Amazing video. Great production, clearly explained both from an intuitive perspective and a mathematical one. Best ever explanation of FMCW radar that I've ever seen. Keep the great work, looking forward to learning more on this topic. Working for years with ADI's AD9361 chips in radio direction finder applications, I'm now wondering if we may need to start developing a radar product in the future.
The audio processing of this video is quite harsh to listen to, there are a lot of high-frequency artifacts and background noise. Your two previous videos with the headphone microphone were clearer.
Yes, it is overdue! I am hoping to finish this series in the next 2 weeks. So the next video should post tomorrow, that'll be on synchronization. Then I've got the CFAR one already recorded, and I just need to edit and post. So I can do that soon. And then I just need to record the drone radar video. I had to wait to get my FCC part 107 license to be legally allowed to post drone footage to youtube. But I've got that now! So hopefully we'll wrap this up soon.
Great video as always! I was looking to get an adar1000 chip to just play around with, but it seems they are not really all that available. In fact, most of the high-end rf radar, dds, phase shifters, and pll stuff are not really available on mouser, digikey, etc. I mean, they are there, but behind order limits or price quote walls. Probably ITAR right? Is there like a less expensive adar1000 dev board? Anyway, I can't wait for the SAR stuff! This is so interesting. I appreciate all your expertise and the time you take to make these videos.
Yes the RF parts are a bit high end and with the PCBs and everything it would probably end up being more expensive than just the Phaser kit. But I know that's a lot for someone just exploring. I would recommend checking out the Pluto. You can follow my other videos to make that into a 2 channel beamformer--and you can do a lot with that. And you can also do the radar pieces we're talking about in this series--just at lower BW.
In simpler terms, when a FMCW radar is modulated with a sawtooth wave, it tends to suppress one of the sidebands because the radar signal only increases or decreases in frequency in one direction before abruptly returning to the starting point. This creates more energy in one direction of frequency change, making one sideband stronger than the other. The abrupt return to the starting frequency doesn't add much to the opposite sideband, so you end up with one dominant sideband and the other being much less pronounced. This effect is due to the way the sawtooth wave modulates the signal, focusing the energy in either frequencies above or below the carrier frequency, but not both equally.
Jon, thank you for videos about RADAR. I am happy to find them. I asked about connection in LinkedIn. Please respond. In Ukraine we really need this type of knowledge during the war.
I love these videos, I'm learning a lot. Got me thinking, though. Can data be encoded into a RADAR pulse? Could be a way to prevent jamming/spoofing, if every pulse is, say, a randomly generated token. The receiver is expecting the return to be a specific token that would be impossible for the EW equipment to know. Maybe they already do something like that? I love radio technology, there are so many possibilities.
Yes, you're in the right thinking! Instead of FMCW, we could have done some information -- like "Barker" codes. They have a good correlation for radar pulse compression.
Anything that can modulated arbitrarily can be used to transfer information. Modulating “radar” to encode digital information is basically digital radio like Wi-Fi signals we all know and love. There is nothing stopping you from using traditional radar techniques on digitally modulated signals like this, so you could definitely use a digitally modulated radar signal to both track a target and send or receive data. This is the basis of transponders. What is called “secondary radar” in air traffic control uses this technique to ping and interrogates digital transponders (IFF) of civil and military aircraft. Its signal is encoded to request a reply signal (a code) but at the same time it is used to detect direction and distance of the transponder reply. “Primary radar” is just traditional pulse radar at 2700-2900 MHz, which can only detect, but not identify, targets. That’s why these two technologies are used together in both civil and military applications to both track and positively identify tracked targets.
Fantastic, as a mechanical engineer who has an interest in electronics and RF im loving this series! But in the demonstration, why is the waterfall and frequency plot messy? as in whats with all the other noise? Is it an artefact of the mixer or of the digital waveform generator? i know square waves have messy harmonics due to the Fourier expansion ect but would have thought a saw tooth would be a bit cleaner. Also you said the upper and lower waveforms are mirrored around 100khz, but are they actually mirrored around the 100khz + or - the beat frequency due to the mixer?
