Talk 9: Measuring Radar Emissions

Talk 9: Measuring Radar Emissions


– My name’s Frank Sanders. I’m the head of the
Telecommunications Theory division at the Institute for
Telecommunication Sciences, a laboratory of the National
Telecommunications and Information
Administration, NTIA, located in Boulder, Colorado. Over the past 30 years,
we have developed a series of techniques for measuring
emissions from various types of radio emitters. In particular, we’ve developed
techniques for measuring emissions from
radar transmitters. In this series of talks,
I will begin by describing the fundamentals of RF
measurement techniques. And I will culminate
the series by describing our techniques for measuring
emissions from radar systems, techniques that, hopefully,
you can implement yourself. Accompanying this series of
talks are notes available on our website, and also
NTIA reports that you may find useful. We hope that you will enjoy
watching this series of videos as much as we’ve
enjoyed producing them. Welcome to talk #9
in the ITS RF Measurement Seminar Series. The talk today is “Measuring
Radar Emissions,” and when we left off last time, we talked
about what we need in front of a spectrum analyzer–
namely a calibration source, a 0-70 dB RF attenuator,
a filter section and low noise amplifier section–and we try
to put all that as close to the measurement antenna as
possible so that we can allow the gain and the low noise
figure in the front end amplifier to overdrive the
loss through the RF line going down to the measurement
system. Today, I’m going to draw out
the basic way that we lay out the interior of one of our
ITS RF front end boxes. We have one of these
boxes sitting right here. And then in the lab session,
what we will do is actually use this box with this
spectrum analyzer to measure an emission out of the radar. And we’ll show an example of
how we get the radar emission parameters that include its
antenna pattern, its pulse repetition sequence, its pulse
width, and its spectrum using the front end box. So the first thing I’m gonna
do is draw out basically how we try to lay out a typical
ITS RF front end, and this, in a certain sense, is
our secret weapon in the war on radar spectrum. What we do, first of all,
is allow for up to 5 antennas that we can have in our
measurement system, and I’m gonna draw them like this. And what we need to be able
to do with the front end is be able to bring signal in from
any of these antennas like so. And at the front end, we put
a relay right here, which can switch to any of
these antenna inputs. And you might say, “Well, why
do we want multiple antennas?” Well, we want, for example,
a high gain dish antenna so that we can get the highest
possible signal-to-noise ratio in the measurement when
we’re actually doing the spectrum measurement. But we might also want to have
an omnidirectional antenna so that we can get a 360-degree
view electronically around our measurement site. We might also want
a cavity-backed spiral antenna– that’s what we’re gonna use
for the lab session in this talk today, and I’ll
explain why later. And you never know what we
might need for additional antennas, so we’ll just
call these auxiliaries. So, with that said, we have
one other component that we need at the front end, and we
want to have this component as close as possible up to the
measurement antennas, and that is a calibration source,
a noise diode. We talked about this
a little bit last time. I’ll do an entire talk about
calibrating with noise diodes, but for now, just be aware
that in the front end, we also have an input for a noise
diode which itself is powered with a 28-volt power supply,
and this noise diode will be kept normally as a component
inside the front end box– it’s about so big. I showed it in the last talk. So we can switch to the noise
diode to do a calibration, and then we can switch to any
of the antennas that we need to use to do the
actual measurement. The next step in the sequence
is to be able to go from that point through
an RF attenuator, which itself can be switched to a value
of anywhere from 0 to 70 dB. And remember that this is
the place in our system where we achieve our wide dynamic
range, the reason being that we can get up to 70 dB of
RF attenuation here plus 60 dB of instantaneous dynamic range
in the spectrum analyzer that we’re using, which equals
a total of about, give or take, 130 dB of total available
dynamic range in the measurement. Available dynamic range… OK, like so. And we do this, as we said,
by stepping our way through the spectrum–that’s what
we’re gonna demonstrate in the lab session today. Next, we need to be able to
go through a front end filter, and in our measurement systems
where we build the boxes, we actually normally put in
a whole array of such filters. We’ll have one that’s
a yttrium iron garnet filter, which is a tunable bandpass
filter–in fact, all of these are bandpass filters. We’re gonna label this “BPF”
for bandpass filter tunable, and remember that
the control voltage for this comes in from– from the spectrum analyzer
so that this knows, based upon a voltage, that’s something like
1 volt per gigahertz– and it does vary–but something
like 1 volt per gigahertz what its tuned frequency
needs to be. But in addition to having the
YIG bandpass filter, which has an approximately
20 MHz passband… that is, it allows our
measurement system to see about 20 MHz of spectrum
at a time. It’s also useful to have a set
of fixed bandpass filters… 2…and– and for that matter,
a straight-through path. And these fixed bandpass
filters might, for example, go 0.5 to 1 GHz– let me draw it like this–
0.5 to 1 GHz. And then 1 to 2 GHz, 2 to 4, and then 4 to 8–and, in fact,
I’m gonna add one more here, and then we’ll put
the straight-through path in. 8 to 18. We’ll erase this
little label just to make things a little more clear. Now, the idea is that if we’re
working with a radar signal, we’ll probably need to
go through the YIG path because we need to be able,
remember, to isolate the– the part of the radar spectrum
that we’re measuring right in here with the YIG passband
