Hobby radar design

[sckzor]

Hello forum,

I am a software guy, recently I have become interested in writing software that relies on data that can only be supplied by a radar system.  The issue is that every off the shelf radar I have seen has been very expensive and not really what I am looking for.  I want a dead simple analog radar that I can aggregate data from without much work.  When looking for something like this I found a project made by MIT (here https://ocw.mit.edu/courses/res-ll-003-build-a-small-radar-system-capable-of-sensing-range-doppler-and-synthetic-aperture-radar-imaging-january-iap-2011/pages/projects/) that is basically exactly what I was looking for.  The issue with theirs is that it is rather large, clunky and still more expensive than I would like.   For this reason I took the schematics made by MIT and created what (I hope) is a schematic for a similarly spec-ed one made from cheaper surface mount electronics. I have done some simple projects with analog electronics and a little bit of RF stuff but I have never made a project as complex and RF intensive as this before.  Before I dump a large chunk of change on the parts for this board though I would like to get some input on whether or not it will work as I intend.  I have placed a link to a picture of the schematic for the project as well as the five different uncommon ICs that I put in it.   All other parts should be generic or common parts like 555 timers and op-amps.

If anyone has time to check this I would really appreciate a fresh, more experienced set of eyes looking at it.

Many thanks,

sckzor

Link to schematic: https://files.sckz.org/uploads/1660268499506579982.png


Raltron RQRA-2328-2536-CR VCO: https://www.raltron.com/webproducts/specs/VCO/RQRA-2328-2536-CR.pdf
KYOCERA AVX AT 0603 Attenuator (3db): https://datasheets.kyocera-avx.com/AVX-AT-Series.pdf
NXP BGU8052: https://www.nxp.com/docs/en/data-sheet/BGU8052.pdf
Analog Devices HMC213BMS8E: https://www.analog.com/media/en/technical-documentation/data-sheets/hmc213b.pdf
AEROSEMI 2A Step Up Converter: https://www.olimex.com/Products/Breadboarding/BB-PWR-3608/resources/MT3608.pdf

[radar_macgyver]

• You have the RF and IF ports on U5 flipped.
• I'm assuming AE2 is the receive antenna, in which case U4 is backwards (you have its output connected to the antenna).
• Connect all the NC pads of the VCO to ground.
• Get some good antennas to go with this. L-com is one place to look for 2.4 GHz wifi gear.
• The LM324s will run on +5V, so ditch the SMPS. You could also consider a proper R-R opamp (LM324 only goes down to ground).

Otherwise, it looks like a decent design, looking forward to seeing your results

[Berni]

Do watch out how you do PCB layout for the RF parts as parasitic effects and transmission lines are very important for RF. It can make the difference between this turning into an unstable noise machine or a well performing radar.

This is likely the reason that the MIT students went with ready made RF blocks. You don't need to know anythyng about RF design to just plug them together and have them perform to what the datasheet claims (they are just SMD chips in a metal box with connectors anyway). But yes these RF modules can be pretty pricey. Most of the time they are used for prototyping things or in niche small volume RF equipment.

If you just want a radar to play with these are some ready made radar modules that implement just the RF stuff on a board (LO, Mixers, Amplifiers, Directional patch antenna..etc) and operate in the frequency bands designated for radar like 10Ghz and 24Ghz. What you get out of them is the ready to go IF out of the mixer.
Example of one: https://www.aliexpress.com/item/4001178786384.html?spm=a2g0o.productlist.0.0.7f052399TPu4Cy&algo_pvid=25a6a566-3d13-43e7-8a85-414307386cd7&algo_exp_id=25a6a566-3d13-43e7-8a85-414307386cd7-8&pdp_ext_f=%7B%22sku_id%22%3A%2210000015056303221%22%7D&pdp_npi=2%40dis%21USD%2134.16%2131.77%21%21%2111.62%21%21%40210318cf16602820166238419eaf5c%2110000015056303221%21sea&curPageLogUid=FRkCv64z6iAM

[tggzzz]

Make sure you are legally allowed to transmit the radio signals, and that you won't "inconvenience" other users in the band or in other bands.

