#181

[@MartinBgelund]

This is very interesting, I'm looking forward to see more of the innovative details.
How about pedestrians, skaters runners, etc in the same speed range? How will you avoid counting those as bikes?

[@AndreasSpiess]

I think, It will not be universally usable. For bigger roads with lots of traffic maybe a stationary solution is affordable. For smaller roads, this might be a possibility.

[@anlumo1]

Skaters might be a problem, but pedestrians and runners are slower and it might be possible to filter them out (as long as you're not measuring close to an intersection). The bigger problem is probably groups of bicycles.

[@MartinBgelund]

anlumo1 There are fast walkers and slow bikers, so filtering will be based on assumptions, not on certain attributes.

[@anlumo1]

Yes, but I don't think that a 0% error rate is required. In addition to that, bicycles have a minimum speed they need to operate at, otherwise they topple over.

[@Sir_Trip-s-A-Lot]

Can easily be addressed by magnetic field sensor alongside. The only foolproof way is to distribute free cycle only RFID stickers at city hotspots. That will solve all the problems

[@hvanmegen]

2 years in: still loving the guy with the Swiss accent.. my #1 go-to guy when I want to be informed about IoT, electronics and home automation!

[@AndreasSpiess]

Thank you very much!

[@mbaker335]

A very interesting video in an area I will experiment with myself. I lived in CH for seven years returning last October to retire in the UK. The pictures make me feel homesick in a strange way. Lovely country. I miss it a lot.

[@AndreasSpiess]

Thank you for your nice words about my country. You are always welcome as a guest here in CH ;-)

[@johnwest7993]

I love this channel. It's one of the few things capable of pulling me away from my books and my work-bench. Your body of work here is a modern-day Principia Mathematica for the electronics student, hobbyist, and even a 'How To' for engineers. This is what every RUclips channel should strive for - high quality content. Thank you.
I was just thinking that now, 4 years later, we could probably do all of this, including everything from speed capture, to direction, to identifying the moving object, all with just an esp32 and camera combination or something similar in price, and of course a whole lot of mathematical analysis of the video pixels. Everything has been reduced to ever cheaper sensors and an ever-growing, ever-improving body of sophisticated, free firmware. We live in amazing, (and fun,) times.
You had mentioned to me that electronics engineering has of necessity moved from fundamental physics to ever higher levels of abstraction with correspondingly different bodies of knowledge, and you were so right.
BTW, I once escorted Bill Hewlett down to the security guard at the front desk at HP in Colo. Springs. I worked making CRT's on the night-shift there in a near empty building. He was nosing around the R&D cubicles without a badge, and I didn't know who he was. :)

[@AndreasSpiess]

I abandoned this particular project because long range radar modules still were expensive. And with the pandemic this became worse...
But I agree, we live in a wonderful time for electronics.
BTW: I assume Mr. Hewlett was ok with what you did. You tried to protect his company after all!

[@johnwest7993]

That is exactly what he told the security guard who was ready to eject me from the premises.

[@dl8cy]

What an incident - yesterday in the evening i thought about that and now the Swiss guy shows solutions. Keen on part 2 of the Video. Thank you and 73 from Germany

[@AndreasSpiess]

Maybe it is not the complete solution. Still a way to go ;-)

[@Thesignalpath]

Thanks for pointing to my channel. :)

[@AndreasSpiess]

You deserve it. I like your channel!

[@schmohawk_]

this is hardcore!

[@PanSatyr]

This project inspired me. I now have this setup working with weatherproof neopixels that create a display based on the speed measured. Thank you.

[@AndreasSpiess]

Good luck with your project!

[@TheRealKitWalker]

2:04 "...in wrong hands can cost us a lot of money ": shows a picture of policeman. Hahaha clever and rightly said. 🤭🤭😂😂😂👏👏👏👍👍

[@AndreasSpiess]

:-)

[@TheRealKitWalker]

@@AndreasSpiess hehe :P but I'm glad that you actually make the gadgets I think of making but I never find the right sensors for them. The buying links you provided are a life and time saver. Thanks so much. Love your videos always. At present I'm researching to make a similar project using a 60 Ghz RF CMOS sensor to track and position a moving object in the field of sight and also measure their size, weight, heart beat, etc... with it. Possibly an idea for your next video? ;)

[@AndreasSpiess]

I thought these modules are still too expensive...

[@TheRealKitWalker]

@@AndreasSpiess Hmmm yes, I just found out the exact price + shipping. turns out there's hope for cdm324 to achieve what I want with the help deep learning. Hopefully, I'll start soon on this. Thanks Andreas.

