TSP #101 – Tutorial, Experiments & Teardown of a 77GHz Automotive FMCW Radar Module

TSP #101 – Tutorial, Experiments & Teardown of a 77GHz Automotive FMCW Radar Module


hi welcome to the signal path in this episode I’ve got an interesting little module to show you this is them FM CW automotive radar that came from a car that had an accident but it seems kind of ironic that he had an accident even though it has an automotive radar in fact the person who said this to me told me that this radar helped reduce the impact of the accidents significantly but of course as with any mechanical device there’s at some point you just simply can’t stop fast enough but these things save people’s lives every single day so I want to take it apart see how it works whatever the circuitry inside look like what are some of the microwave components look like and at the end of the video I also have a demonstration of the Doppler effect and we’ll also go over the principle of operation understand the theory behind how these things are supposed to actually do what they do so it should be pretty interesting and exciting and I apologize to the person who sent this to me because it’s been with me for a long time but anyway I finally got to it I’m eager to get started let’s see what it looks like so I was in the middle of making my own slides talking about automotive radar making diagrams and equations to describe the behavior of a Doppler effect in an FM CW radar but then I came across this set of slides from rohde schwarz and these are promotional slides for some sort of advertising packages and it’s great because it describes exactly what I was looking for under public slide so I’m just gonna use these so thank you director or gossipy from one source of putting this together and the purpose of these slides were to show that Gordon Shores has instruments capable of analyzing radar signals from FM CW as well as chirp type of radars and they talk about the equipment and how they fit into designing and characterizing these systems but for us we’re only gonna use a small portion of the slides for what I want to describe so here’s an example of the car with a whole bunch of different sensors in it and not just long and short-range automotive radars but also ultrasound cameras and lidar s and a whole bunch of other things so a stuff driving car would combine information from all these types of sensors to create an awareness of where it is sitting in the environment so we’re only interested in the millimeter wave radar so let’s only talk about that so what are we trying to do well imagine if you have a car that’s not moving stationary and you have a target that’s not moving also stationary and I want to find out the distance between the car and the target now that should be really easy to do because we know the speed of light we know how long things take to go back and forth and then we can find out from that the distance we have to the target but let’s make it a little bit more interesting let’s imagine that you have a railroad that transmits a ramp on off frequency meaning a CW signal a continuous wave signal that whose frequency is changing linearly as a function of time so ramp like that so if you’re starting on frequency F T and we go by a delta bandwidth so BW and then we reach a maximum when we stop and the time it takes for us to get there let’s call that TCP I said it gives you the slope of the ramp and the signal leaves the car bounces off the target and comes back so what comes back is also going to be a ramp obviously because it’s the same Sigma being reflected but it’s going to be delayed in time because it’s a physical distance between the target when the radar is going to create that tower that because of the speed of light so if I were to measure the frequency at any given time at any instance of time I’m going to observe a difference in frequency between what I’m sending and within what I’m getting back let’s call that FB that difference is because of the delay and the fact that we are sending a ramp so we can measure this FB quite easily and I’ll talk about that a little bit later now let’s see what we can use that information so based on this diagram that you just saw I could relate all these parameters together so the ratio of FB to the bandwidth which is the Delta F is the same as a ratio of town to the time it takes to for the ramp to ramp up so this equation we can easily calculate let’s rearrange that and get tau out and then we were left with this equation tau as a function of these parameters all of which are in our control now let’s plug that back into this range equation the range equation is nothing more than the speed of light divided by 2 times the time it took that’s pretty straightforward and then we can get this equation at the bottom and we can rearrange it to bring FB out which is a measured parameter so we get F be equal to 2 over C bandwidth over TCP I and on so all of this are in our control like well except for C because C is the speed of light but then with the control tcpi we control on the ranges but we were looking for an FPS but we were measuring so this is perfect based on this equation I can find out what R is as soon as I measure F be there also know how important bandwidth is and tells you how your resolution in this test but let’s put that aside so having all of this and you should work perfectly fine but this is for stationary targets on a stationary car as soon as you start moving things will look quite different so why will things look different well it has to go with this guy well I should say this guy is not his fault we have Doppler effect this is Christian Doppler is the person who discovered the Doppler effect Austrian physicist who lived from 1803 to 1853 unfortunately died quite long is only only 50 and thus we described this Doppler effect and the Doppler effect this isn’t just using cars and radars this foundation in astrophysics