right in here from the high-power emission that we’re
not trying to measure when we’re measuring out here
in this part of the spectrum. However, in many cases when
we’re doing other types of spectrum measurements–not
measuring radars but, say, doing spectrum surveys–
what we may want to do is use one of these wider bandpass
filters because we can allow the measurement system to see
a wide frequency range–maybe, say, from 2 to 4 GHz–because
there are no strong signals in that entire range and we’re
trying to make our measurement in that range, but we just
want to limit the measurement system’s front end from seeing
strong signals outside the 2-to-4 range. Now, you might say,
“But wait a minute. Why not just use the YIG
for all the measurements?” Well, we could, but
the YIG does have some limits. The YIG will not necessarily
tune across the frequency range that we need to make
a measurement in, and more to the point, the bandpass
filters have a somewhat lower insertion loss. The YIG has–in fact, Randy,
you could probably help me. What’s the insertion loss on
this YIG, just approximately? 6 dB? It’s about… – I think the older ones
we had were, like, 6 dB. – And this one’s even a little
better than 6 dB, right, whereas a bandpass filter
can have an insertion loss of maybe only 1 dB, half a dB,
or maybe 2 dB at the most. So we can get a better
sensitivity using a bandpass filter, as long as there
are no strong signals in the bandpass frequency
range, signals that are strong enough to overload the low
noise amplifier that’s gonna follow this stage. So the way that we handle this
is, in fact, to build a set of these filters into
the box–and this box, in fact, has a set of filters
like this in it. And then we wire these filters
to a relay here, and I’m gonna show
the relay as being something that looks like this. And then we just bring
the contacts out and down… out and over… and down and over,
and all the way down. Oh, and sometimes we want
to be able to look without any filter in front of our front
end, in which case we can have a straight-through path. And we’ve got just enough
room for the straight-through path here. So we can either go through a
YIG filter, a set of bandpass– any one of a set of
bandpass filters, or a straight-through path. And then we build this thing
symmetrically on the other side like so, bring it in. I’m gonna jump over that
line and come down like this. Come over like that,
and that, and this, and finally up and around
and in like that. And then we put another relay
in the system at this point so that we can contact any one
of these switches, and then we have next, in red, a set of
low noise amplifiers… Like so. And, again, these low noise
amplifiers can switch to– we can switch to
any one of these. I’m not gonna belabor it. You can see what
the pattern is here, and we have a little–and
a second switch here. And the reason that we have
a set of these low noise amplifiers is so that, again,
we can pick an amplifier which is especially well-optimized. We might, in fact, have
something like a 0.5 to 18 amplifier, something
that’s really wide, and then we might have some other amps
that go something like 0.5 to 2, and then maybe 2 to–
2 to 4, and then maybe 4 to 8,
and then, say, 8 to 18. And then coming out of the
amplifiers, we go through one last switch…like so, and the one last switch takes us
out of the box and on– and on to the rest
of the measurement system– the spectrum analyzer,
in this case. And so we get the wide dynamic
range with the attenuator. We get the preselection at
this point in the measurement system with a set of
bandpass filters in here. And then finally, we get–
we get low noise figure… at this stage in the system,
in the front end. And so we build the entire
array into a box, which, electrically speaking,
has these boundaries from here all the way over
to the edge of the board. Everything we have drawn
in here, except for the set of antennas at the front end,
is built inside the box. And you can see that,
basically, this box consists of a set of relays… [Ring tone plays]
We’ll let that go. Hate it when that happens.
[Door closes] So the box consists of
a set of relays. It consists of a set of discrete
components between the relays– the attenuator, the set of
bandpass filters, and the set of preamplifiers, and
a calibration source, the noise diode. The box also has control
lines, it has sensing lines, it has some power supplies,
all built inside. And we do have to have control
lines, which control the relay settings, and we have to have
a voltage line for the YIG control, and there are
some ancillary lines that have to be wired in. And in the last talk,
we looked at these boxes as part of the lab session. So that takes care of what we
have to put at the front end. One other thing that I want to
talk about before we actually go ahead and measure radar
emission as part of the lab session is how, when we’re
in a field location, we would actually position our
measurement system, what we would actually do. Typically when we’re working
a field location, we’re working with a measurement system
that is configured inside an RF shielded box, which
I’m gonna draw like this. And the RF shielded box
is built onto the back of a large truck like so,
and we have a couple of telescoping masts like this that go up and allow us to
elevate a measurement antenna. And I’m exaggerating the size
of the telescope mast here, but on our current premier
measurement system, we have one of these masts
mounted at the back. We have a second mast
mounted on the other side of the vehicle at the front. We can put antennas
on either one. Bear in mind, again, that we
try to put–whenever possible, we try to put the measurement
system RF front end–which I’m drawing schematically–as
close to the antenna as possible so that we can
use that low noise preamplification inside
the front end box to overdrive the line loss that’s coming down
and then through a carefully shielded port inside the
measurement system, and which, in turn, goes into
the measurement equipment… right here, which includes,
