Activa JavaScript para disfrutar de los vídeos de la Mediateca.
How to model digital beamforming - Contenido educativo
Ajuste de pantallaEl ajuste de pantalla se aprecia al ver el vídeo en pantalla completa. Elige la presentación que más te guste:
Descripcción del proceso y costes del diseño de arrays de antenas, por Keysight.
Hello. Thank you, moderator. This is Rafael Riva-Torres with Keysight Technologies, and I'm an R&D application scientist.
00:00:01
Today, I want to talk to you guys about how to model, do some digital beamforming phase array, right?
00:00:12
And let's talk about the concept also of phase array, about channel-specific impairment.
00:00:21
So that's the overarching theme for today.
00:00:27
Okay, now my agenda, we're going to cover some phased array trade study, phased array active loading and scaling, and a digital Rx phased array.
00:00:30
Now, this presentation is born out of several interactions, both personal and customer. So the first one, this phase array trade study. You know, we do trade studies and this is actually some, born out of some personal experience.
00:00:43
You're asked for a trade study, you're given a certain set of conditions, and this is one
00:01:04
of those things that people just sometimes, you know, management, program managers, etc.,
00:01:11
think you can just turn this thing around or just sit down, you know, do a few calculations
00:01:17
and you're done.
00:01:24
It never results that way.
00:01:25
So how can we use a tool like Pathway System Design to help us do our trade study, right?
00:01:28
Again, some personal experience.
00:01:35
I also want to talk about some new features that we put into our Pathway System Site 2022
00:01:40
that I honestly believe will do two things for you.
00:01:45
It increases the accuracy of your phased array simulations,
00:01:51
and it also provides some relief in terms of simulation time.
00:01:56
And the simulation time is not all in pathway system design.
00:02:02
We'll look at that, right?
00:02:06
Some of these EM simulations can be really long.
00:02:07
How can we help?
00:02:10
We'll see that.
00:02:12
That's what this whole scaling thing is about.
00:02:13
We'll talk some more about that when we get there.
00:02:15
And the third topic, working with a customer and they come,
00:02:17
oh, you guys have some nice solution for phased array analysis.
00:02:22
But, you know, all phase arrays today are going digital, right?
00:02:26
There's no such thing as, in the case of this customer,
00:02:30
as a phase array where we can plug in, you know, an RF signal
00:02:34
and, you know, do testing, et cetera.
00:02:42
So we would like to actually design that way.
00:02:45
Well, you know, we have a solution for you.
00:02:48
So let's talk about that one, right?
00:02:50
That's our last topic.
00:02:52
Actually, the picture that you're seeing here on the top,
00:02:54
is actually that phase array you can see there's an analog to digital converter
00:02:56
that's going to be our digital output and we're actually going to do the beamforming in the digital
00:03:01
domain it's so i'll leave the details for later so when we get there so let's move on to the trade
00:03:06
study so here's the situation right i i changed some of the characteristics to sort of protect
00:03:14
some of the names but again seriously some of the stuff that as engineers we
00:03:24
live through right here's just your situation there's a phaser a team is
00:03:32
working on some SATCOM program and they've been tasked with a trade study
00:03:35
right you need to determine the cost and performance of the proposed ISA and
00:03:41
and it's going to be used in the ground station uplink.
00:03:49
And here's a proposed design.
00:03:54
Here's somebody put it together for you in pathway system design.
00:03:55
It kind of looks like this, right?
00:04:00
And that's your design.
00:04:01
Let's say that's the case, right?
00:04:02
So what is the trace study all about?
00:04:05
Well, here's the problem statement.
00:04:08
We have the SACCOM ground station phase array.
00:04:09
It has 625 antenna elements.
00:04:12
we're looking for an EIRP of 400 kilowatts and somebody decided to put a
00:04:15
tailor window on it and there's the characteristics of your tailor window now
00:04:22
your trade study is there's two PAs available right and people cannot make
00:04:26
up their mind right which one to use so your task for this trade study is select
00:04:34
either one of the PAs, right, and cheapest is not always the best, nor is expensive always the best
00:04:40
either, right, higher cost. Or you can actually do some kind of optimization and use a combination
00:04:48
of the two. And how do we do that in the phase array analysis, right? How can you even accomplish
00:04:57
we'll look at it. So we have two PAs, one of them is called PA1, and then we have PA2.