[@ErnestGWilsonII]

Wow, an awesome video! I am hoping that you can hook up a display or send a signal back to MQTT with the speed data from the object and maybe even trigger a camera based on the recorded speed of the object. My love of electronics and RF are always rekindled when I watch your awesome channel. As an old school electronics technician myself, I wish we were not separated by such a large geographic distance. I am slightly jealous of your assistant in the video, having access to the master Andreas Spiess, I hope he is aware of how lucky he is!. I am fondly reminded of days long ago when my old crew had a lab and we built various electronic devices, this is why I love this channel! I hope you win the contest! Thumbs up, loyal fan / subscriber, -Ernest 73s W3TCP

[@AndreasSpiess]

You see that my heart is also moving a little faster if RF is involved... Distances fortunately are no more as separating as they were 100 years ago. So, maybe we can once drink a beer together. You never know ;-)

[@johansoloshotuploads8386]

Hi Andreas,
What a nice subject for our makers. A lot of people think of using this technique in other purposes than it was supposed to, just for switch on lights or opening doors.
The thought’s I had where quite te same as most of the repliers, but some questions stayed after reading al the comments and answers and other posts in other channels including searching AliExpress for the most suitable radar module.
My application is quite standard for radar and is also almost identical as what you did.
I want to display the speed of Skiracers at the piste on a display and send the speed to a central server to store the speed, in combination to the time registration from another project.
My first plan was to set it up with two optical reflective gates, at a distance of some 5 meters and calculate the speed. This is common technology, but the racing suites are often white and do also reflect the optical beam when passing the speedtrap.
So a speedtrap with separate beams and detectors looks more reliable, but takes to many physical components to set up. Besites that, synchronization of the time base on both the traps for precise messurement and setting the speedlength is also a hassle.
Then, as a recent viewer of your video’s, I found this item about speed measurement with cheap components. It surprises me, how simple the units are and how accurate the measurements can be. But when reading more about these units and solutions people talked about the maximum speed that you can measure with this solution. Our racers can go as fast as 120Km/h, so the limit can be an issue. Also the measuring distance can be an issue, but 10 meters is quite nice for this. But are the racers then not passing the sensor to quick to make a reliable measurement?
And if the sensor is ‘always’ on, for detecting movements, how can I trigger the right moment of passing at the highest speed, without showing zero’s after the racer has passed. These last questions are for me to solve as a maker, but the question about the limitation of the speed measurement is tricky for me. The second thought is which sensor to buy. In the meantime some more sensors where added to AliExpress and there is a wide range of types now, 10,525Ghz, 24Ghz, 5,8Ghz etc.
The new CDM324 is 24Ghz and is advertised as 15 meter detection range. It looks as this unit is simplified version of the CFK024-5A unit, which is much more expensive, but has more pin’s and options, including FMCW/FSK/CW.
So if anyone had built a similar speed measurement unit without that works, please point me to more details to prevent me from spoiling precious maker hours.
In the meantime I’m looking forward to the sequel of this amazing video series.
Regards Johan

[@AndreasSpiess]

We worked a little further on the project, but it is not ready yet. The most difficult part is the reach. The max speed should not be an issue.
You need a very high amplification of the signal (>10'000x) and have to pay attention that you do not introduce too much noise. After that, we used an ESP32 with FFT to measure the frequency. You measure the frequency all the time and distinguish from the pattern if it is valid. Your application will most probably show 0 for most of the time and increase fast, and fade off to zero again. So you can detect the maximum. Pay also attention that the angle of measurement has to be integrated into your calculation. If the angle changes the result also changes.
All-in-all I think this will not be possible with this device because of its short range. You would have to place it in the direction of the racer in a distance of less than 10 meters, which is probably too dangerous.

[@johansoloshotuploads8386]

Hi Andreas,
expected something like that. Nice that the speed will not be a problem.
What will happen with the accuracy if a racer moves by in a elliptical curve instead of a straight line? For the noise I can try to get a low noise amplifier with a high gain for that. I first wanted to build my own parabolic antenna, but the gain factor of the module is already quite good. If I manage to create a more directional antenna with a better passive gain, I’ll let you know! Probably I will use something like a feedhorn for the receiver.
Regards, Johan

[@AndreasSpiess]

1. The measured speed can be calculated by trigonometry. The angle has to be absolute constant.
2. I cannot imagine you will be able to connect an amplifier because this device sends and receives at the same time. I assume a horn antenna would help if done properly.

[@johansoloshotuploads8386]

Hi Andreas. Yesterday I contacted a Colegue who is a maker and a Ham specialist in microwave connections.
On internet I found the Schwarzbeck horn-antenna's with an incredible passive gain. My collegue suggested that such a horn will solve the reach issue. Even when you use a horn that is not very good adapted for this frequency, it can give a huge gain. So I'll try that first. Just build a horn with a divider in the middle, with the antenna's in the bottem. The mesurements of the bottem are larger than the calculated dimensions at this frequency, but the expectation is that a reach of 50 Meters is possible. Amplification would be possible, but takes a complete rebuild of the unit. More expencive units on Aliexpress use more stacked antenna's to reach a higher gain. The radar used by the police uses a horn. The horn will also give a smaler openings-angle, but for my usage that is an advantage.
Regards,
Johan

[@JoeBee999]

To messure only frequency, the signal is very often staged throug a comparator. This gives a nice square frequency. As far as i know, Atmel AVRs have one build in. Or just use a OPamp. Then feed to an input capture pin and the time os messured.

[@AndreasSpiess]

That would then be version 1. I have also to check if the digital pins of the ESP32 are Schmitt-Triggers

[@augurelite]

This was one of the most interesting videos I've seen in a long time! Very cool!!

[@AndreasSpiess]

Glad you enjoyed it!