among other things, a spectrum analyzer, an oscilloscope,
sometimes a vector signal analyzer, a PC computer to
control everything and so forth. And–I drew this with a little
too small wheel right there. All right. Now, we’ll try to get
this right because we’re gonna drive this forward,
if you will. Now, bear in mind that when
we’re trying to get the wide dynamic range on the radar
measurement like this where– where the spectrum is gonna
look like this– frequency measured power–
we’re trying to get this– if possible, if possible, we’re
trying to get this power level up to as much as 130, 130 dB above the measurement
system noise floor. Here’s the measurement
system noise floor down here. We want to get that signal
power level from a radar at least up, or as
much as, 130 dB out. Now, to do that, with most
radars–even high-powered air traffic control radars–we
have to try to get ourselves to a location on the ground
with the measurement antenna where we get the highest
possible power level coupled out of the radar, and where
we’re coupling the maximum power level into our system
by picking an optimal antenna type. And so the antenna type
that we will usually pick is a 1-meter-diameter
parabolic dish. The reason that we don’t go
bigger than a meter is it gets hard to mount a bigger antenna
up there and have the mount be rigid enough and have the
telescoping mast be rigid enough such that as the
antenna pitches and yaws back and forth like this, the radar
that we’re trying to measure doesn’t go out of
the antenna beam. If we make the antenna bigger,
the beamwidth gets tighter, and there will be a tendency,
as the antenna moves ever so slightly, for the radar
signal to get out of that beam a little bit, and that causes
the power then to fluctuate that we’re measuring in the system. The other problem is that as
the antenna gets bigger and, therefore, heavier, this
whole system will, in fact, sway back and forth more,
which exacerbates the problem of the radar going out of
the antenna beam–never mind if something were to topple over. So 1 meter turns out to be
a good tradeoff between getting a high level of gain in
the measurement system versus being able to keep the radar
in the beam as the mast and the antenna mounts
pitch and sway and yaw back and forth a little
bit in the breeze. And, um… Well, and so that’s that. Now, here is the next part
of the problem, and I just alluded to it. I’m gonna go ahead
and erase this diagram. Just remember that we’re
trying to get to a place in the radar radiation pattern
where we’re getting the maximum possible
amount of power. Drawing this out
schematically, we can imagine that we have an air traffic
control radar located over here on top of a mast. And it’s not unusual for
airport surveillance radars or long-range Air-Route
Surveillance Radars to be mounted on towers or platforms
which put the radar antenna at a height of about
80 feet above the ground. AGL, 80 feet
above ground level. They’ll have a great,
big antenna mounted up here… and this antenna is
mounted on a rotating joint. And you might think that
the place we’re gonna get the maximum amount of power
out of this antenna would be at the closest possible
distance from the radar. In fact, people commonly think
that the closer you’re coming to a radar, the more power
you’re going to get. Ah, but there is a problem,
and to explicate the problem, I’m gonna redraw the
measurement system of the mast at something that’s a little
more appropriate in terms of height scale, and to do
that, we’ll draw the mast first. If this is 80 feet, we can
get our antenna on our system up to about, typically,
30 feet above ground level. So there’s half of 80,
so come a little below that and come out here like that. This is… roughly the mounting
configuration for our dish antenna on a mast mounted on
the back of the box for our measurement system like so. And so you see that we’re
a little bit below–typically, we’re a little bit
below the radar height. And even if we tip our antenna
up a little bit, what we’re working against is the fact
that this radar’s radiation pattern looks
basically like this. If it’s an air traffic control
radar, there will be a big lobe down here. There’ll be some smaller
lobes down here like this. And so as we move away from
this radar, what we find is that we get lower power in
close, and as we come out away from the radar and we’re
measuring power level from the radar–and we’ll assume
that this radar rotates every 5 seconds. What we see is every 5 seconds,
as we’re driving this system away from the radar,
we see the radar look at us and we get a certain amount
of power, and then 5 seconds later, it looks at us
and we get this power. 5 seconds later, it looks
at us and we get this much power as we’re driving away. 5 seconds later, here,
and we see that we actually get more and more power as
we’re driving away from the radar, and the reason is
because we’re coming out into locations in the radar’s
radiation pattern where we’re getting higher
and higher coupling as we come out and away. So we keep driving–in effect,
we keep driving forward, and we keep getting
more and more power. And it turns out for most
air traffic control radars– and this will vary for
other types of radars, but for most ATC, air traffic
control radars… we get maximum coupling,
maximum power coupling–that is, max power out of the radar–
max power coupling occurs at distances… between about… 1/3 to– 1/3 to 2/3 of a mile… Which is to say around
1 kilometer, give or take. When we get out about 1/3
to 2/3 a mile, something at a distance of around
1 kilometer, we get maximum power into
our measurement system. If we go beyond that distance,
the power starts to drop off again just because of
1 over R squared factors. So this is what we try to do,
is get out into this kind of a distance range, and we
actually will do a power measurement, if we’re in any
doubt about it, where we actually will start driving
away from the radar and we actually will measure the
power, and we’ll see this power come in every 5 seconds like so as the radar looks
in our direction. We’ll actually trace that out