Thanks! That freq messiness is all the radar return clutter -- so all the items in my small office are reflecting that transmit signal, and they all show up on the FFT plot. So each of those spikes is from a book, or a chair, or a lamp, etc. In the next video, we'll work on a way to suppress that clutter so that only the "real" targets show up -- that the CFAR algorithm.
This series is helping me understand the FMCW app. I'm lashing up something similar to the MIT example and am struggling with the digital ramp/chirp generator. You said that the round-trip time, at least for your close-range examples, is some tens of nanoseconds, but the step time on your ramp is 1 microsecond (07:23 ramp_time/num_steps) so the frequency would be likely constant over that time period. Is there a strong filter on the DAC output, or am I missing something?
I'll answer my own question here. After looking a few lower-level FMCW radar pages, I see that if you don't use an actually function generator (like the XR2206 used on the MIT project) you use a low pass filter, commonly based on an op-amp. I'm using plain old R & C and feeding that to an op-amp to get the output up to 20 V for my VCO. My DAC has a 5 us settling time so it really needs some smoothing to get an actual ramp over a few ns. If you take a microsecond to change frequencies, it's not really modulated over the out and back timeframe.
Hey Jon, can you explain why there is a 4 divisor for my_phaser.frequency and .freq_dev_range and .dev_step in all the python examples? It obviously works as I’ve got all the examples running on my Phaser. I just can’t understand why it's needed.
Yes, that's a good question. It's because the VCO we use, HMC735, has a divide by 4 in its feedback output. So what the ADF4159 "sees" is the VCO freq divided by 4. That's a good question though. Actually, its better to post questions to the link below, that way everyone can search it and see the answer. But thanks, and hope the Phaser is working well for you! ez.analog.com/adieducation/university-program/
I know it doesnt have anything to do with this, but what do I need to make a drone GPS system that transmits coordinates in real time to a ground based reciever so i can display them over a map? Any help would be greatly apreciated.
Yes, great point! Actually the video tomorrow is on syncing those sweeps to smooth out the FFT and also remove the non-linear part of the ramp. Let me know what you think!
I find your audio fidelity lacking with many instances of audio fall off, and missing words. Even on replay it's difficult. and close caption does not capture. There is also severe mid to high range clipping, from some sort of compression you're employing. this topic is too important to have this problem.
I really do appreciate the feedback! I've been trying to get better with new mics and software. Often it will sound good on my computer, but then on my TV it doesn't sound great. I'd like to figure it out and improve it though. Are you able to give me some specific time codes where you notice an issue? Then I can find a set of speakers where I can hear that same issue at those time codes and see if I can clean it up.
@@jonkraft ..@5:23 when you employ the 's' sound and the 'ch' sound it really ramps up...but it's throughout. Looks like you have a lav mic on. ? maybe look at the input settings and look for settings that are good for full spectrum music, or such.
@@TheoSmith249 Yeah, I see what you mean now.... The youtube subtitles seem good, if that helps. But I need to fix this audio. I think I'm done with that lav mic, I'm going back to my trusty old 1990's boom arm headset. But I did remix the sound, and applied a "de-ess" filter. If I could impose on you again, does the video posted at the link below sound any better? Or worse? Anyone else can also evaluate -- just give thumbs up if you like the audio better, or thumbs down if the audio still sounds bad. Thanks for the feedback! th-cam.com/video/jM7Kh-TetuM/w-d-xo.html
@@jonkraft the sound is flatter, with less high end. the lisps of the S and CH are still there but under some mild suppression filter. The original is brighter and easier to listen to. Ham operator here, so audio is a sensitive subject. 😏
Hi Jon, I'm trying to get an understanding of the basic principles of radar, and your series has helped me. I have a general question regarding radar and cars. If your radar was setup in car moving, and a tractor trailer got in front and blocked both transmitting and receiving antennas. Would the radar be able to pick up your rig moving back and forth?