[@raghu9270]

Hello Guy with the Swiss accent,
I am a great fan of your channel and thanks to you I have built a lot of projects and Every one was successful. You are truly a great guy.. Can you do a project on building a speed gun radar detector to be used in the car using commonly available sensors and microprocessors. It will be really great project. The currently available sensors in the market are pretty expensive. Also as a maker it will be great fun to build one on our own.
Thanks

[@AndreasSpiess]

Unfortunatly, the cheap modules are not good enough for that purpose and the good ones are very expensive :-(

[@charlesdorval394]

That was an awesome video Andreas! Thank you!

[@AndreasSpiess]

You are welcome!

[@deltaxcd]

Rather than doppler effect, it is easier to understand as interferometry because you are sending and receiving the same signal.
So this sensor can do a very accurate measurement of movement distance, just count output pulses

[@AndreasSpiess]

These sensors only can detect speed, not distances.

[@deltaxcd]

@@AndreasSpiess
well, you should just try it then. like I said rather than connecting speaker connect led and you will see how it will flash when you move reflector by about 1.5 cm
I tried it and it works, although it was lower frequency radar

[@JerryEricsson]

In 1974 the Department I was working for got our first updated digital display radar. It was X-Band. I was on duty one morning (read night shift near ending) and since there was nothing happening in the town, I parked the squad car under the sun shield over the A&W drive in, to watch traffic on the Highway that went through town. I was not running Radar at the time since the angle was too great to get any accurate readings, but it was on, and the volume on the head was turned up a bit. I began to hear ZIP! and a bit later ZIP! there were no cars but on the digital read out I was getting 104 MPH and faster flashes. As I studied this phenomenon I noticed that the swallows who nested in the over head cover were diving down to the ground after worms or insects, the radar was reading their speed. I began aiming the cone at the particular birds and the readings got faster as the angle decreased. It was actually a fun thing to do with the big old units.

[@AndreasSpiess]

Nice story! So this radar was quite sensitive. Detecting the water in these small birds.

[@roryfarrell7292]

Looking forward to seeing your implementation of the theory.

[@AndreasSpiess]

It will take a while

[@CroutonBrown]

Abso-smurfly wonderful video you’ve got here, mon capitain! Thank you for providing this content!

[@AndreasSpiess]

My pleasure!

[@BloodBlight]

Was waiting for this video!!! Awesome! Looking forward to knowing the accuracy of these two sensors at different speeds!

[@AndreasSpiess]

They are accurate. But the range is quite limited.

[@AdilKhan-jp7hn]

Excellent video and very interesting sensors. Please let's build a radar gun. Awesome

[@AndreasSpiess]

Thank you!

[@deangreenhough3479]

Another fantastic video😀😀👍
You make it sound so simple and after the next video. I am sure it will. Bravo Andreas and congratulations on your 257 Patrons which appear to be growing by the day.
As always your outstanding efforts are always worth the time to absorb.

[@AndreasSpiess]

I am not sure if the next video will be easy. I already made some investigations... concerning Patreon: I am still astonished about the support I get from my subscribers.

[@Lasseu]

Hello from Finland. I would love oscilloscope, maybe I'll buy one after I get work, if I manage to get one.. :)

[@ericdormer2196]

One comment I will give.... since Patreon takes a cut of your "earnings" you may want to elimininate the middleman and just take funds directly. In my case I would be happy to send you a monthly amount from my paypal acct, and skip patreon. I am just one subscriber so its not many dollars, but if a lot of people took this approach, it is more money in your pocket. Perhaps worth thinking about On the other hand, perhaps you want to assist in funding the Patreon site, (everyone needs income of course) so that might be a reason to keep using them. Cheers.

[@deangreenhough3479]

Eric Dormer
I would also be happy to pay direct

[@gweliver]

Great project, can anyone direct me to the follow-up video (pt2), thanks for all the great content and detail(s) you provide.
You are someone we can trust, for sure!!!

[@AndreasSpiess]

There is none for the moment.

[@tylerrourke50]

I would also like to see a part 2. Great video!

[@ZlayaCo6aka]

The purpose of the "unknown material" at 12:10 is to prevent any resonance within or oscillation due to the shielded enclosure; in this case, it appears to be the ubiquitous "conductive foam."

[@AndreasSpiess]

Thanks for clarification!

[@Zplizor]

The bicycle demo is really good to explain it

[@AndreasSpiess]

Thanks!

[@benediktkohler28]

Your videos always make my day, danke Andreas

[@AndreasSpiess]

:-)

[@spikeydapikey1483]

Looking forward to part two. Cheers :o)

[@AndreasSpiess]

Will take some time...

[@WolframWebers]

I seriously like your videos and learned a lot about electronics (I'm more into software engineering). Please, keep up. You're awesome.

[@AndreasSpiess]

Thank you!

[@rklauco]

Excellent idea! I have to test it. Fun fact: I thought it's elektrovelo before I realized the sound is produced by the amplifier and speaker :)

[@AndreasSpiess]

I am still too young for an e-bike ;-)

[@noweare1]

I was confused. When you showed the first picture of the mixer I thought the output went to the transmitter but it actually goes to the op amp for amplification. Ugh ! I must of watched that part of your video 4 times until further on the video cleared when the receive & transmitter antennas were shown in the schematic. Awesome work. Thank you.