00:05:05
You can see the characteristics for gain and frequency are very similar, and then you look
00:05:11
at the output saturation power. There's one at 2 watts, and there's one at 20 watts, right?
00:05:17
That's 10 watt difference, and there's like a 6x difference in price, and then a really
00:05:24
large difference in terms of power, almost 15x or so.
00:05:28
And by the way, we didn't magically pull these numbers, we actually did a quick survey around
00:05:36
several vendors and we looked for vendors that had PAs that work in this frequency range
00:05:43
and kind of what the difference in price and power was and this is what we came up with.
00:05:49
of average of everything that we found so this is kind of reasonable to think
00:05:57
that a 20 watt versus 2 watt PA would be like a six times the cost now you have
00:06:02
to compare these two PAs and make and have even a third option where you
00:06:12
optimize where you put one PA or the other in your array right 625 elements
00:06:16
is a 25 by 25 array. Now you might be inclined to say oh I'm just built three
00:06:22
three schematics and I will run three simulations I get you know three data
00:06:30
sets and I can compare them all right but in pathway systems I you you do not
00:06:37
have to do that you can create one schematic and then sorry about that you
00:06:43
can then use certain properties that we have that we allow you to select
00:06:52
different components programmatically so you see this small snippet of MATLAB
00:06:57
code this is MATLAB code the one that's built into pathway system design no
00:07:04
extra license needed it comes with the tool so I have this variable select
00:07:09
notice assessor select one two three I have enough these components I have them
00:07:13
such that when you select number one number one is active the other two are
00:07:17
open number two the middle one is active the other two are open and so on right the same thing with
00:07:21
the third one um and this just makes life easier right what you have is one single schematic you
00:07:26
just go to this script you tell it to run and it actually does that right it says okay this time
00:07:35
i'm going to use a data set for the 33 dbmpa it runs the analysis then for the 43 and then there's
00:07:40
this thing called a group PA. What's this group PA thing all about, right? Well, I'll
00:07:47
talk about that in a second. So the whole idea is you have a swift analysis, single
00:07:53
schematic, you know, be working with three schematics. You make a change in one, you
00:07:58
have to go to the other to make the same change, and all this stuff, you avoid all that. So
00:08:05
this PA is something new that we came out in our last release it's called an
00:08:10
array RF amp and it's an array of amplifiers that's two things for you
00:08:16
right number one it allows you to have all the amplifiers with different values
00:08:22
when you run your Monte Carlo analyses right so in the past we were you the RF
00:08:31
amp and sure you can run in a Monte Carlo analysis but all 625 would have
00:08:36
the same gain and same 1 DB compression point etc because it was just one
00:08:42
component right this one actually you have 625 amp the flyers in this
00:08:51
component which is the way our face array solution works and each one when
00:08:56
you run the Monte Carlo will have, you know, whatever within the statistical variations that
00:09:02
you specify will be different. So that's one use case. The other advantage of this is you can create
00:09:07
groups of PAs, right? Remember one of the options, one of the three options for your trace study,
00:09:12
you can use a combination. What that means is some of your elements can be fed by PA1 and
00:09:18
some other group of your element, antenna elements, can be fed by PA2. So that is a third option.