I am so glad I found your channel! This is awesome! I can't wait to build my own! I'd like to adapt it to my home security system just because having the pizza delivery guy pulling up on my radar while I'm in my lab would be SO FREAKING COOL! Thank you! As my TH-cam handle suggests I'm a candidate for Massachusetts 1st Congressional District and I would LOVE to find federal funding to get these types of projects into our schools and really give kids the opportunity to sample a plethora of technical projects along these lines. I would have loved to have a class on software defined radio as an option in high school!
Nice video series. Could you provide an approximate weight and dimensions of the device? I've been looking around the analog website and can not seem to find this information.
Hi Jon! Great videos!! I ordered a Phaser kit at the end of last year, but it hasn't arrived yet. The supplier (DigiKey) keeps pushing the delivery date forward every time we ask for an estimate-it's always 'a week' further in the future. Do you have any idea of an actual delivery date, or suggestions on how I could obtain a unit? 🙂
@@jonkraftHi again Jon! The kit has now arrived! Thank you very much for your information! A question: Do you know if it's possible to connect a standard laptop USB-C charger to the kit (a charger that can output multiple voltages between 5--20V)? Does the kit communicate with the power adapter to ask for 5V, or do I have to use the included Raspberry Pi adapter?
@@KulMedVideo Glad you finally received that kit, because I think ADI is almost sold out of them again.... For any technical support on Phaser, just post your question here: ez.analog.com/adieducation/university-program/
@@jonkraftThank you Jon for the link! I hadn't seen that one before! Currently I'm setting up the Kupier system, but I know you provided a pre-made SD-image for the Phaser-kit that I cannot find right now. Maybe you mentioned the link to the image in one of your YT-videos, I'll keep searching for it!
@@KulMedVideo You can find the SD card download and instructions here: wiki.analog.com/resources/eval/user-guides/circuits-from-the-lab/cn0566/quickstart
It is a great demonstration. Thank you for this. I have a question about the range of the compressed signal. Why do we not see a clear sinc function for the target? There are very high sidelobes. Multi-paths may cause this, but they seem pretty constant and moving with the target. Do you think it is a hardware issue?
I have a problems with flying drones around apartments. I have no idea if somebody is spying or just flying for fun but it makes me sometimes nervous. Is possible to do kind drone detector if fly close to your window? ❤
Yes, that's a good question. I went through it briefly in the CW Radar video, but it was probably easy to miss. The 100 kHz is the baseband "information" that we are sending. So Pluto outputs 2.2 GHz (that's its LO) and then adds 100kHz sine wave to that (so 2.2G + 100kHz which is 2.2001 GHz). Then mixes that up to 10GHz + 100kHz, and then ramps that up to 10.5GHz + 100kHz. Then the receiver gets that chirp back and reverses those freq shifts -- eventually mixing back down to baseband (0Hz), leaving just the original 100kHz +/- the beat freq. I don't know if this explanation is any better, but hopefully it makes some sense!
@@jonkraft Ah yes that makes sense, thanks. Because I guess if you didn't add that 100kHz then the beat frequency might be near zero and that could be difficult to process
Hey Jon, At 9:06 you said "This is about one meter..." That rail looks like almost the double. A foot is like 30 centimeters. One meter is 100 centimeters. You look like someone who understands what he's speaking about, so please don't say that rail is about one meter long, because it can be seen that it is much longer.
Amazing video. Great production, clearly explained both from an intuitive perspective and a mathematical one. Best ever explanation of FMCW radar that I've ever seen. Keep the great work, looking forward to learning more on this topic. Working for years with ADI's AD9361 chips in radio direction finder applications, I'm now wondering if we may need to start developing a radar product in the future.
Thanks!
Wow you make these concepts super easy to understand! I can't wait for the next episode!
Great work. You are really great in explaining complex subjects. Waiting for every update
The audio processing of this video is quite harsh to listen to, there are a lot of high-frequency artifacts and background noise. Your two previous videos with the headphone microphone were clearer.
Awesome demo! Thank you. Can't wait for the next part.
This is a fantastic demonstration, thank you!
patiently waiting for the next one!
Thank you for the video. I love it!
When do you think you'll do the 4th video? I really like how you break down these concepts in radar.