[@AndreasSpiess]

You are welcome!

[@betojasz]

Mannn, your videos are magic!! Great idea to put sound to hear the distance!! This is more than just science!! its fun!! Thanx a lot!!

[@AndreasSpiess]

You are right. We had some fun, too...

[@judgeguilty]

Excellent subject and project idea. You pedal that bike around like a 20 year old. I was waiting for tricks - riding on one wheel, standing on the seat, etc.

[@AndreasSpiess]

When I was younger...

[@ZillionPrey]

Thank you for your video, I am also planning on measuring cars next to my house. It is allowed to drive 100 here, but most people drive 150 if not faster.
So I wanted to make an education project out of that and measure cars and save it in a database. This helped me!

[@AndreasSpiess]

You most probably need a stronger device because of the needed range.

[@ZillionPrey]

@@AndreasSpiess we have tree directly next to the street, it's about 3 meters to the center of the street, do you think the HB100 is enough?

[@schluderjupp]

Servus Andreas, Here you amaze me once again with a topic that I did not see coming. 😀 Keep up the good work. I'm looking forward to seeing more on this very interesting topic.

[@AndreasSpiess]

Thank you!

[@chuxxsss]

Msybe I should have done this, by the way use the 24 Ghz and some range sensor in software. That is correct the screw is a fine adjustment. You need range gates for distance as well. Doppler radar maybe you could use pulse radar. Nice work Andreas.

[@AndreasSpiess]

In the description, they talk about pulsing the radar. I do not know why I should do that other than saving energy.

[@kd4pba]

Wow, This is how you make a video. Impressive on so many levels.

[@AndreasSpiess]

Thank you!

[@rodinsocaciu]

Highly appreciated your videos. Great when such initiative could get helping. Have you considered looking over a AC Volt&Amp&Watt Meter WiFi logger?

[@AndreasSpiess]

You can buy one from Sonoff. I made once a video about it. But maybe I will cover this topic in the future again.

[@Tarbard]

It makes a nice sci-fi noise too.

[@AndreasSpiess]

:-))

[@argcargv]

I like the direction this project is going. :-D

[@AndreasSpiess]

:-)

[@davidkempton2894]

I was so worried about you on that bike that I decided that I had better become a Patreon!

[@AndreasSpiess]

To support my widow if I crash? ;-) I drive around 4000km per year with my bike. So no problems for these short tests.

[@davidkempton2894]

Andreas Spiess I was hoping that it would encourage you to drive even faster! 😂

[@noamw3841]

My mind is blown, can't wait for more.

[@AndreasSpiess]

:-)

[@koherer3732]

You don't need an expensive microwave osciloscope to watch the frequency of HB 100. You can get a satellite tv LNB, inject 12v into coaxial using a bias tee (inductor to stop RF getting into power line and a capacitor to block 12v from receiver). And then connect it to RTL SDR receiver. Then tune to 600-700 mhz range.
Signal is well... not that unstable to jump to 9 GHz. If you feed the audio trought a capacitor to +5v in of HB-100 while powering it at the same time, you will get FM radio at 10.5 ish ghz.
Device is low powered but if you worry about how legal it is, there is 3cm ham radio band you can think you're using, if you already has a license. I think any polarization LNB will work, I used linear. And aware of harmonics, because i've received some weak signal at 400 mhz and thought it's working that bad, but real signal was higher.
SAT LNB lowers frequency by 9750 Mhz (i guess, google for precise value). BTW, this signal is slightly out of LNB band, but everything works ok.

[@koherer3732]

PS. Yes, signal is drifting when you touch the module, and has wery low drift when just listening. I don't know if it's a HB100 or LNB, maybe they both drift. I have to tune frequency when listening to WFM audio, but I saw a guy using some sort of "peak hold" function in SDRSharp on youtube.

[@AndreasSpiess]

These modules are used to open doors. So I assume the power is below the law. But I am no layer ;-) Thanks for the tip with the LNB. BTW, I have a HAM license (since 40 years).

[@AshMcKenzie]

Cannot wait to see more in this series!

[@AndreasSpiess]

:-)

[@WPBusinessclub]

great explanation and wait to see more. Love the idea of a budget speed camera

[@AndreasSpiess]

It is not so easy as the range is very limited :-(

[@wowtah]

Very interesting! Can you please go into more detail on the direction detection? With the Ali sensors in particular :) keep up the good work!

[@AndreasSpiess]

We will share our results in a later video.

[@klassichd10]

Thanks for this introduction and the Pictures from inside the CDM324. I did not open the mine. I only had to open one of my HB100 arriving with a loose rattling disk of this dielectric resonance oscillator.
The arduino FFT works with the ESP32 just fine out of the box. Even with 1024 and 2048 samples.
I am very curious how the InnoSenT quadrature radars will perform and what they cost.
Also the AGC amplifier could help. Sounds promising.