and we’ll actually find a zone where we got maximum power, and we saw the power
begin to drop off. Then we’ll drive back into the
middle of that zone and get ourselves set up for
maximum power out. Now, for some other types
of radars–for example, maritime surface search radars that
have a beam aiming down– this distance is probably gonna
be substantially shorter. This distance might be just the
distance across a parking lot– maybe 100 feet or
300 feet or 400 feet– so it depends upon the radar. But remember that the key,
the goal here always is–and I’ve got the lack of hair to
prove it–that we are trying to get the highest
possible power level into our measurement system. And if at all possible,
we’d like to get definitely 110 dB,
and if possible, ideally we’d like to get up to 130 dB
of dynamic range on that measurement, and we do that by
getting to an optimum location where we’re getting the maximum
power to the radar and by using a parabolic
dish antenna. Parabolic dish antenna,
typically about 1 meter, again, because that’s a good
tradeoff in terms of gain versus the mechanical
weight and size. Now, for the lab today,
which we’re performing here at the lab in Boulder,
we don’t have the luxury of being able to put our
mobile measurement system at an optimum distance from
an air traffic control radar. Instead, we’re gonna couple
the energy from an air traffic control radar that’s about
30 miles from here into an antenna that’s sitting in
a window just outside this room. So we’re not going to get
this kind of dynamic range in the measurement,
but we will use this as an opportunity to show how the
stepped routine actually works with one of our RF front
ends with our software that controls it, and we’ll also
look at how we get other emission parameters
out of the radar. And so here are the parameters
that we plan to measure out of the radar. We’ll look at the antenna
pattern, such as it is at this distance. We’ll look at the pulse
repetition sequence. We’ll look at the pulse width. And finally, we’ll look at
the emission spectrum using the stepped algorithm. There. Like so. One thing to note, the reason
that we’ll look at these parameters before we look at
the emission spectrum is we need to know what bandwidth
optimizes our signal-to-noise ratio in the measurement. And by looking at these
2 factors first, especially the pulse repetition sequence at
the radar center frequency, we’ll be able to determine the
bandwidth empirically where we get the maximum
signal-to-noise ratio. Remember from one of the earlier
talks that theoretically, this optimum bandwidth for
the measurement of the spectrum occurs when the bandwidth is
equal to about–or as close as possible–1 over
the pulse width of the radar. If it’s got a simple pulsed
CW emission, it will occur at a place where the bandwidth
is equal to the square root of the chirp bandwidth
divided by the pulse width if it’s a radar that’s
chirping, that’s FMing during the pulses, et cetera. For the lab today, we’re just
gonna look at a radar that puts out very simple pulses
where the optimum bandwidth is just equal to the nearest
that we can come to 1 over the pulse width, but we’ll
actually go through that in the–as I say,
in the lab session. So at this point, we’re about ready
to transition into the lab. Are there any questions about
what I’ve given you, so far, on these 2 boards before
we transition into the lab session, the lab part,
the demonstration part? All right. This concludes the whiteboard
talk–the whiteboard part of talk #9, and now
we’ll transition to the laboratory part. All right. This begins the lab
session for talk #9. This is where we will
demonstrate the collection of a set of parameters
on a radar emission. The parameters that we’re
going to look at will be the type of beam scanning
the radar does, the time that it takes the radar to do one
complete beam scan, the pulse sequencing in time, a
measurement of the pulse width, and then, finally,
a measurement of the radar’s emission spectrum. In this part of the talk,
we’re actually measuring emissions from a radar that’s
located roughly 30 miles from our lab. So this is not an optimum
location for the measurement, but it will demonstrate
how we get the data. We’re actually using
a cavity-backed spiral antenna in a window of our lab
building about 20 feet away from where I’m standing,
and initially I’m going to show you what we see
on a spectrum analyzer display. We’ll then transition
into the use of the computer-controlled
system. This is what we would
typically see if we were sitting somewhere
near a radar. We’re actually looking at
the radar center frequency, which we’ve earlier
established as being 2,760 MHz– 2.76 GHz–and we’re seeing
the radar as it’s looking at us every few
seconds in time. This is 0 hertz, trace
on a spectrum analyzer. So we’re looking at what’s
happening in time across this trace, and we’re looking
in 5-second intervals. The sweep time on the spectrum
analyzer right now is set to 5 seconds. Now, the bandwidth that we’re
looking at it with is 8 MHz. This is probably not
the optimum bandwidth to use. You remember in an earlier
talk we said that we need to use a bandwidth that’s as
near as possible to 1 over the radar’s pulse width
so that we can optimize the dynamic range
of the measurement. We’ll check that by actually
looking at this radar’s power in 8 MHz and in a succession
of smaller bandwidths until we’re able
to actually reduce the amount of power
that we’re measuring. So take note now that we’re at
a reference level of -10 dBm. The radar’s coming in at
a power level of about -31 dB. So now we’ll change
the resolution bandwidth to a value of 5 MHz. There we go, and we look at
what’s happening to the radar, and we have not
lost any power. We’re still measuring
about -30, -31 dB. So now we go
to a bandwidth of 3 MHz, and we wait, and we’re at about -31 or -32. So we might’ve lost
a fraction of a dB, but nothing appreciable. But note that while we’re not
paying a penalty on the radar power here, we are getting
a lower noise level over here because this noise level