Yes, it is overdue! I am hoping to finish this series in the next 2 weeks. So the next video should post tomorrow, that'll be on synchronization. Then I've got the CFAR one already recorded, and I just need to edit and post. So I can do that soon. And then I just need to record the drone radar video. I had to wait to get my FCC part 107 license to be legally allowed to post drone footage to youtube. But I've got that now! So hopefully we'll wrap this up soon.
This is an amazing series!
Gread video, It's really help me about to know Knowledge of Radar. From Asia
Great video as always! I was looking to get an adar1000 chip to just play around with, but it seems they are not really all that available. In fact, most of the high-end rf radar, dds, phase shifters, and pll stuff are not really available on mouser, digikey, etc. I mean, they are there, but behind order limits or price quote walls. Probably ITAR right? Is there like a less expensive adar1000 dev board? Anyway, I can't wait for the SAR stuff! This is so interesting. I appreciate all your expertise and the time you take to make these videos.
Yes the RF parts are a bit high end and with the PCBs and everything it would probably end up being more expensive than just the Phaser kit. But I know that's a lot for someone just exploring. I would recommend checking out the Pluto. You can follow my other videos to make that into a 2 channel beamformer--and you can do a lot with that. And you can also do the radar pieces we're talking about in this series--just at lower BW.
In simpler terms, when a FMCW radar is modulated with a sawtooth wave, it tends to suppress one of the sidebands because the radar signal only increases or decreases in frequency in one direction before abruptly returning to the starting point. This creates more energy in one direction of frequency change, making one sideband stronger than the other. The abrupt return to the starting frequency doesn't add much to the opposite sideband, so you end up with one dominant sideband and the other being much less pronounced. This effect is due to the way the sawtooth wave modulates the signal, focusing the energy in either frequencies above or below the carrier frequency, but not both equally.
Jon, thank you for videos about RADAR. I am happy to find them. I asked about connection in LinkedIn. Please respond. In Ukraine we really need this type of knowledge during the war.
I love these videos, I'm learning a lot. Got me thinking, though. Can data be encoded into a RADAR pulse? Could be a way to prevent jamming/spoofing, if every pulse is, say, a randomly generated token. The receiver is expecting the return to be a specific token that would be impossible for the EW equipment to know. Maybe they already do something like that? I love radio technology, there are so many possibilities.
Yes, you're in the right thinking! Instead of FMCW, we could have done some information -- like "Barker" codes. They have a good correlation for radar pulse compression.
Anything that can modulated arbitrarily can be used to transfer information. Modulating “radar” to encode digital information is basically digital radio like Wi-Fi signals we all know and love. There is nothing stopping you from using traditional radar techniques on digitally modulated signals like this, so you could definitely use a digitally modulated radar signal to both track a target and send or receive data.
This is the basis of transponders. What is called “secondary radar” in air traffic control uses this technique to ping and interrogates digital transponders (IFF) of civil and military aircraft. Its signal is encoded to request a reply signal (a code) but at the same time it is used to detect direction and distance of the transponder reply.
“Primary radar” is just traditional pulse radar at 2700-2900 MHz, which can only detect, but not identify, targets. That’s why these two technologies are used together in both civil and military applications to both track and positively identify tracked targets.
Fantastic, as a mechanical engineer who has an interest in electronics and RF im loving this series! But in the demonstration, why is the waterfall and frequency plot messy? as in whats with all the other noise? Is it an artefact of the mixer or of the digital waveform generator? i know square waves have messy harmonics due to the Fourier expansion ect but would have thought a saw tooth would be a bit cleaner. Also you said the upper and lower waveforms are mirrored around 100khz, but are they actually mirrored around the 100khz + or - the beat frequency due to the mixer?
Thanks! That freq messiness is all the radar return clutter -- so all the items in my small office are reflecting that transmit signal, and they all show up on the FFT plot. So each of those spikes is from a book, or a chair, or a lamp, etc. In the next video, we'll work on a way to suppress that clutter so that only the "real" targets show up -- that the CFAR algorithm.
Sounds good, i cant wait!
