The fuses were extremely interesting. I wonder what's the impact of the extra load introduced by those self test paths once the fuses are blown, since they would turn into somewhat long shunt open stubs. Guess it should be fine when your output power has a range from 2 to 10 dBm :D
Wow. RF stuff is crazy. Damn, there are so many things to learn in this world. I guess I have enough projects for 5 years ahead, but there is always something cool in a completely different domain. Great video btw!
Good luck with conference and please cover some RF metrology there, if you plan to make conference coverage video. I'd expect it is also important to ensure all GHz are correct and traceable for successful designs. What are metal boxes fill around the empty areas? Just to help with uniform etching during chip lithography and production? Doesn't look like they trying to hide anything.
The fill is normally used to improve planarity during manufacturing. The chip is polished flat after each layer so that the surface is smooth, since the lithography system has a very shallow depth of field. But metal and glass have different hardnesses and polish at different rates, so to get good uniformity it's important to have about the same percentage of metal per unit area across the whole wafer. For copper damascene processes this is even more important because the metal is patterned by etching trenches into the glass, filling with copper, then polishing flat. The top layer is exempted from this in some cases (or has much looser requirements) because there's nothing on top of it except for glass cuts for bond pads (which have very large feature sizes).
awesome microscopy....Very versatile unit you have allowing for numerous perspectives in doing layer analysis....glad you have the will to accumulate & use all the various technologies you have in your lab......geek candy @ the highest level. It also is quite enlightening to see how the various devices are built @ microwave frequencies...golden as always
Excellent video, as always. I'm using this IC to make a 24 GHz FMCW Doppler radar. I'm going to extend range by adding a much larger antenna array, a power amplifier, and my own receiver amplifiers before and after the IC ports.
Excellent. Please remember to use the full name for the components and not just their initials. Us noobs need that help to be able to understand what you are teaching.
Can this motion sensor be set in a kind of “listen-only” mode, so as to detect radio sources emitted by other sensors, and see the signal with RTL-SDR or the like ?
I was recently looking at a patent for a tuned resonator that looked like a rectangular box with several compartments, but the text was talking about GHz range frequencies, implying that it was a very tiny box indeed. I was excited to notice this structure at 23:24. It's sitting just to the right of some inductors, and to the left of the amplifier that's just off the nose of the fuse-tuned structure. When I was looking at the patent, I didn't realize it was talking about microchip-sized structures, because it was off-subject to what I was searching for, so I didn't look too closely. That was pretty exciting to find an accidental discovery "in the wild", so to speak, but I am not so naive to believe that search engine miss didn't influence this video suggestion later.
25:47 how do they "solder" these wires on there? also, on the beginning of the wire it's easy, just point and stick, but how do they fix the other side of the same wire, how do the rotate it 180º and get the "beginning" again.
Great video. Love the high power shots. I’m blessed to get to play with a Keyence VHX and an older Nikon at my job inspecting RF/Microwave modules and it never gets old.
I worked with the P2G DK of Infineon for my master’s project (last year) but I think most of the DK for radar of Infineon are discontinued by now 😅 and they were not good with tech support from my experience during the development of my project
Those tiny imperfections you show near the end might give your device a very (almost infinitesimally) small signal that could be used to distinguish between two devices of the same type. Like if you took 10 of those modules, you could probably select 2 specific modules that you could always tell apart. I wonder if this type of signals analysis is feasible for keeping track of which source is sending... (not just with these modules, but in general)
@@chromosundrift That was the gist of the thought. Being able to use side channel signals to catalog and track locations passively in several types of situations, including radar. The example of discriminating between 2 that you cherry pick was to demonstrate that there _sometimes_ was a big difference which would be easy to detect, but often would be a smaller amount which requires more sensitivity in receiver.
Next month I get a small 1.5kW X-band radar for a sailboat to play with so I am gathering information. The antenna looks like that on the doppler. Is that stripline or a coplanar waveguide ? Can you make a video about normal radar like used on a ship
Did you decap the ic or did you have a bare die that you used? Also for videos like this it would be useful to have some kind of scale to get an idea of the size of the components on the chip. I am guessing they are not too small since you did use a microscope. Still I am not sure if what you were showing were tens or hundreds of micrometers in size.
I have decaped the IC. It was in a QFN package. The total die area is less 1mm by 1mm, including the pads. The highest magnification was over 200x. So these structures are pretty small. The lines are a few micron in width.
I love when things are explained well. Thank you for taking the time to share. Not everyone has access to such analytical techniques. It's one thing to just describe a circuit with block or circuit diagrams. It's completely another to pull things apart and show the logic of physics in construction. Thank you.