for determining the motion of galaxies and this is how we know whether they’re accelerating or how fast they’re moving and whether they’re moving toward us or away from us and we know that because Doppler effect happens to be a function of how objects move to each other at another what frequency they move toward each other and the opéra effect is something you see an experience every day every time a car comes towards you moves away or a train comes towards you moves the way you can you can hear a shift in the frequency as they approach you and as they move away and that’s the Doppler effect because the upper effectors are fighting waves and it will work on e/m as well as it will work for sound waves so if you are moving towards a target the waves get compressed together so you’re kind of catching up to them so the frequency goes up if you’re moving away from them they get stretched out so their frequency goes down so now that we know this and there’s an equation that describes the ratio of how much the wave changes as a function of the speed and the wavelength and that is pretty well understood by Doppler you describe that so we can use that information and now let’s look and see what happens in the inner situation of the car so now imagine a car is not moving but your target is moving towards you so there’s a relative velocity between your car and the target you want to measure and we’re going to exactly the same thing we’re gonna send our ramp out and the ramp is going to come back at this at some Delta Tau delay time and we’re gonna measure some difference in frequency let me make that measure but the problem is that this difference in frequency we are observing isn’t just because that this the target is away from us but it also is because the target is moving so fp1 the measured difference in frequency is a function of two things now at our which is because of the distance an FD which is a Doppler effect so if I were to go and plug that back into our original equation now we have the original component from before but now we have this 2 over lambda times VT which is the velocity which is a Doppler effect so now there is two unknowns in this equation because we don’t know what VT is and we don’t know what R is so we cannot solve this equation on its own and that’s going to give us some trouble so we have to find a different way of resolving and separating the difference between FD and fr and it’s quite straightforward people already figured it out and how do we do that well let’s continue ramp back down so we go up with the ramp and then down with the ramp now when we do that something interesting happens because we have a Doppler shift the the observed frequency between FP 1 and F DF B 2 is going to be now different and that’s the trick of course as long as the direction of the object isn’t changing in the middle of the ramp but the ramp is very fast much faster than the movement of the object you are trying to to measure so we have to make an assumption which is pretty easy to do so now because of the fact that we have to F be measured fp1 and fp2 now we have two equations these two equations now have the same parameters in them but there’s going to be a difference between fp1 and fp2 now we have two equations now we have two unknowns we can solve for both the range and the velocity at the same time and that’s the trick which means that F MCW radars send out a triangular wave ramp of frequency that’s exactly what they do and this is exactly how it works looks really straightforward once you describe it and understand it but that’s the foundation of FM CW radars now there’s a lot more going on because there’s multiple targets that can be there there’s multiple reflections there’s a lot of other things that need to be resolved in the ESP and sometimes you have to have phased array so you can change the direction of the beam so you can look at some places and if there’s uncertainty you move the beam around and try and resolve that so there is a tremendous amount of research and complexity in digital signal processing and the background of radars not just manipulators but also lidar and this is I’m not going to cover that in this video of course but that stuff is even more important than this but it’s the foundation of how it starts and what is the basic principle of operation of these radars so let’s go back take a look at the hardware and see how it works so let’s take a look at this interesting piece of technology now obviously the plastic that used to go over the antennas is broken off and it’s quite a bit of damage on this and that top cover obviously has to be plastic because it needs to be transparent two millimeter wave frequencies but the body itself is made of cast aluminum and there are a lot of interesting features on this body there are mounting holes obviously some thermal management and a connector design to get the signal back out of this into the cars computer you can see this connector with all the pins inside has to be waterproof and sealed properly now automotive radar applications the specifications for the a6 for shark temperature and other things are much tougher than from consumer electronics it makes the design cycle of a sticks that go into these applications quite a bit more difficult and sometimes further apart in terms of these difficulties so it’s cool to take a look and see how much effort they have to put into just design in this chassis for to survive in a car so let’s go ahead and take this off so we can see what’s inside of this I’m not going to touch the bottom and we will find out in a second why and here’s the inside of this if you look here you can see several pedestals this obviously had two boards in it but the other board which was broken I did not receive but these pedestals is a little bit of thermal paste still left or here which would touch the other ICS that were in there and perhaps some DSP or some