[@AndreasSpiess]

I also played a little with FFT and the ESP32. So far I do not yet know which amplitude gives the best result nor which is the right sampling speed or # of samples to get max speed for the frequency range needed. Do you have any info about these factors?
The InnosenT costs around 30 Euros at Conrad. AGC amplifier is on order...
BTW: I opened mine because I killed it somehow. Don’t know how :-(

[@klassichd10]

I used a biased signal with 1.6V virtual GND. Try to get the signal in front of the output cap.
The ADC of the ESP32 is poor (www.esp32.com/viewtopic.php?t=1045) , so you have to use the preamp to get several hundred mV.
Using 24GHz you get a Doppler shift of 3400Hz @76km/h and 40Hz @0.9km/h.
To avoid aliasing you should sample at least 6400 samples/s (Nyquist-Shannon). Try 8000 samples/s which could work. Otherwise use 4000samples/s which works but limits you to ca. 37km/h.
If your radar also sees faster vehicles like cars, you should use a low pass filter as an anti-aliasing filter.
To be precise @ low speed you should measure 40Hz with maybe 10%, that means 4Hz.
To achive a resolution of 4Hz @ 8000 samples/s you need N> 2000, that means 2048 samples per conversion, or 1024 @ 4000samples/s.
This leads to a window time of ca. 1/4 sec in which the speed should be approximately constant. Looks reasonable.
You find some formulas e.g. in home.zhaw.ch/~rur/dsv1/unterlagen/dsv1kap3dftfft.pdf public.beuth-hochschule.de/~mixdorff/mmt1/files/spektralanalyse.pdf home.zhaw.ch/~dqtm/dsv1/vorlesung/dsv1kap3_dftfft.pdf
You should use a Hanning window (arduino FFT provides it) to reduce the effect of a finite oberservation time (rectangular window).
arduino FFT is a complex FFT. You feed the ADC samples in the real part and flood the imaginary part with 0 - when using the cheap radars.
Using the the quadrature radars you feed the quadrature signal to the imaginary part.
arduino FFT performs the complex FFT and has a function to calculate the magnitude vs. frequency and to pick the highest magnitude. This maybe a bit crude and you may add a function finding several local minima.
And extracting the phase (leading to the direction) is not provided by arduino FFT out of the box as far as I remember.

[@guilhermefraga4349]

Thanks Andreas, Great info as always

[@AndreasSpiess]

Thanks!

[@peterleather6866]

Another fascinating and informative videos Andreas! I wonder how the sensor will cope with more than one bicycle at the same time? Will the sensor output multiple frequencies for the different speeding objects? Also if two bicycles were close to the same speed (side by side for example) would it be impossible for the sensor to tell. A very small limitation perhaps but a very interesting sensor.

[@AndreasSpiess]

It will have its problems with multiple bicycles, for sure. But maybe it will be exact enough for smaller roads where a cheap temporary solution is just right.

[@Gunshot115]

This is a wonderful video. Thank you sir!

[@nathanas64]

What a terrific educational channel!

[@AndreasSpiess]

Thank you!

[@3dgussner958]

Thanks for the video! Just checked the price for your new toy, WOW not cheap !

[@AndreasSpiess]

As I said: only possible with support.

[@Rcbeacon]

You could try adding a horn to the antenna to make it more directional and longer range. I have a couple of old 10GHz speed guns dating back to the early 70s which use microwave diodes in cavities attached to a horn, dedicated 7400 series logic and nixi tube display.

[@AndreasSpiess]

10 GHz and antennas is not easy. Each millimeter counts. And I am no specialist. Maybe somebody knows a reference design?
Your radar guns must be very valuable these days. With nixi tubes. Cool!

[@PhG1961]

Great and surprisingly interesting video. Btw, I defenately would like to have that Keysight equipment. I remember the HP and Agilent as being excellent... but also very expensive.

[@AndreasSpiess]

Yes, it is a nice device. And I am happy with it :-)

[@GeoMikeCache]

Nice oscilloscope! A little bit pricey, but good. thumbs up.

[@AndreasSpiess]

Depends on where you get it from ;-)

[@Tolissere]

What an interesting project. Great work!

[@AndreasSpiess]

Thanks!

[@ZlayaCo6aka]

The "small screw" you mentioned is not part of a capacitor. At about 6:08 in the video you see a small circular white object; that's a ferro-resonator. The "small screw" is ferrite and its proximity to the resonator adjusts the resonant frequency.

[@AndreasSpiess]

So I was „half right„ with the purpose at least. Thanks for the clarification!

[@gandsnut]

The change in towards/away sound (in this demo) seems opposite to 'Doppler' effect... As the ambulance siren comes towards, its pitch sounds higher, when the siren is just by us, it sounds 'normal', then when siren passes away, the pitch sounds lower. But this is a moot point, because we can interpret the radar sensor output however we want.

[@AndreasSpiess]

The relevant for me is my measurements. And there direction does not matter ;-)

[@joinedupjon]

the difference between this radar and sound waves is that you can hear the pitch change directly for the ambulance whereas the radar reflection is processed to produce a 'difference' output and it's the difference output that you hear in the video. negative difference sounds exactly like positive difference. also we'd expect the difference signal to be zero at the instant the bike passes closest to the radar - for the siren this would be the instant we heard the 'true' frequency.

[@RichardT2112]

Great video as always sir!

[@AndreasSpiess]

Thanks!

[@JungleJake1664]

Excellent presentation and infomative, well done.

[@AndreasSpiess]

Thanks!