00:09:25
right so we did that we did this analysis and let's look first at let's
00:09:32
say PA 1 and PA 2 so PA 2 is a yellow one and I'm gonna be consistent with my
00:09:41
colors and PA 1 is a blue one and the figure of merit that we're gonna look at
00:09:45
today is gonna be C and DR which stands for character noise and distortion ratio
00:09:53
like your signal to noise ratio but including distortion and the
00:09:58
dots that you see here there's 625 of them so it represents each channel in
00:10:06
your phasor right so there's 625 elements so there's 625 channels and you
00:10:12
can see that the blue one right by the way CMDR is one of those parameters that
00:10:19
the higher the better and you'll notice that the yellow which is the more
00:10:24
expensive higher power one has a much better on average CMDR if you want to
00:10:29
think of it that way whereas the blue one you know it has some some challenges
00:10:36
right it kind of dips below 20 this line here is 20 might be a little bit hard to
00:10:42
see but my scale here alright so it drops almost all the way down to you
00:10:47
went to 15. So that's not a good thing, right? Definitely you can see where using all PA1s is
00:10:52
not as good as using all yellow ones, but all the yellow ones, which is PA2, they're more expensive,
00:10:59
more DC power. So what if we could do something where we say, replace all the PA1s with PA2s
00:11:04
where and only if CMDR drops below 30 for PA1. If it stays above 30, and here's our 30 line,
00:11:13
right, which is quite a few of them, we're going to leave it alone and we'll only replace it when
00:11:24
it falls below 30, right? So we sort of did that. And so here's where we're going to use group PA,
00:11:29
we're going to use both of them again the color scheme holds right blue dots are pa1 and yellow
00:11:39
dots are pa2 notice that when you replace them everything says above 30. all right and um well
00:11:46
how good is that right and let's be able to come we should be able to compare all three solutions
00:11:55
now that's what our trade study that's the outcome or the desired outcome of our trade study
00:12:00
so I'm going to use the traffic light indicators here for what's good right red means not good
00:12:05
yellow means somewhere in the middle and green means the best right sorry so here's PA all PA
00:12:15
ones ERP does not have a color it's not highlighted because all options meet the EIRP
00:12:24
Notice that even with PA1, right, the problem with PA1 is it has the lowest CMDR RMS.
00:12:31
So we did an RMS across all 625 channels, and we came up with 34.7, right?
00:12:38
What it has going for it, it's the lowest cost and the lowest power consumption.
00:12:46
Now, here's the version with PA2.
00:12:54
Again, EIRP is not a factor here.
00:12:56
It also meets the requirements handsomely now, right?
00:12:58
Notice the CMDR RMS went up to 39, pretty good, where it's not so good as in cost and DC power, right?
00:13:02
We know that the overall cost here is six times more than with PA1, which, you know, the program is not enjoying right now.
00:13:12
All right, so here's the version with the group PAs.
00:13:22
Remember, we replaced the blue ones with yellow only when the CMDR dropped below 30.
00:13:25
and notice that I have the parameters here all yellow by the way notice that
00:13:31
yeah he we only drop 0.1 DB compared to the all PA twos we're just kind of
00:13:36
fantastic and fabulous result by the way cndr did drop a little bit but still
00:13:42
it's better than just all PA ones and you can kind of judge for yourself in
00:13:49
terms of power and cost right and this is a visual from pathway system design
00:13:54
it is actually our new phase array visualizer that does this right so you
00:14:05
can actually tell it oh yeah I want the colors for this this or that parameter
00:14:11
it numbers and it does a few things that helps you visualize your phase array
00:14:15
right so by the way we're using CMDR because it is inversely proportional to
00:14:18
EVM so if you're doing any kind of digital comms and things like that you
00:14:26
know that here's where the higher the CMDR the better for yourself for you
00:14:30
right and we end up with the conclusion that amplifier grouping is the best
00:14:35
compromise in terms of performance and cost now of course if it happens to you
00:14:41
like it happens to me when i do these trade studies uh your boss says um can we lower the
00:14:47
the cost some more and still kind of meet the requirements of course the answer is oh i just
00:14:53
need to rerun the trade study and what do you want me to change the lowest uh cndr for right so we
00:15:00
did it at 30 you you know what if you want to experiment let's say with 28. you can go back
00:15:07
and do that and do these trade studies and all these what-ifs right and you the most fabulous
00:15:12
thing is this analysis uh took me like uh from scratch setting it up um running the simulations
00:15:18
and even putting the slides together it takes like four hours um and to be honest with you
00:15:28
the simulations are incredibly fast the whole three simulations that was like somewhere around
00:15:34
12 seconds is where we clock that at which again on my computer so it's pretty fast uh accurate
00:15:42
and um it doesn't really take that much to set up and by the way once you have it set up guess what
00:15:49
you can do all these what ifs that your boss is gonna or the program is gonna ask you anyway
00:15:54
okay so let's talk about a couple of really nice features honestly uh for our phased array solution
00:16:02
which is active loading and scaling and they actually go together somewhat but let's talk
00:16:10
about what we mean by active loading so in the antenna array pair uh panel excuse me an antenna
00:16:17
array panel active loading is a result of em coupling between the 10 array elements so there's
00:16:26
all kinds of coupling between those array elements i'm showing four of them here you can see already
00:16:31
with four of them, how much coupling you have.