computational platform power management and so on we’re on here which are now gone and they will interface with this connector and go out of the car so really cool to see the data 2015 or 407 so it’s not that old yeah so this is a pretty neat I’m a mechanical engineer but it’s still pretty cool to see at the amount of effort that goes into designing one of these so let’s put that aside and let’s take a look at this and let’s flip it to the other side and check it out all the radar things that you need they are all here and we’re gonna take a close look at them we’ll take a look at the datasheet for this part and a datasheet for that part but aside from those which are off-the-shelf components here’s the connector that connects to the other board and another little part we’ll take a look at the all the radar is handled by these three icees now a couple of things to note these Asics are not packaged they are not flip-chip they are just water bonded directly onto the PCB as low-cost as possible and there are three of them and it this again goes to the heart of the fact that these things don’t have a very quick refresh cycle because once it works and meets the automotive air of capabilities you don’t want to take him out these things are sold by the millions and they make tons of money from these and these look like to be done in the Infineon city by similar so they’re just a ciggy HPD process and yeah there’s three of them and so what I was going to do initially I went and looked at the microscope very closely and found reverse engineer didn’t look at the feed line and all that but unfortunately I can’t talk about that because it’s getting a little too close to what I work in my professional life so I can’t go into the details of the ASIC design and reverse engineering and talk about the feed line and the antenna connections and so on because it’s just going to be too much of a conflict with what I do but nonetheless there’s a lot of other stuff we can cover just enough to know that these three Asics are basically responsible for the radar and you can see in a couple of signal interactions between them these two ICS are identical and that one is different and this signal coming out of here you can see it’s split and it’s fed into the two other ICS there and several points where the signal leaves this IC and goes into the board and comes out of the other side into the antennas and the transition that it takes is also really complex so we can look at that under microscope and figure out a couple of these components now will be very interesting to see if you recognize any of these microwave control components on here and see if we can do any measurements and to find out what frequency this is supposed to operate on and I don’t mean electrical measurements just dimensions because these are microwave things you should be able to figure out the frequency by the size and the dielectric constant of this material so once the signals leave these ICS and go to the other side they’re fed to the antennas and if you look at the antennas they’re also really beautiful and they look like a 1d array of dipoles on each line there and making a 2d array in total some of these are rx some of them are TX this is either capable of multiple beams at the same time which i think is unlikely it’s most likely just to us beam forming and there’s some interesting differences in the distances between these very cool matching that work yeah just a really neat to the array and on this side again the signal is bouncing coming back then are processed by those RFI C’s so let’s take it one step at a time let’s look at this under the microscope see if you recognize any of these microwave components and it’ll be cool to take a close look and talk a little bit about these ICS and then I have an experiment we can do with actual table type Doppler radars now before we look at this now let’s take a look at something else that’s really neat that somebody sent me now you’ve all seen these rulers before I’m sure this is an eighth of food ruler these are PCB rulers that have let’s say footprints of various components on them as a reference essentially and people seem to love these a lot of people make them and I quite liked them myself but somebody sent me a pair of rulers which were just super cool and I apologize the person who sent me these because they send this to me a long time ago and I’ve been waiting to make this video so I can showcase this and I want you to take a look on this and see if you recognize any of these components on that automotive radar so these rulers unlike traditional rulers have all the microwave components on them and they are just awesome so all the components that you see me talk about on the channel when we take equipment apart they’re all here for example directional couplers here here’s the Wilkinson dividers various stops you know I talk about radio stops as well as a various filter here’s the capacitor is the butterfly stubs we see that quite a bit a whole bunch of quarter wavelength transmission lines on on here and as well as the holes there on this side and whole bunch of other things very cool stuff usable of Ivaldi antennae I actually use with all the antennas all the time in our experiments we’re gonna see them again today some filters happen filters which I just talked about in the previous video I made somehow see stop filters over here you can see various types of antennas are also here’s the wideband dipole there here’s an ultra wideband monopole there is a micro strip filter detector is a hybrid ring copper or a rat race coupler and folded dipole these are awesome everyone who’s ever seen these has just loves these rulers and so I’m gonna put a link in the description of the video if you want to buy this for yourself they’re pretty awesome and obviously they’re just good rulers also but if you work in microwave we just like these components there to show people that