Authorities tend to become ratty and punitive if they have to discover why people are complaining.

[geggi1]

I have seen presentations where cheep SDR hardware are used as radar on 2,4Ghz.
Some radardesigns are passive radar type based on WIFI ewuipment.

[sckzor]

Thank you to all that responded to help.

I will address everyone's points:

First, radar_macgyver, I will use a list like you.
1) Upon further inspection I do indeed have the RF and IF ports reversed on the mixer.  I don't really know what I was thinking there, good catch.
2) You are also correct about the RF amp being backward too, I think I just copied and pasted it from the other RF amp and forgot to reverse it.  I would have never noticed that.
3) I was planning on connecting all the NC pins to GND, I just did not want to clutter the schematic or make it confusing.  At one point I had a note in there but I must have removed it or something.
4) The L-com antennas look very nice, I will seriously consider them for the project.  Thanks for the advice.
5) For the op-amps I copied MITs project exactly, they were using a discontinued part that probably only used 12v and I just dropped in the LM324, I didn't really pay attention and I will change it.  That said I don't think I really need a R to R op amp and getting the -5v I need for the R to R would just add unneeded complexity.  If there is any advantage to the R to R please let me know.

I will keep you posted on progress in this thread.

Secondly Berni,

I know that these RF parts are really particular about their board layout.  I did a little bit of reading about this before posing and I think the general consensus is that without all of the fancy lab grade equipment the best strategy is to use a 4 layer PCB, keep your RF stuff well spaced on the first of the four layers and separate it from the bottom layer using a ground plane and a voltage plane.  Finally use the final plane for excess signals that wouldn't fit on the top and hope for the best.  If you have any better suggestions then please let me know.  To address your second point, I did look around a bit for some pre-made radar but almost all I found looked were either doppler radar or had conflicting information.  I want to do SAR with this so I don't think that these systems would work.  Even if they do work for SAR I have learned a lot a lot about analog electronics and RF already from just reading/listening to the technology surrounding this topic and now I am interested to see if I can do it my self, even though it will probably not perform as well as a store bought unit.

Third tggzzz,

The 2.4 GHz band I am using is within the US unlicensed spectrum, the most the parts could even possibly over shoot even if I did not tune it to be within that spectrum is like 100 MHz and with such low power I am operating at and the fact that I will only be turning it on intermittently I think using a microwave oven would be more destructive to things around me than this project.  I don't intend on making this a commercial product of anything so even if it is technically illegal I don't think anyone will even notice, much less mind.

Finally geggi1,

I had never heard of using SDR hardware for radar before.  I looked into it after your post and it looks really interesting.  It looks like it is pretty undeveloped inaccurate though, I want something that has at least some chance of working, where I am not breaking all new ground.  Even if it is accurate though, like I wrote in my address to Berni, I am having fun with this and I would like to continue.

I want to thank everyone again for helping out here.  I won't be able to revise the schematic until Sunday as I will be out of town until then and do not have access to my main computer.  When I revise it I will send it here for anyone who is interested.

- sckzor

[radar_macgyver]

Using a rail to rail opamp does not imply a negative supply - it just means that the output stage of the opamp is optimized to allow operation all the way to the supply rails. The LM324 can operate down to the negative supply, but not the positive (the data sheet says it can go to 1.2V below the positive rail). This limits the output voltage swing if you're powering it from +5V.  A true RRIO opamp will give you a bit more dynamic range, that's all. An LM324 will work fine too.

To Berni's point, you could also consider adding shielding cans to the board. Leadertech and Laird make some decent ones. A bandpass filter on your TX path (eg: Mini-circuits BFCG-252+) will prevent radiating harmonics.

I got myself an ADI "Pluto" SDR kit, but found that there's no reliable way to time-sync the transmit and receive sections - but that was after only a couple of days of tinkering with GnuRadio - maybe I missed something.


One last thing: make sure either that there's nobody around while you're testing outdoors, or that people are aware of what you're doing. Folks with strange antennas and "OMG, he's got wires!!!" get unwanted attention from law enforcement. Ask me how I know...