00:16:36
This coupling will depend on how close, the proximity between the elements, the farther
00:16:40
away the less coupling, right?
00:16:44
And there's other design things that you can do to try to reduce the coupling between them.
00:16:46
We can capture this coupling with S-parameters.
00:16:51
So if you're running an EM analysis, pick your favorite 3D EM analysis, either ours
00:16:54
or the competition we support several of our competitors a CST HFSS you can
00:17:01
bring in those files into our tool and if captured a copy matrix you know the
00:17:08
couple signal levels is what sometimes people mess right so what I mean by that
00:17:17
is how much of this signal and port two couples back to port one and from port
00:17:21
3 to port 1 and so on and so forth right so how much how much is that so it
00:17:27
doesn't only depend on the coupling matrix it also depends on the actual
00:17:33
value of the signal remember in the prior trade study we had a Taylor window
00:17:39
so that means that the amplitude that goes into the end each other antennas it
00:17:44
conceivably different right the other thing is this excitation voltage yes
00:17:47
will change with your scan angle so as you're scanning also these voltages
00:17:56
these are all the effects that you need to capture right and again you bring in
00:18:02
the antenna patterns you bring up in your coupling matrix and our tool will
00:18:07
consume that and allow you to do that analysis now let's talk about this
00:18:11
coupling matrix and antenna patterns and let's say that we want to do a large
00:18:16
array but in my case here I'm gonna say that a large array is a 10 by 10 array
00:18:23
which depending on your EM simulator what you set up the frequency and all
00:18:28
kinds of other things that can be a fairly long simulation right but it gets
00:18:34
to a point with your race starts to grow even further than this there's things
00:18:40
and the different EM tools that help you speed up things you can do like an
00:18:45
infinite array etc etc all these kind of things and that's not really nice and
00:18:49
good but you know when you have to come back and do an array like the analysis
00:18:55
like we do it in pathway system design that also means you have to bring all
00:19:00
that data right not only did you have to wait for the data sometimes even days
00:19:05
right you also have to just transfer all that data now we came up this is a
00:19:11
innovation from us in pathway system design and we can actually scale small
00:19:18
matrix to a big array right so our case here let's say we have a five by five
00:19:25
and we can scale it up to a ten by ten or even bigger and what is the what is
00:19:30
the the phenomenon how is it that we can do that well if you look at a phased
00:19:38
array and you consider any element as soon as you get to two elements away
00:19:43
from it by the time you get to the third one the coupling from that one back to
00:19:48
your original one let's say I'm looking at 28 here right by the time I get to
00:19:51
58 58 is going to show you a lot less than 38 and 48 right so we can say that
00:19:55
I can approximate 28 and coupling to its neighbors with the next two elements
00:20:03
right and then if you look at the bigger array that means that a lot of these
00:20:12
antenna elements in terms of the coupling from their neighbors look very
00:20:17
similar so for example 28 looks a lot like 38 why because 28 has two elements
00:20:23
to the to the left to the right up and down and you can see even a diagonally
00:20:29
it looks that's very similar to the position so what we end up doing when we
00:20:34
scale is we say that everything with a similar color here they have a coupling
00:20:40
babies they all have similar coupling to their neighbors right and the same thing
00:20:45
with all the different colors what that means is if I can come up with a smaller
00:20:54
matrix right that does that that captures that effect I have I only have
00:20:58
to run an EM analysis of the smaller array, which
00:21:05
saves incredible amounts of time and simulation in your EM tool.
00:21:10
And remember that thing about transferring data?
00:21:16
Data files are much smaller, right?
00:21:18
So this is the concept, right?
00:21:21
So we save you time.
00:21:23
Not exactly all the time is saved in the pathway system design.
00:21:24
It's actually saving you time and running your EM analysis.
