TSP

In this episode Shahriar takes a detailed look at two different automotive 77GHz radar modules. Each module design is presented along with the overall architecture and functional blocks.
The working module measurements are presented with discussions on how to capture an FMCW waveform at mm-wave frequencies. The teardown of the faulty module reveals the overall synthetic aperture design, antenna placement and design considerations and the digital backend processing. A close look at the Infineon 77GHz RFICs are also presented.
Please consider supporting The Signal Path:
www.TheSignalPath.com/Donate
/ thesignalpath
Other links:
www.TheSignalPath.com
/ thesignalpath

[@Thesignalpath]

*To clarify a few things and avoid confusion on terms such as synthetic aperture, MIMO, and SIMO:* 1) You could move a radar aperture with precisely controlled steps and take individual images. This would be as if you had a repeating pattern of the antennas in space. The collection of the data can then be processed at the same time to create an image equivalent to a much larger aperture and thus “synthetic aperture”. 2) In a MIMO system, multiple TX antennas can transmit coded radar waveforms at the same time and the receivers can use digital beamforming and MIMO processing to differentiate each TX channel. Since each TX channel is carefully spaced apart, you can equivalently create images with a wide field of view. 3) In SIMO you do the same thing as MIMO except one TX at the same time. Here you also create a larger equivalent aperture. Although I did say that this technique also creates a larger synthetic aperture, I want to make sure this is not confused with the moving antenna situation. This is easier than MIMO because you don’t need the complex coding of the transmit waveform. *As long as you keep the above in mind, the operation of this particular radar module should be clear. The equivalent aperture in all cases is larger than the physical antennas.*

[@vikramkumarp]

Please try NXP chips all TX and RX and processing in one chip

[@EEVblog]

Measuring 77GHz like a BOSS.

[@user-sd3ik9rt6d]

3.9 cm at the speed of light.

[@antoineroquentin2297]

@@user-sd3ik9rt6d 3.9 mm!

[@indian.techsupport]

​@@user-sd3ik9rt6d*3,9mm or the antenna would be way longer

[@greekstraycats]

6 GHz I am done .... 😞

[@tubosolinas]

@@user-sd3ik9rt6d 3.9 mm 😊

[@ryanwans391]

The only content worthy of pausing my studying for final exams for

[@gentleclash]

exactly lol

[@JonaJona]

I love how you explain many of the RF concepts in a much more approachable and thorough way compared to some of the older videos!

[@pupperozucca4994]

He was born in RF planet, thanks for choosing planet earth as your home.

[@andymouse]

LOL!!

[@appercumstock3017]

Master Course 77GHz Radar in 33 Min. Outstanding!

[@cyril1522]

2 GHz every minute, fantastic

[@FindLiberty]

...enjoyed best at 1.25x speed. Pure gold.

[@boyinlove2k]

Your RF knowledge never ceases to amaze me. That was so interesting - thank you.

[@pyrokinetikrlz]

insanely good video! Those photos of the RFICs were mind-blowing!

[@guerrillaradio9953]

You perfectly fill in my general EE classes with the high end RF knowledge which fascinates me! Now I only need a channel like yours for FPGAs to fully satisfy my need for extra input! XD

[@lbgstzockt8493]

One of your best videos! I especially love the IC analysis, you could go into even more detail with those.

[@mohsentabouna6401]

I wish I had the chance to attend one of your courses back in toronto. you must be an amazing professor

[@xtlmeth]

You never disappoint!

[@user-ig8pp3st4v]

That stitched real-time mode is really nice!

[@cbiz8]

Excellent analysis!

[@w5cdt]

Nice tear down! Thanks!

[@willernst8376]

Such a good video!

[@gerryjamesedwards1227]

Ti have been making a big thing about their mm-wave radar being put into cars and robots, but I haven't seen anyone playing with them yet, er I mean testing them, so I'm excited for this!

[@EEVblog]

I've got one of those TI mm-wave radar devkits somewhere

[@gerryjamesedwards1227]

@@EEVblog have you done a video on it?

[@catalino8010]

i find so beautiful the science of the RF

[@leviathanx0815]

Ah yes... that's one of the projects I was part of a few years ago, or rather an earlier generation of it; my project already had Ethernet. But it's interesting to see how people try to reverse engineer these. These TI SoCs were custom defined. That is, if you are a bigger customer you can select a base chip and tell them what modifications you want. Like you could let them disable a core in order to pump up the clock engine and get more speed on a different core which would be more beneficial for your application. Their full datasheets, as well as the one for the Infineon analog radar chipset, are typically NDA and therefore not accessible to public. Not much of a state secret as this is what almost every ECU supplier does in the end for specialized projects. My latest project involved fully digital radar SoCs. Those are beasts on a whole different level once you manage to tame them. (Also, no need to ask me for in-depth details as everything is NDA- and IP-protected.)