[tggzzz]

Thank you to all that responded to help.

I will address everyone's points:

Third tggzzz,

The 2.4 GHz band I am using is within the US unlicensed spectrum, the most the parts could even possibly over shoot even if I did not tune it to be within that spectrum is like 100 MHz and with such low power I am operating at and the fact that I will only be turning it on intermittently I think using a microwave oven would be more destructive to things around me than this project.  I don't intend on making this a commercial product of anything so even if it is technically illegal I don't think anyone will even notice, much less mind.

(Deinterleaving your responses is a pain)

What makes you think you are transmitting in only the ISM bands? Hardware non ideal behaviour means your transmissions can be splattered outside the ISM band. If you don't understand RF specifications (e.g. IP3), then please pause for thought.

Whether or not it is a product is irrelevant. Disrupting another service it the point.

How often the transmitter is on merely affects the  hamce of interference, detection, prosecution.

[Berni]

Secondly Berni,

I know that these RF parts are really particular about their board layout.  I did a little bit of reading about this before posing and I think the general consensus is that without all of the fancy lab grade equipment the best strategy is to use a 4 layer PCB, keep your RF stuff well spaced on the first of the four layers and separate it from the bottom layer using a ground plane and a voltage plane.  Finally use the final plane for excess signals that wouldn't fit on the top and hope for the best.  If you have any better suggestions then please let me know.  To address your second point, I did look around a bit for some pre-made radar but almost all I found looked were either doppler radar or had conflicting information.  I want to do SAR with this so I don't think that these systems would work.  Even if they do work for SAR I have learned a lot a lot about analog electronics and RF already from just reading/listening to the technology surrounding this topic and now I am interested to see if I can do it my self, even though it will probably not perform as well as a store bought unit.

Yep that is a good PCB strategy for RF.

As for the radar modules, most of them are FMCW and this is mostly what you have built. CW is for Continuous Wave (as opposed to radars that emit pulses) and FM is for Frequency Modulation. This means that the LO is actually a VCO that can be controlled somehow. If you don't use FM and just use a static LO then you get a Doppler radar that will only see moving targets. If you feed a ramp waveform into the VCO then you get frequencies out of the mixer that correspond to the distance to the target (due to phase shift of distant targets moving faster at a given frequency sweep rate). Doing SAR radar then just means taking these FMCW snapshots and running them trough the math.

Tho if you are looking to get long ranges out of your radar it might be best to DIY, since you can give it more transmit power and use lower frequencies that can travel farther. The modules tend to use high frequency because it makes them more accurate in the shorter distances they are designed to be used, while building your own radar that works that high into the GHz starts to get into quite some RF magic.

[tggzzz]

This blog is interesting and might be relevant: https://hforsten.com/ There are blessedly few entries; clearly he is someone to be respected for not succombing to the "look at me" and "I opened a box and it works as expected" syndromes

MIMO radar antenna arrays
Short introduction to multiple input multiple output (MIMO) radar antenna arrays

Backprojection Backpropagation
Even better focused homemade SAR images with timedomain backprojection algorithm and automatic differentiation based autofocus.

Synthetic-aperture radar imaging
Differentiable synthetic-aperture radar image formation with Tensorflow. Including very fast image formation and autofocusing utilizing GPU.

Third version of homemade 6 GHz FMCW radar
New and improved version of the 6 GHz FMCW radar with two receiver channels.

TRL measurements with homemade VNA and open source software
Measuring transmission line parameters of 50 ohm microstrip on OSH park 4 layer process and S-parameters of 1 nF SMD capacitor.

Improved homemade VNA
Second version of the homemade 30 MHz - 6 GHz VNA with improved performance

RF power detector and scalar network analyzer
Next on the list of homemade RF test equipment is RF power meter. I decided to make one to calibrate the output power of my VNA and to also measure output power of the previous radar projects.

Cheap homemade 30 MHz - 6 GHz vector network analyzer
Vector network analyzers are used to measure high frequency circuits, unfortunately they are too expensive for student budget so I decided to make one myself for small budget.