the different kind of microwave components you can directly put on PCBs just keep in mind look at the dimension of this and there is one of these here you can’t even see it from that angle so I’m gonna talk about this and we can see how these things just scale with frequency so you see if we can recognize anything else from these rulers on that board so after the microscope we go alright look how beautiful this is so this is one of the IC is the one that’s different from the other two now so you can see here a couple of very interesting features first of all here’s what rat-race coupler and we saw that on the ruler as well and in this case they’re using this to do differential – single enough conversion and single ended to differential conversion because the inputs to this are fic all seem to be differential so you can see that here’s a feed line to the antennas and it becomes differential going into the chip or vice versa differential coming in going to become single ended and to the feed line so you can see there are multiple feed lines coming in and out so this again leads to believe that this is a phased array of some kind or it can do multiple beams now if I go follow this other line you can see we have another rat-race coupler by the way these capacitors here for tuning we can continue on and you can see they have a test structure of some kind or a through a line of some kind and there is another component that we should be recognizing there it is I’m sure you saw this on the ruler as well this is a branch line coupler so it divides the signal so which means that the signal flowing either in this direction or in this direction would be split into two equal power so you can see one goes this way and the other one goes this way and these two artifices are identical so let’s look at one of them and look at this one so the signal comes from the branch line coupler becomes single differential again goes into this I see again connected with multiple places to the antenna feed line sir – terminated points so these two are outputs are not used or inputs are not used we continue on and on the other side we can see that all of them are indeed used so this is a complex structure and how the signal goes from this to the antenna I can talk about it it is it’s wonderful I have to say that this is quite a bit I’m sure there’s some material you can find out I just can’t talk about it and also what’s on the RFI see sadly can’t discuss that but I don’t know that you can see how wonderful and beautiful this design is now let’s go ahead over here and do some measurement whoops if I can get this thing to actually cooperate with me and move on to the center the microscope sorry about that there we go so what was I gonna do here we wanted to look at this branch line cover now this special coupler is perfect for us to measure its dimension and by measuring its dimension we can figure out the frequency of operation because we know how branch line coupler works sorry but this is not very good video-making here so let me is the one this perfect so here’s a branch one couple in the middle now I wanna measure its dimension is now a branch on coupler uses quarter wavelength sections so by measuring that I should be able to estimate the frequency of operation of this pressure on coupler so this put the line on that I can get it ah that is not good this thing is trying to catch the edge for me it doesn’t do a very good job out there you got cut so somewhere up to let’s say right there that’s good enough for me so that distance is about 0.49 two millimeters let’s say you know half a millimeter or so so it’s the quarter wavelength and half a millimeter now the dielectric constant of this material is most likely somewhere between 3.8 to 4.2 so we can say you know roughly for 3.9 now we can compute from the dielectric constant and a quarter wavelength what frequency it is and the equation is obviously very straightforward so we’re gonna divide 300 divided by the measured distance times 4 so 0.49 to 0.49 2 times 4 that’s the quarter wavelength times the square root of the dielectric constant of the material now I think the dielectric constant is about you know that’s a 4 if I multiply that I get 76.2 gigahertz and that lines up perfectly with the 77 gigahertz Automotive radar standard so this is indeed working 77 gigahertz and I could have done the same thing I could have measured the dimensions of this rat race coupler and that rat race couple dimensions are also related to the wavelength except if you have to measure it radio list is a little more difficult but this branch line coupler tells us everything we need to know about the center frequency that’s how easy it is to be able to determine that and everything else is quite server look at this ASIC so beautiful and I definitely see by CMOS or CDHP T process yeah so very nice and on the other side the antennas are not that interesting to look at under the microscope you’re not going to see anything special I can’t lay this flat but you can see some details of the material used which is also pretty interesting than individual dipoles there so anyway so let’s go and look and see what about the ICS that are there so let’s zoom out a little bit and if I can get this focus here’s our first IC 88 to 83 so that’s one of them and the other one is H 701 so let’s go to the internet and find out what they are alright so let’s take a look at the par star on this force the first one was an h MC 701 which is an analog devices part used to be hittite and this part is a 8 gigahertz 16 with fractional and PLO and it’s exactly what we would expect to find this particular PLO is intended for FM CW sensors as well as automotive radar so it has the ramp functionality already built into it and if I go that we can see the block diagram of it and it’s clear that there is a sweep control block built into it so the sweep control block allows you to define your