO) cellular systems like 5G.   Note that they use switched chains of attenuator, phase shifter, Power Amp or LNA (Low Noise Amplifier), phase shifter, attenuator.  Transmit path is the top four in the dashed oval; receive path is the bottom four.  It’s also possible to approach this by only switching the amplifier between LNA and PA, using half as many attenuators and phase shifters. 

A 5G cellular antenna array based on the Anokiwave AWMF-0139 quad core IC.  (Source)

Unlike the parabolic dish antenna, I can’t provide simple equations for gain and beamwidth.  They’re far too configurable for simple things like that.  They can, however, be designed to provide similar values for both.  In many applications, parabolic dishes have come to be considered as having too many disadvantages and are being replaced with phased arrays, even if they are largely mechanical and aren’t electrically steerable.  There are phased arrays of microwave antennas that are flat panels instead of parabolic dishes used in radar systems, for example.



11 comments:

  1. Thank you. I think I want to go back to school.

    ReplyDelete
  2. We used to say that the only antenna formula we needed to know was: Weight of the aluminum in pounds times height above ground in feet equaled foot-pounds of torque on the ionosphere.

    ReplyDelete
    Replies
    1. and you're not done until you can get the ionosphere to glow.

      Delete
  3. Thank you for the basic explanation of flat plane phased array antennas. Back when I was in electronics school in the mid-70s, these were not in common use. I don't believe they even existed yet, but I could be wrong. I have wondered how these work for quite some time.

    ReplyDelete
  4. Here's a link to a description of the first phased-array radar that I'm aware of, the AN/FPS-85 radar located near Eglin AFB, FL. As an Air Force captain, I served as test director for a major upgrade in the mid-to-late 1970s. Most of what I observed/learned there I have long since forgotten :-). https://www.globalsecurity.org/space/systems/an-fps-85.htm

    ReplyDelete
    Replies
    1. Thanks for that story. When I saw the web page, I thought it was the original PAVE PAWS system, which I had heard about somewhere back then. I guess it was the predecessor to those.

      That history would be worth diving into.

      Delete
    2. The FPS-85 was, and I suppose still is, very productive as a space track system since, facing directly south, it would eventually see and track every satellite in orbit, subject to size and distance limitations. Systems facing in other directions could only see objects with a high enough inclination that they would pass through the coverage area.

      Other than the FPS-85 and Pave Paws, another notable phased-array radar is Cobra Dane, which has its own wikipedia page as well as a novel available on Amazon. I had little to do with Pave Paws, but was a project officer on Cobra Dane. Never had the "good" fortune to visit its location on Shemya.

      Delete
    3. That is, all satellites other than geosynchronous satellites that were stationed at longitudes outside the coverage.

      Delete
  5. So SiG, what is the actual physics at play that causes the resulting "bend" in direction? I've been listening to an audiobook about Einstein- his life and work- and people keep glossing over the "how" part for a lot of observable physical phenomenon. They say "field theory" and leave it at that.

    Why do the emissions from each antenna interact with the emissions of the antenna next to it at all? Or am I making some sort of conceptual error when I think of the delayed emissions as "dragging" the emission (lower antennas on your drawing) toward the delayed antenna? How do the separate emissions interact, and through what medium?

    I can't help but picture the beam steering as similar to what happens with a stream of water from the faucet when you touch one side with a finger, it slows that side and 'pulls' the stream toward the finger. In the case of the water stream though, there are easily understood forces affecting the water.

    n

    ReplyDelete
    Replies
    1. This is hard to explain without hand waving and a ton of figures, which the Blogger comment engine won't let me do.

      Let's start with the figure showing the eight antennas. Think of the amount of delay between each antenna and the bottom one, so the top antenna is the most delayed (the wave is closest to it) and the bottom not delayed at all. Each of those antennas has a pattern of its own and the signal spreads out like the ripples from a rock dropping into water. The step in delay from antenna to antenna is what makes the angle of the wavefront change.

      The emissions from the antennas don't interact with each other. There is no medium they interact with. It's just that the farther you get from the eight ripples spreading out in eight circles, the more those circles look like a single circle. At the receiver, the voltages induced do add the most in the desired direction, while off that direction, they add to smaller totals because of the phase differences.

      Electromagnetics guys refer to antennas having a near field and far field. When we talk about antenna patterns it's always far field. Each one looking like spreading ripples is the near field view. In practical terms, it's hard to define where near field stops and the far begins, but think "a few hundred" wavelengths. In the far field, the combination of all of those ripples smooth out to look like a smooth wavefront pointed in that direction. In the near field, it looks more chaotic.

      Naturally, someone came up with a single number for where far field begins but it's really not that sharp. To give you an idea, the number is 2*((antenna size)^2) over wavelength (the units for measuring antenna size and wavelength have to be the same). I happen to have an example I did while working on a 10 GHz weather radar. The wavelength of 10 GHz is 3cm, the antenna I was using was 46cm diameter. The equation said the far field as starting at 1393 cm. 464 wavelengths.

      Hope that's helpful.

      Delete
  6. Thanks, I clearly need more time pondering...
    n

    ReplyDelete