measurement system is dropping at the rate of 10 log of
this bandwidth, and that means we’re seeing more and more of
the radar’s rotational pattern. We’re getting more dynamic
range on the measurement. Now we’ll take it down
to a bandwidth of 1 MHz, and we wait, and we get a slight reduction
in power, about a dB. We’re now down to about
-32 to maybe -33 dB… dBm, I should say. Decibels relative to milliwatt
is what we’re measuring here. All right. So at 1 MHz, we may have paid
a very slight penalty in measured power, but
not more than about a dB. Now we’ll take it down to
a bandwidth of 300 kHz. Ah. Now we’ve
dropped down to -20, -30, -40, -41, -42 dBm. So 300 kHz bandwidth is
clearly substantially less than 1 over the pulse width. Now we are paying a penalty
in dynamic range because this power level through the radar,
as you’ll remember from an earlier talk, drops as
20 log of the bandwidth, where the measurement system
noise level drops as 10 log of the bandwidth. So that causes us to pay
a penalty on the measurement system dynamic range. So we’ll take the bandwidth
back up to 1 MHz, and this is the point where we get
the optimum dynamic range where the noise level is as low as
it possibly can be relative to the power that we’re
getting out of the radar. Now, if this were an actual
field site location, we would’ve also worked to
optimize the measurement system’s location on
the ground, as I said earlier in the whiteboard section
of the talk, and we would hopefully be getting
about +30 dBm. Here at this location,
we’re getting about -31, -32, -33 dBm. Now, having seen this,
the next thing that we’re gonna do is look at
how this radar scans in space, and to do that, we’re going
to go to a slightly longer sweep time. We’re going to go
to let’s say a– oh, we’ll make it
a 6-second sweep time. And then we’re gonna go to
the single mode on the spectrum analyzer, and then I’m going
to hit the “single” button and start the traces. And I watch to see what
happens in the first interval, and I wait until I get
the radar to hit me right in that first interval. There. Now we’re going to get one
complete rotation on the radar. Boom! Beautiful! The radar looked in our
direction, some time elapsed, it looked in our
direction again. If we were to watch this radar
over and over, we would find that this interval is always
the same, and we’re seeing no other beam scanning
occurring in that interval. So we know that this radar has
a simple mechanical rotation, and it probably has
a 1 over cosecant squared fan-beam radiation pattern,
which tells us that it’s probably an air-search radar. Now we’ll get its rotation
time so that we can note it in our measurement notes,
and for that, we turn on a marker function. We go to peak search here,
and then we go to marker “delta,” and then
we find the next peak over. And the rotation interval
turns out to be… 4.6–4.61 seconds,
so we would note that. This is a way of
identifying radars. Even if a set of radars are
all the same model number, they’ll typically vary
ever so slightly in their actual rotation times. So 4.61 seconds–we note that,
and at this point, we would also record this scan because
this scan shows us the antenna radiation pattern at
the location of our measurement system. We’re not getting a lot of
dynamic range on that 30 miles from the radar, but we are
seeing some of those side lobe structures
in the radar pattern here. Of course, we’re also at this
location seeing some multiple– excuse me, seeing some
multipath effects. But we would record this
for antenna pattern and for antenna rotation time
and for the beam-scanning technique that the radar
is using because these all help us to identify what
type of radar this is. They’re also useful for
spectrum management purposes. Next, we’re going to look at
what’s happening right here, right inside the main beam
of the radar in time. Now, air traffic control
radars typically put out about 1,000 pulses per second. We know that they
typically put about 20 pulses in the beam because that’s
where they get their optimum signal-to-noise ratio. So if we want to see what’s
going on inside the radiation pattern of this antenna right
close to the point where it’s looking at us, we want to
go for a little more than a 20-second–
a 20-millisecond interval. Let’s look at about a 30,
a 30-millisecond interval, and we want to trigger a sweep just as the radar
is looking at us. So we’re gonna set a trigger
threshold in here that’s at about, oh, -35. So we go to “trigger,”
“video,” and here we have a trigger threshold
that’s set not too bad, but we’ll bring it up
just a little bit. And now we’ll go to
single-sweep mode. Now, we saw just the end of
the pattern as it came around. We want to see the beginning,
so we’re going to go to a trigger delay
here of–whoops. Gonna go
to a trigger delay on– go to a trigger delay of,
oh, say, -4 milliseconds. 1, 2, 3, 4 or so
milliseconds– it’s not terribly important. We get another sweep. It may take up to 5 seconds
for the radar to look in our direction. I’ll bring the trigger level
down, just slightly down. OK, and “single.” There we go. Now we’re seeing
into the middle part of this radar’s
radiation pattern. In fact, let’s take
a maximum-level pulse right here and count out until
we’re down 3 dB. And what I’m gonna do is,
in fact, go to peak search with a marker, go to “normal.” All right. And now we’re
gonna go to “delta,” and we’re going to go from
that maximum-level pulse out to a point where
we’re about 3 dB down, which is going to be
about there. Ah, 1.49–not quite enough. Not quite–there! Eh, we’ll go back to this point
where it’s a little more than 3. OK…there. OK, that pulse is
2.7 dB below the peak. Now we’re going to go to the
next pulse over here that’s equal to that amplitude,
and that is going to be this pulse right here. There! OK, those pulses are
almost equal in amplitude. The time between them
is 17.6 milliseconds, and these pulses are all within
about 1.5 dB of each other. None of them are more than
3 dB down from the peak. This is where the radar is trying to really
find the target. It’s looking for a target right