00:21:28
shorter em simulation and measurement time okay and you can run as big of an array as you want
00:21:32
right i did a 10 by 10 here because in order to kind of prove to you guys that this kind of this
00:21:39
this works i am going to do two things right i actually we actually did run the 10 by 10 array
00:21:45
in in our em solver em pro from keysight technologies and we did run the five by
00:21:51
five and i'm going to compare them side by side all right in typical fashion i like to run these
00:21:57
scripts right i have three options so why not uh single schematic three options and i have my
00:22:07
script again that sort of controls my simulation single schematic i can do everything now we have
00:22:13
three cases the first one is no active impedance so we just put it together and we just we close
00:22:20
our eyes and said there's no active impedance right there's a case and that
00:22:26
we consider the less accurate it it works it gives you results and there's a
00:22:33
lot of good things about it but there is the missing piece which is this active
00:22:42
impedance right then there's the full active impedance that's the one where we
00:22:47
ran the full EM for the 10x10 array and then there's the scaled one which is the one where
00:22:52
we ran the 5x5 and then that's the data that we consumed inside of pathway system design
00:22:58
and for directivity right and remember the active z is the most accurate we were getting 25
00:23:05
and it seems that all three versions get pretty close right now look at the EIRP right
00:23:10
So, the no active C gives you 53.1 dBm.
00:23:17
The most accurate that we have is the active C, that's a by 10, 51.8.
00:23:23
Kind of the difference with the no active, right?
00:23:28
And notice the scale.
00:23:31
The scale gives you 51.7, very, very close to the active C.
00:23:32
So, while still an approximation, right?
00:23:37
But it gives you really good results, fairly accurate, right?
00:23:41
but then you save tremendous amount of time and resources transfer again data
00:23:45
files and that sort of thing here's a few patterns for you to look at on the
00:23:51
left is the sex in case that we were just looking the prior slide so not the
00:23:59
scan angle of zero and you can look at the pattern and you can see for example
00:24:03
the no active impedance have the really deep nulls but that doesn't really
00:24:11
happen in the other array in the actual active C array and you can see that the
00:24:15
scale one also does not have the really super deep nulls right in in the
00:24:23
side the beam width and the main beam all three solutions looks very similar
00:24:29
right and again we feel that the scale and the active C are closer than the no
00:24:33
active and the same thing happens when you scan again inside the main beam
00:24:41
there you begin to see some differences right again we feel that our scale
00:24:45
solution is very good and again those deep notches that happen with the ideal
00:24:52
one which don't happen with the other two right you will notice that farther
00:24:58
away you get from the main beam in both cases the other two solutions do
00:25:03
begin to depart from the active see the full-blown analysis if that were ever
00:25:10
important to you you know consider still using the full array right but if you
00:25:17
want to focus on the main beam and several side lobes near the main beam
00:25:25
this is an you know the scale array is a fantastic option all right so we talked
00:25:29
about some advantages that we give you and let's talk about the now about this
00:25:38
phased array digital design right and we're going to do the RX again this is a
00:25:44
customer is they want we all our phased arrays are moving to digital how can we do
00:25:50
that but they still need to do some RF lineup right and understand some Monte
00:25:56
Carlo some variability statistical analysis this is all stuff that we will
00:26:01
do with pathway system design so part of the the situation here is some of the
00:26:05
control is and some of these are the you know phase arrays do some of these
00:26:16
being former ICS many of them are complete receivers right they have gain
00:26:22
control they have even program of both filters they have integrated analog to
00:26:28
digital converters etc so let's say you have one that looks like this and then
00:26:36
you have many many of these because you have probably have one per channel some
00:26:40
just bring two of them inside four of them inside sort of help reduce the chip
00:26:44
count but still you have many of them right which is the point and how do you
00:26:50
begin to your model this right then include it in your design I went ahead
00:26:55
and make some assumptions totally convenient for this demo right hopefully
00:27:03
some of you can relate to to the nature of it so in yellow is our beamformer IC
00:27:10
so notice I have the mixer with the yellow I have I have some filter and I
00:27:18
have a digital step attenuator to help with that level control or to gain
00:27:23
control um now i wanted to do a minus 60 that i have an input signal um uh after the antenna panel
00:27:26
right uh right at the input of my lna uh anywhere from minus 60 to minus 100 so i need a 40 db
00:27:39
dynamic range a little bit of this is made up but kind of in that range let's say that this is
00:27:47
uh kind of what you would expect out of uh the antenna elements in your array anywhere between
00:27:53
in that range um now the thing about it is you have to be able to be in that range and at the
00:27:59
output of this digital rxic right the only thing that's missing here by the way is the analog to
00:28:05
digital converter so i don't know the digital converter let's say it has to say about 10 dvm
00:28:11
of input power which is about it's one volt and if we assume 50 ohms it's one volt and whatever
00:28:17
right whatever it is that you need let's say you wanted to have the right power level okay
00:28:23
now i did a quick some analyses um and things like that and i figured out uh that i needed some more
00:28:29
games so i brought in this vga mimic it also has some game control um and uh i ran a few analyses
00:28:36
and I said, okay, at minus 60, this is what I have to set the two DCAs,
00:28:44
and at minus 100, this is what I need to set the two DCAs.