[@rjrodrig]

Why can you look at the phase of the signal. You can use a down converter approach by using I and Q signal split and get the phase information. I suspect you can detect direction. Using an FPGA you could implement a down converter

[@AndreasSpiess]

May

[@faidularcs]

Awesome project

[@AndreasSpiess]

Thanks.

[@MrAtarifreak]

Nice video, I look forward to the next video. Thank you for the Information and for sharing it with us.

[@AndreasSpiess]

You are welcome!

[@richardphillips2405]

Great video. I will have to try this. I was wondering about using the microwave sensors for range detection (distance).

[@AndreasSpiess]

These sensors only detect speed, no distance.

[@richardphillips2405]

@@AndreasSpiess This is sort of what I thought. It probably could get real complicated to build (diy) a distance measuring radar. What about FMCW (frequency modulated continuous wave type sensors).

[@AndreasSpiess]

The FMCW shoud do the trick. I one bought one, but unfortunately do not find it in my lab anymore :-( They are quite expensive...

[@offgridsweden]

Great video! Greetings from Andreas on Off Grid Sweden

[@AndreasSpiess]

Greetings back!

[@NorbertHarrer]

Awesome video. I was also using an HB100 in combination with an arduino for simple motion detection. I am looking forward to your solution.

[@AndreasSpiess]

We will see how much we get out of the data. I hope more than detect motion...

[@NorbertHarrer]

Yes, that's a lot harder. I have built and used the reference amplifier that is specified in the HB100 data sheet (the one with the two LM324 opamps). It outputs a square wave with a frequency proportional to the speed of the moving object. The square wave is basically a digital signal. So one could count the 0-1 transitions per second with a moving window. Might be easier then FFT. I only used it for motion detection though. Also note, that I had to clean up the signal a bit. At no movement the signal sat somewhere between the low and high point. And also when an object was just about to enter the field of view the signal was not a clean square wave but some garbage. I used an analog input and cleaned the signal up in code with a threshold detector for the high and low levels. - Greetings from Austria.

[@steven_jahn]

I won an 2 channel oscilloscope last year after someone was caught cheating. They let me know in May I had won on the redraw and I got my scope in July!

[@AndreasSpiess]

Thanks for your feedback!

[@ebb2421]

the screw of the HB sensor does shift the freq, confirmed with HP5350B, have fun

[@AndreasSpiess]

Thanks for the feedback!

[@sylvainlazare1114]

Good work and very convincing presentation. What happens in the caseof more than one moving target? you would need a more discriminating software? or do you have simple solution?

[@AndreasSpiess]

I do not think we can discriminate several bicycles.

[@VideoDetection]

Super interesting subject and well presented!!!!!

[@AndreasSpiess]

Thanks!

[@thenooscoper64]

This video is really helpful, thank you!

[@AndreasSpiess]

You are welcome!

[@gonzalodanielperez87]

Gracias! Excelente video! 👌

[@AndreasSpiess]

You are welcome!

[@XTronical]

Wow, great video and so informative. One thing, the diagram near beginning is showing stretched frequency in forward direction and compressed travelling away. Think it should be opposite.

[@AndreasSpiess]

I think, this was a signal from the bench. Depends on the speed of movement

[@XTronical]

Ahh, right, thank you :)

[@albygnigni]

Great video, I suggest you to use two sensors facing opposite directions at least, then with some calculation you can understand who is entering and who is exiting using the Doppler effect formulas.

[@AndreasSpiess]

That is another possibility. Then we also would need two ESP32 if we go with FFT, I think.

[@albygnigni]

the ESP32 has two ADCs so theoretically, if the FFT algorithm is lightweight, it should be possible to do some stereo FFT, maybe using the other core, too. If your point is to build a low power device then this way should be the right one, otherway a Pi Zero W will have the right hardware to do so.

[@mohammedahmed4637]

Great channel I subscribed

[@AndreasSpiess]

Welcome aboard the channel!

[@petermeiner3574]

Moin Andreas, da erwähnst Du mal so nebenbei, das Du ein neues Oszi hast.....aber was für eins!
Herzlichen Glückwunsch! ( ein bischen Neid ist auch dabei)
Bin gespannt auf deinen Test. Da musst Du aber ein gutes Abkommen mit Keysight gemacht ;-))

[@AndreasSpiess]

Ich habe kein Abkommen. Die wollten mich einfach unterstützen (und vielleicht hat sie das Siglent Logo auf meinen Videos gestört ;-)

[@ThomasHaberkorn]

now I know where Knight Rider got their sound effects :)

[@AndreasSpiess]

:-))

[@user-zt9us6ny3x]

First of all, thank you, Andreas, very much for such cool videos. You teach us a lot!)
Please, correct me.
1) "Two zero crosings of the wave" - is a half-period?
2) Period is 150ms. So 1/150 ms = 1/0.150 s = 6.6(6) Hz, not 67 Hz?

[@AndreasSpiess]

1. Yes, correct

[@joinedupjon]

I was going to suggest you fabricate an inside corner cube reflector to increase the detection range... but if you're interested in detecting general cycle traffic rather than optimising your own speeding it's not relevant :)
Audio output sounded like an old fashioned bicycle dynamo IMO

[@AndreasSpiess]

You are right: It sounds like a dynamo!