This series is helping me understand the FMCW app. I'm lashing up something similar to the MIT example and am struggling with the digital ramp/chirp generator. You said that the round-trip time, at least for your close-range examples, is some tens of nanoseconds, but the step time on your ramp is 1 microsecond (07:23 ramp_time/num_steps) so the frequency would be likely constant over that time period.
Is there a strong filter on the DAC output, or am I missing something?
I'll answer my own question here. After looking a few lower-level FMCW radar pages, I see that if you don't use an actually function generator (like the XR2206 used on the MIT project) you use a low pass filter, commonly based on an op-amp. I'm using plain old R & C and feeding that to an op-amp to get the output up to 20 V for my VCO. My DAC has a 5 us settling time so it really needs some smoothing to get an actual ramp over a few ns.
If you take a microsecond to change frequencies, it's not really modulated over the out and back timeframe.
Amazing!
Hey Jon, can you explain why there is a 4 divisor for my_phaser.frequency and .freq_dev_range and .dev_step in all the python examples? It obviously works as I’ve got all the examples running on my Phaser. I just can’t understand why it's needed.
Yes, that's a good question. It's because the VCO we use, HMC735, has a divide by 4 in its feedback output. So what the ADF4159 "sees" is the VCO freq divided by 4. That's a good question though. Actually, its better to post questions to the link below, that way everyone can search it and see the answer. But thanks, and hope the Phaser is working well for you!
ez.analog.com/adieducation/university-program/
Can you please do the project about ground penetrating radar (GPR) using the Phaser? or Through Wall Sensing / wall thickness measurement
how do you get the beat frequency?
I know it doesnt have anything to do with this, but what do I need to make a drone GPS system that transmits coordinates in real time to a ground based reciever so i can display them over a map? Any help would be greatly apreciated.
You are not aligning the sweeps. That is why your spectrum is sampled.. You can make the graph much smoother and also measure doppler
Yes, great point! Actually the video tomorrow is on syncing those sweeps to smooth out the FFT and also remove the non-linear part of the ramp. Let me know what you think!
I find your audio fidelity lacking with many instances of audio fall off, and missing words. Even on replay it's difficult. and close caption does not capture. There is also severe mid to high range clipping, from some sort of compression you're employing. this topic is too important to have this problem.
I really do appreciate the feedback! I've been trying to get better with new mics and software. Often it will sound good on my computer, but then on my TV it doesn't sound great. I'd like to figure it out and improve it though. Are you able to give me some specific time codes where you notice an issue? Then I can find a set of speakers where I can hear that same issue at those time codes and see if I can clean it up.
@@jonkraft ..@5:23 when you employ the 's' sound and the 'ch' sound it really ramps up...but it's throughout. Looks like you have a lav mic on. ? maybe look at the input settings and look for settings that are good for full spectrum music, or such.
@@TheoSmith249 Yeah, I see what you mean now.... The youtube subtitles seem good, if that helps. But I need to fix this audio. I think I'm done with that lav mic, I'm going back to my trusty old 1990's boom arm headset. But I did remix the sound, and applied a "de-ess" filter. If I could impose on you again, does the video posted at the link below sound any better? Or worse? Anyone else can also evaluate -- just give thumbs up if you like the audio better, or thumbs down if the audio still sounds bad. Thanks for the feedback!
th-cam.com/video/jM7Kh-TetuM/w-d-xo.html
@@jonkraft the sound is flatter, with less high end. the lisps of the S and CH are still there but under some mild suppression filter. The original is brighter and easier to listen to. Ham operator here, so audio is a sensitive subject. 😏
Thx for sharing your impeccable ideas and unventions
Hi Jon, I'm trying to get an understanding of the basic principles of radar, and your series has helped me. I have a general question regarding radar and cars. If your radar was setup in car moving, and a tractor trailer got in front and blocked both transmitting and receiving antennas. Would the radar be able to pick up your rig moving back and forth?