Heartbeat detection with radar
Phase of the radar baseband signal is very sensitive to movement allowing to detect small movements caused by breathing and heartbeat.

Homemade synthetic aperture radar
Second version of the 6 GHz FMCW radar now with SAR images

Horn antenna for radar
Designing and fabricating better antennas for the radar

6 GHz frequency modulated radar
Making a cheap and simple radar with a range of about 100 meters.

[sckzor]

radar_macgyver:

Using a rail to rail opamp does not imply a negative supply - it just means that the output stage of the opamp is optimized to allow operation all the way to the supply rails. The LM324 can operate down to the negative supply, but not the positive (the data sheet says it can go to 1.2V below the positive rail). This limits the output voltage swing if you're powering it from +5V.  A true RRIO opamp will give you a bit more dynamic range, that's all. An LM324 will work fine too.

I only recall seeing rail to rail opamps being used with negative supplies.  I suppose that is where my misconceptions about them arose, I wrongly assumed that there were not advantages beyond the ability to go negative.  Thanks for clearing that up, I am still a little fuzzy on everything surrounding op-amps and I am excited to read more about them next time I get a chance to.  With regard to the circuit, knowing this, I don't see why I wouldn't go with a R to R opamp to get better dynamic range even if it is a few cents more money.  Something like the LMC6484 seems like a good, popular choice.  Thanks!


Berni:

Yep that is a good PCB strategy for RF.

I am happy to hear this should work! I would hate to have my circuit mysteriously not work for some reason I can't detect.

As for the radar modules, most of them are FMCW and this is mostly what you have built. CW is for Continuous Wave (as opposed to radars that emit pulses) and FM is for Frequency Modulation. This means that the LO is actually a VCO that can be controlled somehow. If you don't use FM and just use a static LO then you get a Doppler radar that will only see moving targets. If you feed a ramp waveform into the VCO then you get frequencies out of the mixer that correspond to the distance to the target (due to phase shift of distant targets moving faster at a given frequency sweep rate). Doing SAR radar then just means taking these FMCW snapshots and running them trough the math.

This is very interesting, I would love to read more about these topics, thank you for letting me know about this, as soon as I get a chance I will look more into this.

Tho if you are looking to get long ranges out of your radar it might be best to DIY, since you can give it more transmit power and use lower frequencies that can travel farther. The modules tend to use high frequency because it makes them more accurate in the shorter distances they are designed to be used, while building your own radar that works that high into the GHz starts to get into quite some RF magic.

This makes sense, I would like a long-ish range device, at hundred feet of somewhat accurate range is my hope, since 2.4 WiFi is rated at about 300 outdoor this seems pretty reasonable to me.  Thanks for your help!


Tggzzz:

This blog is interesting and might be relevant: https://hforsten.com/ There are blessedly few entries; clearly he is someone to be respected for not succombing to the "look at me" and "I opened a box and it works as expected" syndromes

That blog is indeed very interesting from what I saw of it.  I am a little short on time right now so I did not have a chance to read all of the entries you mentioned but I look forward to doing so in the future.

What makes you think you are transmitting in only the ISM bands? Hardware non ideal behaviour means your transmissions can be splattered outside the ISM band. If you don't understand RF specifications (e.g. IP3), then please pause for thought.

 
I will read more into this, I do have somewhat limited knowledge of all RF specifications, as I continue designing I will take this into account.  Your concern is appreciated.

[sckzor]

Thank you very much for taking all the time to explain that evb149,

I will beg to differ here, and I will tell you how / why in case it is helpful.  Even since probably the 1940s when "real" professional radar systems are designed before building prototypes I think they analyzed / simulated some performance parameters of the system based on mathematical models to get ideas for achievable range, noise, power, performance, etc.
These days especially it is very feasible to do any level of precision of computational electromagnetic modeling as to what a radar could / would do given a particular model of a particular environment without ever building a radar prototype and collecting real data.