00:28:51
And in fact, in this workspace, if you guys get it,
00:28:55
you'll notice that I have an equation that I can figure out any value
00:28:58
for the DCAs between minus 60 and minus 100.
00:29:01
For this demo, I'm going to stick to minus 60 and minus 100 levels, right,
00:29:04
to not complicate things too much.
00:29:08
but you know you can go to any value in between right and even outside of here it's just that
00:29:10
my equations when I set up the workspace you'll notice that hey if you try to go below minus 100
00:29:17
it fixes the DC at 4 and 1 for example and the same thing if you try to go above minus 60
00:29:23
it'll fix the DCA values. Now the first thing that I did is I said oh I have let's say a 1
00:29:29
megahertz wide signal and we're going to run it through our spectra says our RF
00:29:37
analysis tool on the left it's a the lower spot lower power minus 100 dbm
00:29:42
input power and we're looking at the output and you can see I'm about 10 dbm
00:29:49
pretty close feel pretty good about it you'll notice that the noise level seems
00:29:53
kind of high come back to that in a minute and then and the minus 60 again I
00:29:58
got really close to the 10 dBm that I was looking for
00:30:03
Then the noise level was a lot lower. Why is that? Well, we started off with a higher input power, right?
00:30:07
so
00:30:16
The higher input power means I don't need that much gain. So I have lower gain
00:30:17
Remember I started with higher single level
00:30:24
So, going through the circuit, the noise does not see the same amount of gain as it does
00:30:26
when the signal power level is less.
00:30:33
So that's why the signal power, when you have minus 100, the noise power is that much higher.
00:30:36
You have much more gain here.
00:30:42
Now prove that in the next slide.
00:30:44
Here is the line up, or cascaded analysis of our receiver.
00:30:49
notice, right, the minus 100, you definitely had to give it more gain in order to get to
00:30:54
that output power level, and the output power level here is the green line, right, so we're
00:31:02
trying to get to 10, here's the 10, right, so you need a lot more gain to do that.
00:31:06
Other things on here is the cascaded noise power, those are the sort of purple lines,
00:31:12
and then these teal lines, that is the carrier to noise and distortion ratio that we talked
00:31:18
about earlier. I just like to plot those. We are convinced that unavoidably people like to see data
00:31:22
also in table format. We provide table formats and things like that. You can see here, I always like
00:31:33
to look at the center frequency, desired channel power, cascaded noise figures, character noise
00:31:39
ratio, and this TIMP stands for total inner mod power. It gives you an idea of where your inner
00:31:45
mods are being created, which is kind of a nice result.
00:31:52
I'll focus, won't spend too much time here.
00:31:56
Notice that the output power, again, somewhere close to 9 dBm,
00:31:59
actually more, closer to 9 dBm than there, to 10.
00:32:03
But this is the nominal value.
00:32:06
What happens when things start to vary?
00:32:09
We will look at that coming soon here.
00:32:11
Of course, this is a phased array.
00:32:17
And now we have phased array analysis, right?
00:32:19
We want a phased array, so why don't we go ahead and convert our schematic to phased array.
00:32:22
So I'm using the famous array RF amplifiers.
00:32:27
I replaced all my amplifiers by those, right, just to be able to do that.
00:32:30
Remember, I told you there was an advantage when you do Monte Carlo analysis using those, right?
00:32:34
And that's that the whole bank of all the amplifiers will be different,
00:32:39
will have slightly different values when you run your Monte Carlo, which is a huge advantage here, right?