[@antoineroquentin2297]

Thanks for the insights, very interesting.

[@PCBoardRepair]

radar is cool....i worked at Siemens Milltronics as a tech in thier Radar units like their LR250

[@OxTongue0]

Thanks for your efforts, it was interesting for me.

[@falconeagle3655]

Awesome content

[@sampad3204]

awesome video

[@GurpreetKakar4873]

First module was built by my company. Nice work.

[@michaegi4717]

This sensor looked really special to me (for a automotive sensor). Can you give some background? What's the purpose of this, it doesn't look like a series sensor to me.

[@dtiydr]

@@michaegi4717 "Built" is not necessary the same as "designed".

[@michaegi4717]

@@dtiydr Yes, I also don't design the sensors that my company produces. But I still know such essentials. :-)

[@szymon0900]

​@@michaegi4717 What exactly strikes you as "special"? To me it looks like usual modern automotive radar module, but maybe I'm missing something. Thanks!

[@michaegi4717]

​@@szymon0900 Hi, those that I usually handle are smaller. I miss heating elements for the radom (or the connections were just not shown in the video and the heating is in this thick part of the radom? I also wondered why the part in front of the antenna is especially thick, as this generates additional disturbance. I also wondered why the housing is made from plastic, as I imagine that it's hard to get rid of the heat. The CAN connector looks bulky to me and I wonder if this is really water prove. I can't see how this will be mounted, there are no screw holes and I can't positions for any clips. All this caught my attention, so I asked, because I thought that GurpreetKakar might be able to give some background.

[@alklapaxida850]

great video thankyou.

[@kouidriabdelkadir6600]

Good work. thanks

[@luigicirelli2583]

brilliant

[@Baatory]

Thank You

[@ornithopterindia]

👍Thank you sir.

[@McTroyd]

This is awesome stuff. Looking forward to the reverse-engineering of the blindspot detector. 👍

[@xDevscom_EE]

It's pretty crazy that all this technology is now commonly available in the cars/bikes just to detect proximity to objects! I wonder if it's also a valid attack side-channel, if someone very dedicated comes next to a car with a specially purposed mmWave signal generator? And drivers still might ignore the engineering and drive distracted while texting :-D

[@obsidianjane4413]

Its not a communications channel. Any signal you send would get chopped up and mutilated by the signal processor. I think the most you can do is jam or spoof the radar into thinking something is close or far to the radar.

[@Thesignalpath]

Indeed jamming is the most you can hope for. In some really rare situations one may be able to "fool" the radar - but I would imagine the DSP would run multiple ambiguity resolving before making decisions.

[@liam3284]

One way to stop a car with this tech is to open an umbrella face. The car "sees" a wall. Useful if crossing a multi lane where drivers dont like to yeild.

[@keithhearns3783]

RF circuit design very interesting

[@neonkev7866]

Keysight should send you a 110 GHz UXA with (at least) the 2 GHz digitizer so we can get to the bottom of that 77 GHz radar! Do analyzers like the UXA or FSW use an unpreselected path for the internal wideband digitizer, so you still need to deal with mixer images?

[@Thesignalpath]

The R&S FSW has a pre-selector YIG up to about 90GHz. But that instrument is about half a million dollars. :)

[@tommihommi1]

the preselector "only" goes up to 85 GHz apparently :D

[@neonkev7866]

@Thesignalpath That's pretty impressive. I found a quick app note that describes a 5 GHz BW analysis setup. You need the FSW85 along with an external 6 GHz scope for the digitizer: scdn.rohde-schwarz.com/ur/pws/dl_downloads/dl_common_library/dl_news_from_rs/219/NEWS_219__01_FSW85__EN.pdf Not a cheap setup but you can get leveled and equalized FMCW analysis at 77 GHz in real time!

[@0io1]

They must use the unpreselected path for wideband capture, the internal preselector only have 80MHz bandwidth at 50GHz. I think the preselector only useful for out-of-band signal check in very wideband capture.(Narrow Sweeping) With good frequency planning(wide IF bandwidth eliminate sweeping,good LO&IF Frequency for IF filtering,non-harmonic mixer), If no out-of-band signal near origin image frequency, no image assured, so we can bypass preselector. But SAs use harmonic mixer at high frequency ,and it's difficult to use in wideband capture

[@tommihommi1]

@@0io1 very good points, thanks. Now that you say it, I saw a block diagram of the FSW in a presentation at some point, and you're right, the preselector is bypassed.