We all see the videos / films of realistic but never real scenes like dinosaurs running around and aerial views of cities that never actually existed and photo realistic renders of 3d modeled environment scenes that are purely CGI but are not artistically drawn but are actually precisely 3d modeled in geometry and material characteristics and then ray-taced with simulated illumination and reflections etc. etc. to generate an image of what a virtual camera would see in scattering, reflection, dispersion, attenuation, etc. by every single light ray scattering, diffraction, reflection, refraction in the scene according to geometrical optics.

Did you see the Nvidia presentation by their founder when they launched the RTX3000 series GPUs?  Or other examples of 3d-computational electromagnetic modeling or ray tracing etc.?
https://developer.nvidia.com/rtx/ray-tracing
https://en.wikipedia.org/wiki/Computational_electromagnetics

I have a decent amount of experience working with hardware accelerated 3d graphics.  I have worked with Vulkan and OpenGL somewhat extensively and while I am certainly no expert, I understand how it works.  I watched the presentations about the 3000 series GPUs and even picked the 3060 Ti up a few days after launch in order to experiment with all of its features, so I have worked with RTX before.  Since visible light is at the end of the day just EMF radiation I don't see why modeling 2.4 GHz radiation with RTX on a GPU would be all that much different than modeling light, as a matter of fact, because the antenna originates from (basically) a single point I would think that doing this would actually be easier.  That said, I have not looked into using GPU hardware for modeling EMFs.

If you look for open source CEM and GPGPU stuff I'd be surprised if there weren't CUDA or OpenCL or similar analytic codes for SAR the real scientists use for real systems or at least useful academic models / research project open data / source.

Your GPU is basically designed for this kind of stuff and should eat any reasonably simple model for breakfast in terms of running in real time with scant code optimization and analytic analysis; just good old number crunching would probably suffice even on something analytically inelegant like FDTD data compared to other CEM approaches.
And tying the analytics into your favorite higher level tools like Python, Julia, Scilab, Octave, Jupyter, R, OpenCV, matplot, VTK, HDF, whatever should be easy.

I am sure you are correct in thinking that there exist tools to do this already that someone in the OSS community has created, and tying this to a higher level tool would likely not be super difficult.

I did not consider simulation because I think for me hardest part of creating something like a SAR imager is the hardware due to my lack of experience.  If I cannot figure out how to make this then there is no point in spending time writing all of the software to support it and do calculations for it.  However, I think that something like this could come in extremely helpful when testing the software, not needing to turn something on or go to a location in order to test on a new data set or something. Thanks again.

[sckzor]

Make sure you read up on Shannon-Hartley theory. 300 feet is for WiFi data transmission; the same amount of power could go many orders of magnitude further if you're only receiving/sending information slowly.

I am familiar, I was making a general guess as I was writing the post,  as the signal propagates it is inevitably going to become harder to detect because of the affects of propagation but also, and probably more importantly, the background noise on that band.  I think it would be hard to expect more range than my original estimate unless I was in a field in the middle of nowhere or somewhere else with minimal background noise.

[sckzor]

First I apologize for taking so long to do this, I got super busy during this week.  In the spare time I did have though I was able to make the corrections to the circuit and rework some things, most notably the triangle wave generator.  I made the ramp speed variable with a potentiometer and switched from using a 555 timer to another quad op-amp for the design.  If someone could quickly glance over that circuit, I would appreciate it.  Normally I would just order a couple of op-amps and try it out but I don't want to pay shipping on everything twice, I am not ready to order the rest of the parts yet and I have run out of op-amps.  I also did some reading/video watching about the things that everyone mentioned, it was all very interesting.

I have attached to this post a link to the updated schematic and I will begin working on a PCB shortly.  Any other comments, suggestions or critiques are welcome!

Thanks.

Schematic link: https://files.sckz.org/uploads/1660945702684486670.png (higher resolution now too)

[sckzor]

Do you plan to use something like OSHPark or hand-mill/etch a phenolic copperclad board?

Yes, probably JLC, I have used that one in the past.  I would like to try making my own pcb at some point but I think impedance matching the traces would be too hard and PCBs from a fab are really high quality and pretty cheap

The difficulties you'll run into are probably not the parts or schematic directly but rather designing/assembling the PCB.