ramp the speed of the ramp the parameters of it and it will automatically perform that ramp function on your pls closed-loop system so that this your signal will come in from the RFI C that’s on the board and then it will go through the PR section and the VCO will be controlled directly from here and this allows it to create the ramp so it makes perfect sense that this is found on this board the otherwise C was an ad what was the part number 88 to 83 and this is a 6 channel LM a programmable gain amplifier an anti-aliasing filter with 180 C Block into it so it has multiple channels of receivers and these are Naser a low frequency chooses after conversion by F and then they have gained control and anti-aliasing and then MUX and a 12 bit ADC and this is exactly what you would use for radar as well because you have this multiple channels coming back from the different sections and terror sections that are down converted and then you can create either a multi being or a phased array or can figure out in digital domain the direction of the beam and so on directly from the data that’s coming out of this a TV camera so it’s fantastic and in fact if you look captain’s adaptive cruise control collision avoidance blind spot detection and so on so this is again optimized and designed specifically for this type of application it said 72 mega samples per second for 12 bits not bad 70 millivolts and then finally the last is just a last ICS used to 24 makers reel-to-reel amplifier with shutdown options just another so nothing to worry about on that one so it looks makes perfect sense and these components are all working together in order to create this radio functionality so it all now comes together and I hope that you have a clear view of how this system actually works now there is one other thing I want to do we should be able to build our own little not a full automotive radar but a Doppler radar in this very simple way so I’ve set up something on the bed let’s go take a look so pooch has been kind enough to set up an experiment for us and here it is so this experiment will demonstrate the Doppler effect it’s not a fool of M CW radar but it’s the simplest one that I could put together so there is a signal and a low signal so a CW signal that’s coming over here goes through this Doppler so here is at 4 gigahertz after the Doppler is at 8 gigahertz going through the amplifier we have a large 8 gigahertz amplitude signal here that is split into two one part goes into this antenna so radiate out and the other side goes into a yellow part of this mixer and the other RF port of the mixer is connected to the other antenna and then the if’ port is taken to the spectrum analyzer so I’m using these two instruments here this bottom one would generate a 4 gigahertz signal for me and this one is looking at frequency content between zero Hertz and one kilohertz so anything that makes them between those two frequencies we should be able to capture so the idea is that 8 gigahertz signal leaves this antenna and if it reflects against something it will reflect back onto this antenna over here so comes out of here and goes back into here now if there is a Doppler effect there will be a difference between the frequency that this one receives and the difference in the frequency this one is transmitting and the difference will be down converted by the mixture and we should be able to see it over there on that screen there so let’s see if we can replicate that effect and see what happens so here’s a piece of copper reflective obviously will reflect the waves coming out of the antenna now I’m gonna just hover it over the antennas like so if I hover over it nothing happens because this is not moving so there is no Doppler effect but watch what happens as I shake it you see the frequency content that’s the Doppler effect so if I move it up I get a response if I move it down I got a response so this movement is difference in velocity between the antenna which is stationary and this piece is creating this Doppler effect so I can continue to do that let me start from here somewhere you can see go down and up and if I go up up and down like that you can clearly see the Doppler effect if I hold it stationary the reason of the opera’ effect in a stationary mode the two frequencies are the same so they’ll get mixed down to DC and you won’t see them because it’s a DC is that the all the way at the left side of the screen there but if I just move it up and down you can see every time I move you get a spike and if I could do this perfectly steady at exactly the same velocity I would get a single tone which of course I can do I can try but no matter what that move that peak is gonna move around no matter what I do but there is your Doppler effect I mean it’s so easy to demonstrate and all it takes just a couple of components and you should be able to do it and you don’t need a fancy spectrum analyzer for this you would just need a simple digitizer but this is why that I see that we saw was using these twelve in a to D converters is to capture exactly that signal along the FM CW where from the spin transmitter so this is a very simple version because it doesn’t have a frequency ramp so it won’t be able to do a full as FM CW but it certainly can capture the Doppler effect then it’s really quite amazing that you can see this physical phenomenon and I hope that you enjoyed that so that’s it I think this would be a sufficient for you hopefully to get you excited to go and read more on this type of on circuits and applications so if you’d like to support this website you know what to do you can give it a thumbs up leave a comment subscribe to it and of course I have a patreon channel which you can contribute to and it will help me buy more equipment do more repair videos in the future and I hope you enjoyed this one I’ll see you next time


Leave a Reply

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