00:32:43
and let's just for argument's sake say that that's where most of our
00:32:48
variability is going to come from right let's say you know that this is a
00:32:52
really good analysis a good way to do that
00:32:55
of course because we do phase array we can always show you the the antenna
00:32:58
pattern this antenna pattern for the nominal case
00:33:01
all right so here's the the Monte Carlo analysis
00:33:07
so on the left that's the desired channel power
00:33:10
remember i'm trying to get the same power no no uh output
00:33:13
no matter what the input power is between minus 60 and between 100 and 6 minus 60 right so notice
00:33:17
that the blue is for the higher power and the green is for the lower power and you can see that
00:33:25
the statistics looks very close which is a desired result here by the way right because we were aiming
00:33:30
for that remember that our average was our nominal was looking around nine or so you can see that
00:33:36
over the variations, at least the ones that I'm using here,
00:33:43
you can see some of them actually
00:33:48
go above 10, which is probably something you don't want,
00:33:50
right?
00:33:53
Probably driving your analog to the other converter
00:33:54
a little too hard.
00:33:58
And then on the lower side, maybe the signal
00:34:00
level's getting a little too low, right?
00:34:02
Maybe this is something you have to address, OK?
00:34:04
I didn't want to look at output P1 dB,
00:34:09
Because this is a receiver, most people are more interested in the 1 PD1 input, PD1 dB, or 1 dB compression point.
00:34:12
Notice that the lower power has a lower one, and the higher power has a higher P1 dB.
00:34:21
Again, that's due to the difference in gain between the two.
00:34:26
And the difference in gain between the two is about 30 dB.
00:34:30
So that's the difference that you see in this plot, about that difference.
00:34:33
right um let me throw in another plug for my pathway system design we used to not be able
00:34:37
to do two-tone analysis and our receive phase right we've been always able to do that and
00:34:47
transmit um and we did add the the capabilities to do that in this last release so now you can
00:34:52
do two tones see and receive two tones it's kind of an interesting thing because are the two tones
00:35:00
coming from the same direction or they come from different directions and things like that we
00:35:06
actually solve that problem uh so if you put them in the same direction with the two-tone analysis
00:35:10
you can put them coming from different directions even from different they can be different obviously
00:35:15
different frequencies close not so close in general it's a very nice general jet in general
00:35:20
uh two-tone analysis and you can do multi-tones right we actually support multi-tones
00:35:28
As you can see here, kind of the different groups, I'm going to zoom into the fundamental
00:35:32
areas in a second here, but you can see you can track the different harmonics and again
00:35:39
going for that equal output power between the two.
00:35:44
So here are the two-tones analysis when we run this.
00:35:56
Notice that when you have the higher output power,
00:36:02
it's definitely you have a lower IP3 than with a lower power.
00:36:10
And again, all this goes back to the input,
00:36:14
the gain of the two blocks, the two options,
00:36:21
you have lower power and higher power okay all right so let's say you're happy with your rf uh
00:36:24
design and uh you're getting ready to now uh do some waveform analysis and actually put some uh
00:36:38
digital beamforming right so our block here is uh our phase array all the electronics rf
00:36:49
that we've just been looking at.
00:36:56
That's this block, which we do through RFLink.
00:36:59
So RFLink is our way of actually incorporating
00:37:02
from the RF domain analysis that we were doing
00:37:06
into our digital or time domain analysis that we can do.
00:37:09
Notice that I actually have
00:37:14
a fully framed digital source here.
00:37:16
I've shown these before in other presentations
00:37:18
and if you want more information,
00:37:21
Please see the documentation, or I'll talk about some references
00:37:23
in a little bit about where you can get more information.
00:37:28
Notice that I have the same slider for the power.
00:37:33
And again, we can do it minus 100 and minus 60,
00:37:35
which I'll do in the following slides.
00:37:38
This block represents an antenna manifold.
00:37:43
Remember, we had an antenna manifold in the phase array analysis.
00:37:45
We also have one here.
00:37:47
This is the block.
00:37:48
And then we added the analog to digital converter
00:37:51
that was missing from our in the RF domain right we don't really have an
00:37:53
analog to digital converter in the RF domain that's a digital time domain
00:37:56
component I have a very basic built-in beamformer I'm showing you the details
00:38:01
down here the actual case and we've seen this a couple of times you know
00:38:08
customers want to come and use their own digital or baseband beamformer you're
00:38:14
more than welcome to do that there's several ways to do that but in this case
00:38:22
I'm gonna use our built-in one which is kind of nice and convenient to do here
00:38:27
and just kind of the way we will set it up I didn't do any scan angles I was
00:38:31
gonna do that but sort of you know have to limit in time so I left everything
00:38:35
as zero so no scanning we're not gonna do any scanning here this game block is
00:38:41
only to scale back down because I want to use the VSA software to demodulate my
00:38:46
signal so what happens is in the analog to geoconverter being a true analog to
00:38:51
digital converter right it's 12 bits so that output center goes between 0 and 2
00:38:56
to the 12 that's a really nice number so I I'm just used to seeing the VSA with
00:39:03
smaller voltages so I kind of went this way and felt that it was a kind of
00:39:11
Appropriate here. You can you don't really have to do in the VSA would be totally happy to still demodulate your signal, right?