I think assembly will be the hardest with lots of small SMD components. I plan on using one of those SMD soldering plates. I have made one in the past for other things, and while it looks a bit janky it works good enough.

[Yusuf Alsaedi]

I may have to make one of those janky SMD soldering plates

[sckzor]

Firstly, I apologize for necro-ing this thread but this doesn't really belong anywhere else so I am putting it here.

Secondly, after doing some more design work, ordering parts and manufacturing PCBs I have finally built the radar system outlined above.  It took longer than I thought it would because I got busy and there was an error in the first revision of my PCB but it did finally happen.

I have not done any analysis of the performance of the system yet but it does appear to be working.  Once I do do this I will post my findings here again.

 I have attached a photo below.

Thank you all for your help,

sckzor

[sckzor]

Good afternoon,

I have started to test my radar system, to do this I have connected the sync pulse and the signal from the video amp to a Raspberry Pi Pico that I have performing snapshot captures like a rudimentary 50 KHz oscilloscope.  This data is then sent over the serial port to my PC.  I then take the data that I capture and feed it through a Fourier Transform to see the frequency breakdown of the data. This system all works well, however, the problem is that the received data seems very odd to me.

Attached are photos of the full screen data capture, the data capture and the data after the FT and an image of the data, the data after FT and a colored historical spectrogram. 

Attached to this post is an image of the scope recording the same wave form to show it's relation to the sync pulse and to prove it is not some sort of fluke with the Raspberry Pi.

Starting about halfway up the chirp waveform there seem to be some response waveform.  This doesn't really make sense to me because the signal being transmitted is continuous I should be seeing a constant response wave form with a relatively constant phase shift of the TX signal shown in the output.
This said, it is pretty obvious that something is happening.  When I move a surface back and forth in front of the device, something happens to the wave.  Mainly it seems that the large lobes on the graph move up and down and change their size.  I really do not know what to make of this change.  There appears to be no appreciable change in the FT breakdown of the frequency when this happens.  I have tried everything I can to help remove interference in the system with shielding, isolated power supplies and such so I am fairly confident that this is not being caused by interference. I have also beeped the entire circuit so I do not believe that it is being caused by shorts or something either.

I have attached an image of the schematics of the final circuit that was printed on the board.  A couple of small changes were made but nothing large (I think).  The only part that has changed since the original BOM was the Raltron VCO which was switched with an equivalent from Crystek because the Raltron part was out of stock.  The link to that datasheet follows if you are interested. https://www.crystek.com/microwave/admin/webapps/welcome/files/vco/CVCO55CC-2230-2430.pdf

Any insight anyone can provide about this problem would be extremely helpful to me.  I do not have any equipment capable of measuring such high frequencies nor do I know some place that does so intuition is the only debugging method I have for the high frequency component.  Hopefully the issue is with some low frequency component as that is much easier to work with.

If there is any extra information that you need or anything you would like me to try, let me know.  I would be happy to do it.  Things have slowed down for me a bit now so I should be able to do it in a relatively timely manner.

Hopefully I do not have to write this project off as a failure.

Thanks,

sckzor

Edit: Embeds from attachments don't seem to be working.  All of the images are attached.  Which one is which is pretty self explanatory.

[radar_macgyver]

My hunch is that something's fishy with the IF amplifier.

I would try to probe the output of the mixer, possibly after disconnecting the 1 uf capacitor coupling cap. Another point to note is that the mixer output is not properly terminated to 50 ohm. Finally, to minimize the number of variables, connect the TX output to the RX input (without an antenna), but use an in-line attenuator. With no antenna, you should really only see a DC output from the mixer, maybe with a small discontinuity at the ramp transitions.

Could you post a scope shot of the mixer output, both with and without a 50 ohm termination? You may need to crank up the gain of the scope quite a bit to see anything.

[sckzor]

Hello,

Unfortunately I do not currently own an RF attenuator.  I will order something like this https://www.amazon.com/Digital-Attenuator-DC-3-8GHz-Parallel-Attenuation/dp/B0828GQ672/ and some extra SMA patch cords assuming none of the information I have below shows some sort of glaring fault in the design.