[@bfx8185]

Nice content with great explanation indeed! I would like to know also some parts numbers like those 77G TX/RX chips. It's not so visible in the video. Your ceiling will be too low soon 😀

[@bfx8185]

Another topic would be as there is probably lot of crashed cars and those radars still intact , how to re-use it for something useful :)

[@smahax]

Fantastic You could have use Frequency mask triggering in real-time to see what was happening around this PLL.

[@paulpaulzadeh6172]

Now , some useful stuff, is it FMCW, or pulse radar? Phase array ?

[@maartenlont1384]

The two LO signals in the RX chip per channel could be used for a bist. Maybe one is coupled to the LNA input and one drives the mixer. This way you could test the functionality of each Rx channel in-situ.

[@Thesignalpath]

This is likely and a conclusion I came to after the video was posted as well. Thanks!

[@farhadmodaresi4182]

Shahriar, Can you do a review on the TinySA spectrum analyzer? Not everybody has the capabilities to push that device to its limits.

[@AndrewZonenberg]

When I first saw it, I was thinking that the TX antennas were being used as an electronically scanned array of some sort: fixed phasing in the vertical axis, but steerable in the horizontal axis by adjusting phase of left, middle, and right elements. How can you tell that's not what's going on here? For example the center chip generating the waveform and driving one antenna plus the LO, then the right channel echoing the waveform with variable phase to each of the right two antennas?

[@RubinGrolsch]

They're too far apart, any scanning will immediately form a bunch of grating lobes. The Rx antenna can digitally scan though, since those elements are spaced closer together

[@AndrewZonenberg]

@@RubinGrolsch Yeah I was originally thinking the left side was a 4 element TX AESA with four separate IF streams coming from the other board. I couldn't figure out what to make of the weird non-uniform spacing on the right.

[@camfocus8888]

Yes, you are right!

[@IvanStepaniuk]

Why are the TX antennas spaced like that? (two are closer, the third one is slightly apart)

[@FindLiberty]

Adds richer focus sauce to the phase for enhanced data details received then compared. Spicy.

[@yaoyucao9601]

Is it possible you tear down the Ericsson mini link 6352 RF board? Thanks

[@donepearce]

I've worked with a lot of circuitry like this, although most of my work has been at a slightly lower frequency. But I have to say I am not too happy about the way that RF board has bee etched. Detail is poor and I think the high impedance lines connecting the patches are over-etched. I found that I had to be fussy about what PCB house made the boards, and they couldn't always maintain quality.

[@jessiepooch]

@donepearce And yet...

[@donepearce]

@@jessiepooch And yet, maybe ultimate performance is not an issue here and they can accept poor matching.

[@KenPaisley]

I spit my coffee out when I heard "blind spot detection for a motorcycle".

[@syuan1038]

The glitch seen between two bands in the spectrum might be the result of using different chirp slopes and bandwidth to resolve the speed ambiguity.

[@halcon2134]

yes

[@ctid107]

Looks like there is a solder bridge at the top of the AFE5401 @ 15:54

[@dtiydr]

Just a thin strand of fiber.

[@Roboticgladiator]

I would expect the radar to be doing random frequency hopping to avoid interference from other vehicle's radar.

[@R2AUK]

🤯👍

[@diyemc7206]

Are you sure, this cover on top of the microwave ICs is "just" to stop unwanted emissions? Non-potted ICs are susceptible to light flashes (those cheap chip-on glad LCD drivers usually reset with a photo flash). I'd guess, with the translucent cover, this is at least one reason for this additional part.

[@SeanBZA]