It might be interesting to show some microwave PCB design by essentially taking an HB100 doppler module as the base, and modifying the design to allow injecting an externally generated IF (in quadrature, if easier) by which the local microwave oscillator or the received signal will be offset, so the downmixing won't center speeds at 0Hz. Also, not having the downmix go to zero-IF quadrature as would seem an obvious alternative, would preserve the precious DC offset of a static reflector a fractional number of wavelengths in front of the antenna array, along with all the other benefits like escaping 1/f noise and far less linearity needed from the ADCs that feed the downmix centered on the IF into further digital processing. If it's the TX side that gets frequency offset, having multiple of these fed from the same microwave amplifier, but different offsets should allow frequency-division-multiplexing in the receivers when the IF's are spaced enough to account for possible doppler shifts but are otherwise close together. Separating them could be done in the digital processing, and with sufficient spacing of the antennas moving objects could get 3D movement computed, instead of the normal 1D limit of a single TX/RX pair. It's just that coming up with a design anywhere near as clever as the HB100 isn't straight-forward.
Loved the video, thanks Shahriar for another one. As to your question, yes always appreciate more depth and explanation. Wish I could attend the conference for fun.
When I see the block diagram I always assume the circuits to be super huge and complex. Very interesting to see how simple they really are. You could draw a circuit diagram for this on a single paper and understand most of it with normal RF knowledge
It looks like there's tiny little test points at each end of the fuse. Hard to tell without more mag to see if there's probe needle marks on them, but I suspect that there's a needle connected to each end of the fuse and current is passed between the two to blow the connection. This avoids an extra tooling step since it can be done on the same ATE system doing die/wafer test, vs having to measure the frequency offset and then move to a laser to actually blow the fuse.
I know that this is probably not the type of equipment that you usually play with, but can you do a review of the tinySA Ultra? I think a lot of people are quite interested in it at the moment, as it is very affordable (~150$).
I got one similar module for experiments with it. Thank you very much for your explanations about these 24 GHz microchip architecture. This helps me a lot.
Blockdiagram show oscillator with single-ended output and functional baluns to convert. But chip layout not. Also curious about the PN of the unlocked oscillator at 20Hz. Seems to be the short trip around time present enough time phase correlation so that PN do not hurt. PN spec at 100k seems only an device emission topic.
Wow I'm soo glad to see that video :o Thank you so much!!!! I know all that techniques but see it and explain it in that way is exceptional. Thank you soo much !!!
Hi, I found the module (infineon DEMO SENSE2GOL, right?) and it only reaches 20m. Is there a similar module, which can read speed up to 100m? I want to stick it in my car and measure if I have a constant distance to the car in front of me. A manual Radar cruize-control retrofit. Or is there a better solution I am missing? cheers :)
More beam forming should help; these antennas are pretty wide-angle. Edit: a car with a vertical back that you're illuminating in the normal direction w.r.t. that surface is also much brighter on radar than e.g. a human, so don't worry too much necessarily about that 20m figure.
I was wondering that. I can't believe that they would do micro probing in production volume, so presumably they were targeted by laser. There didn't seem to be any structure for blowing them within the chip itself, that probably would have been almost impossible.
Shahriar, I reiterate on doing a series on microwave phased array theory. This walk through had everything needed to generally understand the design achievements in silicon as well as the functional aspects.
Great educational videos as usual and very inspirational as well. Thanks It would be nice to see a couple of interviews with other people in the world of RFIC (maybe Razavi, Tomas Lee, ....)
The inset of the patch is for making a nice match right? It connects the trace to the point where the standing wave on the antenna works out to for example 50Ohm. I don't believe the directivity or bandwidth of the patch is really a function of this.
Indeed the return loss impact is significant, but BW of the path at >30GHz is also impacted once the dimensions of the antenna shrink and other imperfections come into play.
I would love a breakdown of phased array antenna design! I'm one of those people for whom RF borderlines on the Dark Arts. 🪄 I'm fairly certain that most of what I _do_ know was learned from your channel. Any knowledge you'd care to impart is definitely appreciated here!
Your RX side drawing may be a little wrong. The baun is the impedance match to the next circuit. On the RX side, it will switch for a small period for the input to the doppler shift. Then, the TX will be turned on for the in this case the anolog frequency to be transmitted. So the arrow was around the wrong way. But I still enjoyed the show.
Every balun in this circuit is responsible for either SE-D or D-SE conversion. Of course the appropriate impedance matching is needed. I don't see anything wrong with the block diagram. Furthermore, both TX and RX are on at the same time, otherwise, you don't have a radar!