[@joinedupjon]

I've remembered a humorous suggestion for defeating speed radar in an old motorcycle magazine... attaching corner cube reflectors to the wheels to produce strong return signals with varying frequency shift, the theory was that this would confuse the receiver. doubt it was ever tried - it doesn't seem practical.

[@maximilianibel]

Interesting idea. Maybe install a larger rotating disk with cubes mounted on outside rim on top of car (wheels are probably too low). Could use this project to measure effectiveness of that strategy (or just use the US style speed indicators). I don't know how focused radar based speed guns are, plus I thought many speed guns now work with lasers pointed at number plates and directly measure return times (like lidar/rangefinders). That's much harder to fool.

[@michaegi4717]

Very nice project. I'm really impressed. Please be aware about radar frequency regulations in your country.

[@AndreasSpiess]

These devices are used to open doors in buildings. So they are probably not too illegal.

[@michaegi4717]

I'm not really a radar hardware expert, maybe I'm wrong. But as far as I understood the frequencies are country specific. These china imports might be legal in other parts of the world. At least for Germany I found some hints that the used 10.5Ghz is not allowed and you have to use 9.35 GHz... but handle this information with care... I might be totally wrong. I'm really concerned about the crazy monetary fines in Germany.
Nevertheless I'm still totally amazed from this video. I'll try to talk to a radar expert and try to find a legal sensor for my country. Thank you very much for this fascinating and informative video!

[@manickn6819]

Very interesting and cheap. I am going to put one of those on my list to order.
Small typo in the description. You have the amplifier as LN386 as opposed to LM386.
Recently I used a LM386 on a guitar amp build. I have all the discrete components to make one and its easy to design but at US$1.15 shipped its just not worth the effort and time. Interesting times for electronics with cheap parts and oscilloscopes. Even phones with all their functionality really are a marvel compared to what I had available when I was young.

[@AndreasSpiess]

LN corrected. Thanks. You are right: Interesting times, 10GHz sensors for 3dollars...

[@rawyin]

I'm curious about the amp. Is this really necessary given the arduino and other controllers can handle frequency measurement up to about 8mhz with freqcounter?

[@spacechicken2756]

This is a very interesting project! I am interested in making a little speed gun that would be able to measure tennis serves / baseball pitches. Do you think these components would be accurate enough? If not, do you have any suggestions for better radar sensors?

[@AndreasSpiess]

So far we abandoned this project because better sensors are over 100 dollars.

[@mattjmwmatt]

Could we use this to calculate distance Mr Andreas? Would be very interesting and useful for a robot maybe.
Maybe some phase detection maybe?
Undergrad(soon to be) Electrical engineer here, love your videos !!!
Cheers!

[@AndreasSpiess]

The doppler effect cannot be used to measure distance. In radar they use the delay time for that. But speed of light is very fast and therefore delay times short. Maybe you could use the phase to detect such small delays. These boards do not do this trick.

[@michaegi4717]

Radars that can measure speed and distance do this usually by changing their frequency (ramp). They are called FMCW-radar.

[@aldob5681]

Image recognition seems better suited for people count. And could be trained to report the direction also.

[@AndreasSpiess]

True. But probably more expensive.

[@zephsmith3499]

Regarding direction detection. The frequency change is not very reliable, being based on the detector being offset from the path, but that will vary. However we can easily note that the volume increases for approaching movement, and decreases for departing movement! So we can either measure the amplitude of the IF signal, or if you do use Automatic Gain Control, measure the feedback signal which controls the amplification. That change in amplitude should determine direction in many cases, with a single object. (Multiple objects are going to be hard in any case).
My question is how much of the reflected signal comes from Andreas and how much from his metal bicycle? Given the normal purpose for these sensors, we know they have SOME sensitivity to plain old people, but how much better is it at picking up a bicycle? Experiments to determine relative signal strength would be very interesting.
(I know you've been running up and down stairs to get your heart sensor activated, and now practicing on your bicycle for this; and with my being of a similar age I hate to discourage you from any exercise - but you could have your young assistant run towards and away from the sensor if you are not ready to show off yet).
And it would be interesting to see if you can (to some degree) tell whether somebody is carrying metal - keys, coins, cell phone, a weapon - by comparing signal strengths at the same distance with and without. Does a person carrying a metal cell phone have twice the radar reflection as one without, or just 10% more? I have no idea, but you are in a great position to begin testing.

[@AndreasSpiess]

We for sure have to do some experiments in the future. But first I need a decent AGC amplifier. And no worry. I drive about 4000 km per year with my bike...

[@ON7ARQ]

Nice explanation, but you need a licence to use it and I doubt if it is legal 1 to use the frequency 2 to use it as radar, I have also the HB100 module but for use as radio-amateur for ATV ( amateur television ) or phone

[@AndreasSpiess]

These modules are used in many buildings to open doors. But I am no lawyer... How far can you transmit with this module and how stable is its frequency?

[@gweliver]

Very interesting indeed! I have been think about a speed for cars that warns you of your current speed as compared to the posted speed limit.
The problem is the price? My community has several elementary schools so wouldn’t it be great is their was an affordable alternative whereas affordable LED’s are available now!