Would say that is a happy side effect of making the cover out of a carbon and iron oxide loaded thermoplastic, likely HDPE. Likely a thermoplastic that is extruded and the biaxially stretched to a thin film, then gets vacuum formed to the shape, and then die cut out. That is very cheap, and automotive is all about cheap, cutting a tenth of a cent every time. The thermal compound pads are more likely there to provide thermal mass, as the main chipset likely only operates in pulses, not continuously, so needs the large thermal mass to absorb the heat pulses as the main DSP does it's half second wake up and few scans, then goes back to sleep, and the large thermal mass allows the heat to dissipate. You can see in the bike one the same, pulsed operation, so the blue thermal transfer compound can absorb the heat and slowly spread it out to the large area of the case. After all a blind spot sensor really is a slow reacting sensor, as the vehicle approaching is doing so at a low relative velocity, and it only has to look for a metallic return that is large enough, in the area covered by the beam formed by the transmit and receive antennas, and then have the CAN interface send a message to the BCM of a vehicle detected, so that the driver display can show a flag, or a tone, and the SRS system knows there is a close in object, so the potential for collision is there. 6 wire interface, 2 are power and ground, 2 are the CAN interface, and the other 2 are for the LED behind the mirror, to light up to indicate there is a vehicle there, and also likely also to drive the turn repeater LED, as it makes little sense to run an extra wire up to the mirror (extra cost, extra wire, and the extra test steps for the incoming inspection to do to verify the loom, plus the simple cost of placing this extra wire into the channel for it) when you already have a CAN transceiver there, and plenty of spare processing capacity in the firmware, plus plenty of IO pins unused, to handle the extra broadcast messages of light on at side, and light off. Heated mirror is simply a PTC element that is powered with the mirror, so that it draws max power at under 5C, and then heats mirror up to 50C in use. With that cover, another thing that affects the unit is the actual reflective layer on the mirror, that also likely has been specified to be avoiding certain thicknesses of coats, and also has a minimum thickness coat, so as to be a good light reflector, and also a max thickness coat, to handle attenuation of this 77GHz signal. Will guess the mirror is a critical part of the system, and the wrong one will likely result in the system not working well, or having continuous faults recorded for signal strength.

[@gacherumburu9958]

👍👍

[@muctop17]

13:50 I'm sure they do coding! Or how do they avoid interferece from other systems close by on the road? Spread spectrum or chirp modulation?

[@tonykara]

How did you acquire all this ultra expensive equipment?

[@Thesignalpath]

Onlyfans. ;)

[@xabhax]

Motorcycles have blindspot monitors?

[@jessiepooch]

@xabhax Diy add-on approx. 280gbp

[@thecriss88]

Dude, you are smart.

[@genericcheesewedge4870]

what is the purpose of converting the signal passing from the main chip to the other ones from differential to single ended and then right back again?

[@Thesignalpath]

A differential co-planer pair of wires would likely be too thin and therefore not only lossy but also difficult to realize on PCBs accurately.

[@0MoTheG]

Conditions inside and outside of the IC are different.

[@farzaadkhaan]

Hearing a beautiful Persian accent 😍😍😍

[@ZergZfTw]

It would be awesome to see you take apart an AESA radar, but I doubt that would be allowed on KZbin.

[@Thesignalpath]

Sure it is allowed. I have taken apart and reversed engineered 60GHz AESA on the channel before. :)

[@ZergZfTw]

@@Thesignalpath Video #207? I was thinking more about the low POI and anti jamming stuff that is on fast jets and navy ships.

[@ionstorm66]

Only ITAR on AESA is for space craft/satellites.

[@lbgstzockt8493]

@@ZergZfTwI think the hard part there is getting one, I don’t think they end up on eBay all that often…

[@Thesignalpath]

You meant military not consumer. If I get my hand on one…

[@calopii]

Very good reverse engineering. I am actually working in development of such automotive radars.

[@worroSfOretsevraH]

Hi. Could you make a video on mobile phone RF side? Infrastructure. For example in a very crowded city with 1000's of people using their phones simultaneously, you would need that many channels on separate frequencies. What amount of bandwidth would that require. How it is done in real life? Thank you.

[@earlyadapter643]

It is insane what complex circuitry is used and what minimalist user interface is behind. I'm sure this radar could do some rudimentary imaging and finally drives one LED on the dashboard, but even a display as the ones for the aircraft nav aids kinda glideslope/localizer would be too much to distract the driver. Ever more complex things are impelemented to keep the user stupid.

[@karlbesser1696]

Did you get the exact functionality of the device from the manufacturer, or are you speculating how the device might work.

[@Thesignalpath]

I am describing how it works based on my own knowledge, there is no information from the manufacturer.

[@karlbesser1696]

@@Thesignalpath Then I congratulate you on your comprehensive knowledge. ✔️

[@TheTransporter007]

Imagine if they used a PA with the requisite DSP. It would be 10x the cost, but probably 10x as effective.

[@FurkanBahadr]

I am surprised by these 77 GHz baluns there. This is the first time I am seeing something like that :D

[@FurkanBahadr]

Are there any papers about this topology? Can someone help me out?

[@ralfschooneveld3186]

From what car is this?

[@robertculver3296]

Thanks again for an interesting 'drive' around a microwave pcb👍

[@aliemlek]

👍👍👌👌

[@0MoTheG]

What is the range and angular resolution on these? They seem overly complicated for presence detectors. 2.4GHz CW radars could do as much for pennies.

[@Thesignalpath]

No, a 2.4GHz radar cannot do the same thing. Not only does it not have the available bandwidth, a similar antenna configuration would be much larger at that frequency. There is a reason why we don't use 2.4GHz to build automotive radar.