in there with those pulses. That should be about
15-20 pulses by radar theory from an earlier talk. Let’s see how many
it really is. 1, 2, 3, 4, 5,
6, 7, 8, 9, 10, 11, 12, 13, 14, 15,
16, 17, 18, 19, 20. Ha ha! Hit dead-on. OK, so they’ve optimized this. That’s good to know. Basically, if you see
the radar with 20– well, basically, if you can’t
see the radar with 20 pulses integrated in the receiver,
you’re not gonna see it at all. That’s what it amounts to. But if you substantially
reduce the number of pulses, you do reduce the probability
of seeing of the target. So we know again
that this radar is working in a sort of conventional sense. Now let’s look carefully
at the intervals between these 2 pulses. The interval between these
2 pulses is 1.05 milliseconds. Now we got to marker “normal”
so that we’re now on that second pulse, now “delta,”
and we’re gonna get the time from the second pulse
to the third pulse. Ah, 1.0 milliseconds–not
the same interval as the first interval. Now we’ll go back to “normal”
on the third pulse, and now we’ll get the time to go
from the third pulse to the fourth pulse. So, remember, 1.05 milliseconds,
1.0 milliseconds… 850 microseconds. All right! So now we have 3
intervals between these pulses: 1.05, 1.0, and
0.85 milliseconds. Now back to “normal,” “delta.” Now we’ll come and look at
the interval to the next pulse in the sequence:
800 microseconds. All right. So now we have, what, 4 intervals
that we’ve measured. Now we look at
the next interval: 800 microseconds. Now the next interval
after that: 850 microseconds. So we see that the intervals
between these pulses are changing ever so slightly. They’re on the order of
a millisecond apart, but they’re changing slightly. This tells us that
the radar is utilizing a feature called stagger. “Stagger” means that
the intervals between pulses in the sequence are not
exactly equal to each other. The reason for doing this is
to eliminate lines that would occur in the radar receiver
if, in fact, the radar were transmitting pulses
at an exact and non-varying interval. For example, if it were
to pulse–were sending out exactly 1,000 pulses per
second, you would get lines spaced exactly 1 kHz apart
in the receiver. This is a problem if you’re
doing Doppler processing. In order to defeat or prevent
those lines from occurring, you have to break up
the intervals between the pulses such that the intervals
between sets, or pairs of pulses, are not equal
to each other and are not even multiples of each other. And this is a technique that
is used in air traffic control radars, which have so-called
moving-target-indicator processing, which is
Doppler processing. So the fact that this
radar has a stagger evident in the pulse sequence
means that, again, it has moving-target-indicator–
that is to say, it’s got Doppler processing. It’s probably an air
traffic control-type radar because they’re worried about
using it to get velocities. All right.
Now, we would note and we would record this
pulse sequence as part of our measurement set on the radar,
and we would note this so-called stagger sequence
of intervals, which we could also play back out of the data
later, having recorded it. Now let’s zoom in
one more time. Let’s zoom in on an individual
pulse inside this sequence and measure
the pulse directly. We could, and oftentimes
would, use an oscilloscope with a detector
to measure this. But if you have to do
a rough-and-ready job and you’re in a hurry, you can go ahead
and use a spectrum analyzer if– as a–sufficiently shorten
sweep time, which this one does. So what we’re going to do is
go to the sweep time of not 30 milliseconds, but rather
a sweep time of 10 microseconds, because we’re guessing that
this pulse width is gonna be about 1 microsecond because
we’ve optimized our dynamic range with a 1 MHz bandwidth. So we determine, in fact,
empirically that the pulses must be about
a microsecond long. And now we’ll go back
to triggering, video triggering, and we’ll actually bring
the trigger–let’s bring the trigger threshold
up a little. We’ll go to “single,”
and we’re going to need to do one more thing, which is
to change this delay from 4 milliseconds to about
2 microseconds. Do that ever so quickly…
2 microseconds, or so, so that we can see
the beginning of the pulse. All right. Take another trace. We might have to wait
for up to 5 seconds. Ah. Now, we’re in
a 1 MHz bandwidth, and we’re seeing the effect
that that 1 MHz bandwidth is actually rounding
off the pulse shape. Here we want the widest
possible bandwidth so that we can get the best possible
resolution and pulse shape. The widest bandwidth we have
available is 8 MHz…boom. Now wait another
5 seconds or less. I thought I hit “single.” Try it again. There! Beautiful! And now we can
use the marker– although I can see that because
this is 10 microseconds across, which means 1 microsecond
per division, this is about 1 microsecond long,
but we can actually use the marker function
here to look at that. There’s that, and here’s
the other side of the marker, and what do we have? 933 nanoseconds, So it’s about
0.9 microseconds pulse width. OK, perfectly consistent
with what we saw before. Again, we would
record this information if we’re doing
a set of measurements. We would write down this pulse
width, and if the radar were transmitting multiple pulse
widths, we would get a set of these types
of measurements. Now, having done all this,
we’re ready to run the actual spectrum measurement,
stepped algorithm measurement, and for that, we’re going
to go back to the bandwidth of 1 microsecond. We’re gonna verify that our
detection is positive peak. We will go to a sweep time of
5 seconds, and the reason is the radar rotates
every 4.6 seconds. So if we have the measurement
system wait for an interval of 5 seconds, thank
you very much, then we know that the radar will look in our
direction every time we take a measurement point. And the radar is tuned
to a center frequency 2,760. We don’t have a lot of dynamic
range available at this location, so we’re gonna measure
across what, John, 80 MHz… – Yep.