Jon -- really cool, I didn't know ADI had Python Libraries for their RF ICs :-)
I am so glad I found your channel! This is awesome! I can't wait to build my own! I'd like to adapt it to my home security system just because having the pizza delivery guy pulling up on my radar while I'm in my lab would be SO FREAKING COOL! Thank you! As my TH-cam handle suggests I'm a candidate for Massachusetts 1st Congressional District and I would LOVE to find federal funding to get these types of projects into our schools and really give kids the opportunity to sample a plethora of technical projects along these lines. I would have loved to have a class on software defined radio as an option in high school!
Very cool project, really good 😊😊
Ah c'mon, can't i do a phased array using magnetrons? Too expensive for someone whose paychecks come in BRL and not in USD. 🤣
Nice video series. Could you provide an approximate weight and dimensions of the device? I've been looking around the analog website and can not seem to find this information.
Very nice and Professional knowledge, thanks!
Cool! 😁
This is really amazing! Thanks a lot!
Does anybody know what the RPI RAM size must be?
Hi Jon! Great videos!! I ordered a Phaser kit at the end of last year, but it hasn't arrived yet. The supplier (DigiKey) keeps pushing the delivery date forward every time we ask for an estimate-it's always 'a week' further in the future. Do you have any idea of an actual delivery date, or suggestions on how I could obtain a unit? 🙂
Yes, sorry about that. We have 65 kits headed to our warehouse now. So I hope by early March they will be shipping out.
@@jonkraftHi again Jon! The kit has now arrived! Thank you very much for your information! A question: Do you know if it's possible to connect a standard laptop USB-C charger to the kit (a charger that can output multiple voltages between 5--20V)? Does the kit communicate with the power adapter to ask for 5V, or do I have to use the included Raspberry Pi adapter?
@@KulMedVideo Glad you finally received that kit, because I think ADI is almost sold out of them again.... For any technical support on Phaser, just post your question here: ez.analog.com/adieducation/university-program/
@@jonkraftThank you Jon for the link! I hadn't seen that one before! Currently I'm setting up the Kupier system, but I know you provided a pre-made SD-image for the Phaser-kit that I cannot find right now. Maybe you mentioned the link to the image in one of your YT-videos, I'll keep searching for it!
@@KulMedVideo You can find the SD card download and instructions here: wiki.analog.com/resources/eval/user-guides/circuits-from-the-lab/cn0566/quickstart
Thank you so much.
Thanks, Jon 👍
:)
It is a great demonstration. Thank you for this. I have a question about the range of the compressed signal. Why do we not see a clear sinc function for the target? There are very high sidelobes. Multi-paths may cause this, but they seem pretty constant and moving with the target. Do you think it is a hardware issue?
What is the practical range to actually detect an object using this setup? Could it do 10km?
I have a problems with flying drones around apartments. I have no idea if somebody is spying or just flying for fun but it makes me sometimes nervous. Is possible to do kind drone detector if fly close to your window? ❤
What is the max distance this radar can track drones at?
Sorry if I missed it but where does the 100kHz centre frequency in the beat frequency come from?
Yes, that's a good question. I went through it briefly in the CW Radar video, but it was probably easy to miss. The 100 kHz is the baseband "information" that we are sending. So Pluto outputs 2.2 GHz (that's its LO) and then adds 100kHz sine wave to that (so 2.2G + 100kHz which is 2.2001 GHz). Then mixes that up to 10GHz + 100kHz, and then ramps that up to 10.5GHz + 100kHz. Then the receiver gets that chirp back and reverses those freq shifts -- eventually mixing back down to baseband (0Hz), leaving just the original 100kHz +/- the beat freq. I don't know if this explanation is any better, but hopefully it makes some sense!
@@jonkraft Ah yes that makes sense, thanks. Because I guess if you didn't add that 100kHz then the beat frequency might be near zero and that could be difficult to process
Hey Jon,
At 9:06 you said "This is about one meter..."
That rail looks like almost the double. A foot is like 30 centimeters. One meter is 100 centimeters. You look like someone who understands what he's speaking about, so please don't say that rail is about one meter long, because it can be seen that it is much longer.
I don't know what to tell you. From end stop to end stop it is about one meter long. You're free to not believe me though....