[@cameramaker]

What are those "via like" pads in the dies? Anyway you could mention that there is some digital part on each die, likely something for i2c setup or so, and it has also a bunch of fuses

[@petergibbons607]

and i used to think 27 MHz was crazy :D

[@rtronicslab1253]

whish my lab was that equipeted,o man,some day, some day.....enjoyed yor videos ,a alway....

[@0MoTheG]

Why is this a synthetic aperture when I can see the aperture? Because it kind of works as if the antenna would be moving sideways?

[@Thesignalpath]

See my pinned comment.

[@shockwave77598]

The thermal pad is thick because it a) conducts the heat good enough, and b) is rubbery to offer some vibration resistance. If it was glued to the case, every bump and every rock would vibrate the chip, and the legs would break within days.

[@Thesignalpath]

It doesn’t need to be glued. It also doesn’t need to be so thick to solve any vibration issue. It lowers the tolerances which I suspect is the reason because everything is thermally welded and has poor tolerance. No screws at all.

[@davidvb3754]

Your Enterprise painting is hidden beyond your gear..... 😭

[@sheffieldhoney]

Those shelves look strained :-)

[@douro20]

They do but they are also very strong.

[@Thesignalpath]

Yeah... A tilt has developed.

[@hempbear]

...and the cool Enterprise painting is being obscured!

[@FindLiberty]

My wallet has a built-in sphincter muscle that pre-empts that much expensive hi-tech weight.

[@davimarsteinarsson6753]

When he says “It is fairly simple design “ I feel stupid

[@dieboodskapper]

curious to know how these automotive radars and..... lidars gonna deal with a multitude of other transmitters from potentially millions of other cars.....creating potential phantom bogus receiver patterns ...pretty much like shining a flashlight into someones eyes.....what do you say?

[@FindLiberty]

Good points; millions!. Doesn't the FCC give them separate play pens? The error detection routines for like-systems must sort all that out when those short pulses happen to line up at exactly the same time once and a while.

[@muctop17]

Have a look into Spread Spectrum of GPS and the 2 Voyager vehicles 😁 There is explained how code separation works. With spread spectrum you can listen below noise level ! But I don't know exactly how it's working here

[@dieboodskapper]

It seems like the only viable solution to this problem is stereoscopic image processing ....

[@dieboodskapper]

I was wondering if you can possibly give a open source design ....to detect the dangerous 60Ghz (active denial systems) oxygen absorption region, kind of like a simple cheap geiger counter. Looking at these designs I wonder if it would be able to slightly shift the RX mixer LO to 60GHz center frequency and convert to a power meter. I sure would buy something like that just to scan the neighborhood. Anybody??

[@BartKus]

28:15 really, 20ms chirps? That seems really long for a short range radar.

[@m.arsyaelnathan9606]

Ea

[@taomicioli]

all the pads are prob for vibration reduction in addition to heatsinking, imo

[@igorrizvic6008]

50 db under how much voltage?

[@federicogemin1715]

How the hell do you know all those things

[@0MoTheG]

KZbin (And books)

[@DrTune]

22:22 "Nonetheless, the architecture of a very basic radar like this, is not so complicated" mmmmmmm....

[@camfocus8888]

RX antenna is actually on the right, Tx is on left sight. I don’t think BSD radar using SAR, it is just FMCW phase array radar to measure the angle and distance. BSD radar doesn’t need forming a SAR radar image and it requires much more processing power to do it.

[@Thesignalpath]

No, RX is on the left TX on the right. As for comparison between SAR, MIMO & SIMO, see my pinned comment. The part number for all the components are also indication what is what on this board. There is no debate here.

[@Thesignalpath]

> FMCW phase array radar to measure the angle and distance. Yes, FMCW as I mentioned in the video and is confirmed by the chirp generator. However, it is not a simple phased array. It is digitizing each RX antenna on the left separately and simultaneously, again, as I explained in the video. There is a considerable processing involved here because we have a digital beam former in the back-end - again confirmed by the analog front-end. We also have the part number of the components - there is no doubt what is what here - RX on the left, TX on the right. See my pinned comment.

[@TheM4man]

Would be super interesting to power that 77GHz board and probe the IF's 🥸♥️

[@ko-prometheus]

How to make an instrument or device to detect the presence of microwave radiation from 5 GHz and up to 50 GHz? Cheap to make.

[@tayyem-org]

the sound has very poor quality, it is better if you focus on slowing down your speed and use high quality mic

[@Thesignalpath]

I’ll adjust the equalizer balance, I think I have better settings. As for the speed, please watch it at 0.75x if you find it too fast.