[@AndreasSpiess]

I think such a device should be possible with cheap supplies from China.

[@phinok.m.628]

Well, the problem with detecting the falling/rising frequency is that you have to a assume the bike is going at a constant speed. A quickly stopping bike may decrease the frequency more than it's changing proximity to the sensor increases the frequency, causing false readings, not to mention the angle from where the bike comes can change, since bikes don't always drive in straight lines. Another problem is, that multiple bikes driving very closely to each other would likely be counted as a single bike. So if you want accurate readings, this is probably not a very good way to count bikes. If you only want rough numbers however, I guess this would be a reasonable way to do it.
And if you want accurate readings, I guess this method would still be useful, but combined with other measuring methods. Since no method is perfect, but many combined can probably eliminate a lot of errors.
Anyway, I would be very interested in a way to measure distance to cars using for example radar. I've been looking for a reasonably cheap and reliable way to measure long distances (say up to 300 m) and preferably many times per second, but haven't found any way to do it so far.

[@AndreasSpiess]

I do not expect exact readings because of all the problems you mentioned. But I hope the errors will be tolerable for a certain purpose. And if not, we at least tried ;-)

[@phinok.m.628]

Yeah, it's certainly an interesting experiment. I hope you don't take me mentioning likely problems the wrong way. I think it's great what you and other youtubers like you do. Regardless of whether it works as intended or not, it's always good to share your experience with others so they can learn from it.

[@phinok.m.628]

@@AndreasSpiess Btw. what's wrong with measuring the change in amplitude to determine the direction the bike it going? Sure it's would be a little more complicated to do, but I think it would be a more reliable way to do it.

[@connecticutaggie]

I have a pair of CDM324s on order and they should be here in a few weeks. I have a couple of applications in mind. One would be for low speeds (

[@AndreasSpiess]

We are still experimenting. But so far, the reach was never bigger than about 120 meters. So 100 km/h will be difficult, I think.
The signal of the sensor is more in the range of 4 mV.

[@connecticutaggie]

Thanks Andreas, the high speed application would actually be close range and the low speed application would be way less than 120meters. I know that answer may seems bit strange. I wish I could go further; but, I am not ready for a public reveal of either idea right now. Well, anyway, the low speed / low accuracy project is first.
Looks like my senors are arriving this weekend so hopefully I will get a chance to play.
BTW, I PM'ed you on Facebook as well.

[@AndreasSpiess]

I made a mistake: Not 120 meters. 10 meters.

[@janisvaskevics93]

Great video! Subscribed.
Had an idea once - try to make a photo radar for a fraction of price of the ones that government buys. Should be very easy.

[@AndreasSpiess]

Welcome aboard the channel!

[@gijsqwerasdf]

This is so awesome, thank you! One question: is it possible to only detect metal objects? Is it depending on the frequency perhaps? I'd like to detect bikes and cars, but not pedestrians.

[@AndreasSpiess]

Conductivity is what counts. So it will not be easy to distinguish between humans and bikes using radars. A camera is probably better.

[@alessi4249]

Was there ever the follow up to this? This was an excellent video!

[@AndreasSpiess]

No, unfortunately not

[@filipeterra8981]

Nice video! You may also try to use a LM2917 if you want to go full analog

[@AndreasSpiess]

I am not sure how the LM2917 reacts to these varying amplitudes. But anyway, I want a digital signal to transmit via LoRa.

[@korokw]

Did you ever think, why frequency is changing when the vehicle is passing?? Because cos α is changing - while the velocity is constant! This is a very important fact!

[@AndreasSpiess]

Of course. Usually you can only use a part of the data where the angle is small

[@HaasLukas]

bin schon sehr gespannt auf das nächste video :)

[@AndreasSpiess]

:-)

[@omassi]

Nice i feel a step closer to build a golf clubhead speed reader.

[@AndreasSpiess]

Should be possible. Lots of metal, if I remember right ;-)

[@omassi]

Andreas Spiess yess, I have to try with a bluepill and the 10ghz radar... But looks to hard for me 😂 and I dont have the oscilloscope.... But sure your videos Will help

[@tinker7722]

Very interesting! I would like to know more about that counting statistics with LoRaWAN.

[@AndreasSpiess]

That should be the easiest part...

[@bardenegri21]

I had no idea you could do that with those sensors. Maybe you can use the stm32 'blue pill' type boards in your device since its super fast. GreatScott did a video on it this week.

[@AndreasSpiess]

I used this board last year. But this year I try to stick with the ESP32

[@FedericoMorenoCh]

Very interesting video Andreas! Do you have any update on this proyect?

[@AndreasSpiess]

No. The range of these small devices was too short. And the bigger ones still are quite expensive.

[@luom23]

considering detecting multiple objects, I would suggest multiple receivers

[@AndreasSpiess]

I am not sure if this would work or if they would disturb each other

[@thanksbuyershoppingmall6747]

Thank you so much for making such informative video. I want to make long range proximity inductive sensor please let me know
Whether this sensors could be use only to detect metals from a longer distance ?
Thank you

[@AndreasSpiess]

10-15 meters max.

[@johannienel1]

Awesome video!

[@AndreasSpiess]

Thank you!