Now that you mention it, back when I first began investigating this problem a week or so ago I do remember taking the hook attachment thing off of my scope probe and pressing it up to the IF leg of the mixer.  When I did that I was just looking to ensure that there was some sort of output and I was not amplifying noise or something.  At the time the output looked correct but now that I take a closer look into it I am not sure that it is.  I would think that because the distance to the object is not changing, the reading should have some sort of standing wave that is not correlated to the ramp frequency, however, it seems that in all cases it is correlated to the ramp.  Stranger yet, both with and without the coupling capacitor (i.e. regardless of the state of the video amp connection) I see the wave form is relatively stable and has a very noticeable and somewhat predictable change when I point the device at different distances.  When I added the 50 ohm terminator I no longer saw a stark difference and the wave was much less stable.  Perhaps the load of the terminator is destroying the useful signal or maybe my cheap probes are attenuating the signal before it ever reaches the instrument?  I have no idea what to make of that change.

As an aside I suppose that when I was designing I assumed that the mixer output was not a 50 ohm signal line hence the lacking terminator, in hindsight it makes perfect sense why it would need one but I never made that connection in the design phase.  I used the 50 ohm terminator built in to my scope for the readings.

Thanks for the help,

sckzor

[radar_macgyver]

So it looks like there's way too much gain on the video amp, causing it to saturate and hit the rails. You may want to breadboard a gain stage that's more appropriate, using the oscilloscope to verify that the output is an amplified version of the input.

The termination resistor needs to be applied to the output of the mixer. Just turning on the 50 ohm termination while you're using a probe is no good, that's meant for if you're using a coax signal fed directly into the scope. Solder a 50 ohm resistor (preferably SMT) to the mixer output, and probe as before. I would then design the video amp to amplify the output of the terminated mixer output.

For an attenuator, get something like this instead: https://www.amazon.com/dp/B091KGWJ2M/

If possible, look for used fixed SMA attenuators on ebay or other places. The Amazon ones may work at 2.4 GHz, but will likely have poor VSWR.

[LM21]

I wouldn't use LM324 anywhere. It is slow and its output voltage range is limited. Like others have said, there are opamps that work with +5V and ground power supplies. LM324 is maybe 50 years old design, and it shows. At 5V and ground as supplies, its output can be only between 1-3.5V or less.

Some PCB houses also offer PCB assembly at reasonable cost.

NanoVna costs only about 60-1500£$€, depending on how high frequencies you want to use.
Tiny Sa and Tiny Sa ultra wont cost much, either.

[sckzor]

The LM324 is not used anywhere.  The two op-amps used on the board are Texas Instruments LMC6484s.  To my knowledge these have better performance than the LM324.  I was testing the chirp VCO and video amp using the LM324 because I had them on hand and that is what others who have done a similar project used (if my memory serves), however I removed them earlier in the design process at the recommendation of the others here.

The NanoVNA and TinySa are very interesting.  If I were to get one I would probably build a NanoVNA because it is probably cheaper and I have all the tools to do it.  If I am still having issues debugging this or I want to go even deeper into RF once this is over I will definitely keep it in mind.

Thanks!

[sckzor]

I have just terminated the signal from the mixer and measured it on the scope.  It does appear that the signal is hitting the rails of the op-amp when it crosses over zero volts.  I will have to redesign the circuit to accommodate this like you said radar_macgyver.  If you have any input on the best way to implement the redesign of the video amp let me know I am open to any and all suggestions.  The photos of the scope readout with the 50 ohm termination are attached to the post. 

I also ordered the attenuators and extra coaxial cable so I should be able to test without the antennas by sunday-ish.  I will update the thread at that point.

Thanks!

Edit: Photos weren't there for some reason

[profdc9]

You might be interested in this development board:

https://omnipresense.com/product/ops243-c-fmcw-and-doppler-radar-sensor/

It is a 24 GHZ band FMCW radar.  Even has built in FFT, but it will return the raw quadrature demodulated return data.