– Total? All right. So go ahead. John Carroll here
is going to actually perform the stepped measurement. And I–let’s start this
measurement–let’s start the measurement, starting with
what the spectrum analyzer sees, and then we’ll
switch over to the computer at an appropriate time. So we’re going to
look at this 2 ways. We’ve only got one projector,
so, first, we’re gonna look and see what the measurement
algorithm–which is computer-controlled–does
to the spectrum analyzer. This analyzer is now running
under computer control, or it will be. – It will be.
– But it will be. It’s going to tune to
a particular frequency. It’s gonna sweep
for 5 seconds. It’s then going to pick off
the highest amplitude point that occurred in
the 5 seconds, correct it for a calibration factor,
store that number and display it on a computer screen,
and then step in frequency by an amount that’s equal to
our own resolution bandwidth so that we’re looking through
a 1 MHz window in width, and we’re gonna step
that window one window with 1 MHz across,
one point to the next. Right now we’re still tuned to
the radar’s center frequency, but… – We’re ready?
– Ready. Ready. This is a measurement
system called RSMS-4G, for fourth generation. We’re going to use
antenna port number 1. Now, we’re not gonna use
the YIG here because the radar signal isn’t strong enough
to warrant using the YIG, so we’re gonna use a bandpass
filter that goes from 2-4 GHz. This is our measurement system
front end as the computer understands it. We’re gonna come in through
antenna port 1 on that box. We’re gonna go through
that first relaying position where we could calibrate
with a noise diode. We’re gonna go through an
attenuator, which is initially set to 0 dB. We’re gonna pass through this
bandpass filter–and these are its characteristics–then
we’re gonna go through a low noise amplifier, and then we
pass out of the front end box and into
the spectrum analyzer. This is the nicest software
that we’ve ever been able to implement. We’ve gotten better and better
ever since the late 1970s when we started doing this. Now, this is the measurement. We’re gonna start
at 2,720 MHz. We’re gonna measure
up to 2,800 MHz. Remember that the radar
is tuned to 2,760 MHz. We’re at a 1 MHz bandwidth,
and we need to take how many steps to go from 2,720 to
2,800 in a 1 MHz bandwidth? The difference between these
2 numbers is 80 MHz. So if we’re in a 1 MHz
bandwidth for the measurement and we’re gonna go through
80 MHz of spectrum, we need 80 steps. The math works out
rather nicely on this. – OK, now it’s not attenuating. – All right. Now,
when the measurement system starts to run, it starts off
initially with 70 dB of “R” of attenuation snapped–
and we can see it over here–and it actually
gradually reduces the amount of attenuation,
looking for an attenuation level where it begins
to see signal coming in. This is something that we have
automated in this system that we used to have to do
kind of manually, so… So it take it a few seconds to
find that nice attenuation level for the start. Now, we ourselves know right
now that what we want is 0 dB attenuation, but as a general
purpose solution, it’s nicer to do it this way. – It’s not setting it, and for
some reason, it went to 70. You can see it’s at 70,
but it’s not stepping down. – All right. Let’s try it again. – [Indistinct]. – For the purpose
of the tape, we can– we can edit this out. – Do you have any idea
what’s going on, Steve? – [Indistinct]. – I–I thought
you had it working. – I had it working before. When you started,
it was working fine. – OK. – Frank, it’d be great if
the RSMS software [indistinct]. – Yeah, but we’re gonna cut
this part out on tape. – How much time have you
spent on the development of this stuff? – This has been going on
for a few years now on this particular
software, yeah. Kill it?