00:39:17
It's just that the power levels look kind of funny. All right having said that
00:39:24
Here's again our lower power result. So this is the ones that we were running that minus 100, right?
00:39:31
And you can see then I'll put powers around a DBM kind of
00:39:38
level we're shooting for 10
00:39:45
and then EVM of minus 20 almost 20 not minus 29 BB looking pretty good right
00:39:47
spectrum and everything you can see a little bit of the silos coming up a
00:39:55
little bit but not too bad now when we go to minus 60 you can kind of see if
00:40:00
you focus on the spectrum here and the bottom left hand corner see the side
00:40:04
lobes came up right the addition channel you can actually I'm gonna switch
00:40:08
between the two you can see also that the constellation became a little bit
00:40:15
rounded again that's the non-linearities at the higher power affecting us
00:40:18
somewhat and then the EVM drop to minus 26.2 right and you can kind of see the
00:40:23
effects of that even in this CCDF plot you can kind of see the the effects of
00:40:30
that I'm gonna switch between the two I can kind of sort of see what's going on
00:40:34
there okay so with that that's the end of the technical presentation for today
00:40:38
before but before I let you guys go let me give you a few video references it
00:40:46
does if you're new to our phase array design or that's me or new to face array
00:40:52
in general not just with pathway system design the there's two videos here on
00:40:57
our YouTube channel from my colleague Dr. Matthew Maka if he covers some really
00:41:03
good phase array theory and you know how to avoid some of the pitfalls and his
00:41:09
videos and then the last two are some playlists from another colleague of ours
00:41:16
Anurag Bhagrava I hope I pronounced his last name correctly sorry Anurag I didn't
00:41:20
so again really good the system view phase array is step by step they're
00:41:27
short videos and he walks you through to other some of the finer details so if
00:41:34
you haven't seen those please I suggest I recommend you send them spend some
00:41:40
time to do that also if you ever are looking for a subject how to do
00:41:46
something in pathway system design even this with just to navigate the user
00:41:52
interface I know I have these short five-minute videos there's 28 of them
00:41:58
just treasure trove of really good information right you're stuck somewhere
00:42:06
oh I wish I could not five minutes just give me five minutes that will walk you
00:42:11
through and some of them are basic and some of them are some fairly advanced
00:42:15
topics you know about five minutes that's all we're asking all right with
00:42:18
that I want to thank you for attending today's webinar and listen listening to
00:42:26
my talk come visit us here's a link if you want more information about pathway
00:42:32
system design and again thank you very much Oh before I leave I want to remind
00:42:36
everybody please tip your moderator and the way you could do that is fill out
00:42:42
their survey they love getting the 100% participation in their survey so please
00:42:49
fill out the survey to make our moderator extra happy today with that
00:42:54
thank you so much and I pass it back to you moderator
00:42:58
- Idioma/s:
- Autor/es:
- Pedo Luis Prieto Zardaín
- Subido por:
- Pedro Luis P.
- Licencia:
- Dominio público
- Visualizaciones:
- 34
- Fecha:
- 19 de marzo de 2023 - 13:04
- Visibilidad:
- Público
- Enlace Relacionado:
- https://learn.keysight.com/how-to-model-digital-beamforming/how-to-model-digital-beamforming?elq_cid=5087815&cmpid=ELQ-24743
- Duración:
- 43′ 06″
- Relación de aspecto:
- 16:9 Es el estándar usado por la televisión de alta definición y en varias pantallas, es ancho y normalmente se le suele llamar panorámico o widescreen, aunque todas las relaciones (a excepción de la 1:1) son widescreen. El ángulo de la diagonal es de 29,36°.
- Resolución:
- 1824x1028 píxeles
- Tamaño:
- 229.35 MBytes