Kill it [indistinct]. Well, we–of course, we’ve turned it off
and turned it back on. That probably got it
kind of confused. So, again, we run
the application here. OK. Antenna port number 1,
bandpass 2-4. Stepped. OK. 2,720 to 2,800,
80 steps. OK, I don’t hear it
switching the attenuator. – Now it’s trying
to do something with mass continuation. [Indistinct]. – Tell you what, Jim,
why don’t you go ahead and stop the tape running. This nuisance is just
wasting tape here. We’ll get this–
we’ll pick it up again as soon as we can solve this. We’ll cut out that–OK. And, billing room, we had
to put a delay in here while we had to take care of
a software crash. We’re now running again,
so I’m gonna stop talking to give you about 10 seconds,
and then we’ll just pick it up. Then you can edit out
everything up to this point where things were
being problematic. OK, so now what’s happening
is the spectrum analyzer is looking for an attenuation
level that it can use where it is not overloading and where we can see the
measurement system noise floor. So it starts off with
a high attenuation level. It steps it down:
60, 50, 40. We’re still coming down. Basically, the algorithm looks
for a point where either the overload flag is set
in the spectrum analyzer or we hit 0 dB of attenuation. At this frequency, 2,720,
where the analyzer’s working right now, it’s gonna hit
0 attenuation and not see the overload flag, and then
it’ll start the measurement with that level of
attenuation invoked. All right, it’s hit
0 dB of attenuation. It never saw the overload
flag, so now it’s taking a 5-second sweep at 2,720 MHz. It pulls off the highest
amplitude point in that 5 seconds, which turns out
to be at -73.7 dBm. Then it steps by a frequency
map that equals the bandwidth, which is 1 MHz. And so we saw 
about -73, then -75. Now we’re seeing a level
at that point of -64 roughly… – Noise? – And that may
have been noise. Now we’re going to… see a point there at -66. Those could’ve been sort of
stray noise points. Or in this particular band,
they could’ve been– and they probably were–
hits from a couple other
radars in the area. They’re probably
very distant radars. I was watching them
on the spectrum analyzer display in time and I think they’re
a couple other radars, really distant. Now what we’re measuring
is measurement system noise down here. This level down in here,
around about -77, in a 1 MHz bandwidth,
peak detected. We checked
and we saw earlier– and you can’t see it on
the tape–that we are, in fact, getting the noise from
the low noise amplifier in the measurement
system front end into the spectrum analyzer
at a level that exceeds the spectrum analyzer’s
own noise floor. So the noise figure of
our overall measurement system, in other words, is being
determined by the low noise amplifier at the front end– that’s optimum;
that’s what we want. We’re still seeing measurement
system noise as we work our way across to 2,740 MHz. This technique may look slow,
but it’s faster than doing maximum-hold mode with
peak detection in a spectrum analyzer. And as we’ve pointed out
before, this allows us to get wider dynamic range because
we can step the attenuation as we move along. That was another momentary
hit from a source that was not the radar. And the reason I can say that
is I’m watching the spectrum analyzer, which we can’t
actually see on the tape. So those are a couple
other stray points. We’re continuing to
work our way along here. We’re up to 2,749 MHz. 2,750. The radar is tuned right here. We’re gonna come up about
40 dB as we approach 2,760. We’re at 2,753 MHz. 2,754. 2,755…2,756. Now we’re beginning to
see the radar. I can see it on
the spectrum analyzer display. 2,757. This is the radar coming in, the radar spectrum
developing here. Here we go. 2,758. 2,759. The system’s automatically
adjusting the dynamic range on the display as we get
higher-level points out of the radar spectrum. Now we’re coming down. This is an air traffic control
radar that uses a klystron, which is itself filtered with
what’s called a diplex filter, so it’s a very sharp
emission spectrum on this radar. So we see that every 5 seconds
we get a point on the spectrum, on
the computer display. That’s a real feature in the
radar spectrum right there, a little bump. Now we’re coming back down into the measurement
system noise floor. Now we’re back down to the
measurement system noise floor. We’ll go ahead and
run this out to 2,800. Had this been a radar
measurement in a good location in the field, the dynamic range
that we would’ve had here would’ve been at least 100,
preferably 110, maybe as much as 120-130 dB. Here, as I said, we’re only
getting about 40 dB of dynamic range because of our
measurement location, but our point in this part of the talk
is to show how the measurement algorithm itself works. And I emphasize that if this
radar were doing some complex frequency hopping, if it were
doing complex beam scanning, it wouldn’t make any difference
to this measurement algorithm. We would get it just as well,
just as long as we wait long enough for every single one
of these measurement points. For a given radar, this can be
anywhere from 1 or 2 seconds to as much as, for some
weather radars, up to 1 minute per point. For most radars, it falls
between 5 to 10 seconds per point. You notice that the way that it
fills in the radar spectrum, we don’t get any gaps. You will typically with
a positive peak detector running at maximum-hold mode
get little gaps through here. That doesn’t happen with
a stepped algorithm because we methodically wait at each
frequency for the radar to look at us. As I said, these points
are not the radar emission. The reason we know that
they’re not the radar emission is I was actually able to
watch the spectrum analyzer screen on each of these data
points as it was developing the spectrum, and we–I saw that the emissions here
looked like momentary pulses. They could either have been
noise spikes or they could’ve been really distant radars. They were not this radar,
and I could tell because I could see what was happening
in the time domain as these points were taken. This screen over here
bears mentioning. This shows us
in real time what’s happening on the spectrum
analyzer display. We have some other nice
features in the software. We can see whether on any
given point we hit overload according to the IF monitor
in the spectrum analyzer. This is a warning
about YIG alignment. We’re not using the YIG here
for this measurement today. And then this green light
shows us when the spectrum analyzer is itself
actually sweeping. So it sweeps for 5 seconds,
pulls off the maximum point in 5 seconds, corrects
the point for calibration plots. It steps to next frequency
and runs across the frequency range designated in frequency
step sizes that are just equal to the bandwidth that
we’re using–in this case, 1 MHz bandwidth equals
1 MHz step sizes. And now we’re finished with
this particular demonstration: radar measurement. So at this point, having
demonstrated how the measurement system works,
are there any questions? All right. Well, thanks
for sitting in on talk #9 in the RF Measurement
Seminar Series.

Leave a Reply

Your email address will not be published. Required fields are marked *