Doppler Radar Speed Measurement

Message

ThomasB · #1 Post by **ThomasB** » 14 Jun 2025, 23:43

I would like to log the speed of the cars that go past my house. It's not a busy road, but the cars often go too fast. My property line is next to the road, so I have good visibility of the cars.

I have some basic requirements. The completed project has to cost under $100 USD. Everything has to fit in a small battery powered project box that is hard to spot at the side of the road, which means I don't want a radar horn on it. And it would be ideal if there was Arduino library code for the sensor to simplify development.

Yes, there are many microwave based people presence detectors on AliExpress and cool handheld baseball radar guns can be found on Amazon. But I can't find anything that would be suitable for my application where I need at least 20 meters range and forward/reverse velocity up to 85 KPH. And at a low cost.

The $50 USD K-LD2 and KLD-7 sensors from RFbeam Microwave GmbH look like possible candidates. But I can't find any existing projects to use as inspiration. So I'm not eager to go with one of them.

In case one of you has traveled in my shoes, do you have any recommendations for a DiY roadside speed detector that works well? Keep in mind that it should work with an ESP32 because maybe it will become an ESPEasy plugin.

- Thomas

TD-er · #2 Post by **TD-er** » 15 Jun 2025, 00:50

Hmm a quick search led me to "HB100", which is also quite cheap. (1,63 euro: https://www.aliexpress.com/item/1005008944070193.html )
https://theorycircuit.com/arduino-proje ... g-arduino/
https://theorycircuit.com/wp-content/up ... Sensor.pdf
https://theorycircuit.com/wp-content/up ... asheet.pdf

Not sure about licensing needed to radiate upto 20 dBm at 10.525 GHz though.

I took a quick look at the datasheet and the Arduino code and it looks like the sensor might output a frequency proportional to the velocity.
May also need some opamp to amplify the signal, but if you look at page 5 of the datasheet, it seems like they placed the signal at the + of the opamp to make it as digital as possible

Only thing I don't know is whether the "1m" resistors are 1 milli Ohm or 1 MOhm.
Well since they write the 'K' with a capital-k I think they have no clue about Si-units

(the 'u' and 'n' are in lower case....)

Anyway, I think you might already be able to use the pulse-counter plugin (P003) to get some test setup running.
Ali Express sellers claim upto 10m range.

Edit:
More interesting read:
https://github.com/dlpoole/HB100-Dopple ... easurement
https://github.com/3zuli/HB100_test
https://github.com/xxJian/HB100_Microwave_Demo <--- KiCAD file for opamp circuit
https://github.com/celikslab/ArduinoHB100Radar <--- Hand-held setup in I think Turkish
https://github.com/GPS1103/Doppler-Radar/tree/master <--- Arduino project to run on Atmega328p (Which could be used to interface to ESPEasy via serial)
https://github.com/dlpoole/HB100-Dopple ... /README.md

It seems to be hard to use it directional, so maybe you should also make some horn-like design using some conducting material like a Pringles can?

ThomasB · #3 Post by **ThomasB** » 15 Jun 2025, 04:52

Thanks for the info. The HB100 does not appear to be the droid I am looking for. It seems to be ideal for person detection in a room. Range is "up to 10 meters" which is another way of saying it will be less than that in real life.

BTW, I have an HB100 in a drawer. I might prototype it if something more interesting doesn't come along.

- Thomas

Ath · #4 Post by **Ath** » 15 Jun 2025, 16:38

Seems like the HiLink LD2451 24GHz Radar detector is capable of meeting your specs: https://manuals.plus/hi-link/hlk-ld2451 ... ule-manual
It has a detection range of up to 100m, and configurable speed-trigger of 0-120 km/H
There's an Android app and an iOs app available for testing (probably connected via Bluetooth)

Available at Aliexpress in the 12-15$ price range.

The technical working looks a lot like the LD2410 supported by ESPEasy, but most likely there are 'irreconcilable differences', not allowing it to work with that plugin without adjustments, just like other models from that LD24xx family

.

ThomasB · #5 Post by **ThomasB** » 15 Jun 2025, 18:06

Thanks! The HLK-LD2451 looks like a winner. Will order one.

- Thomas

Ath · #6 Post by **Ath** » 17 Jun 2025, 22:59

Did a little more investigation, and found the data specs for the unit, at the Hi-Link site, using their Google Drive share: https://drive.google.com/drive/folders/ ... ErEnPA2w4d

Also found this forum thread, elsewhere, that's less enthusiastic about this unit though: https://www.eevblog.com/forum/projects/ ... d-with-it/

ThomasB · #7 Post by **ThomasB** » 18 Jun 2025, 01:32

Seems that the more expensive HLK-LD2415H would have been a better choice. But the HLK-LD2451 I ordered has shipped, so I will focus on it.

I installed the HLK-Radar-Tool app on my smartphone. The app launches, but since I don't have a LD2451 device I can't get far. The interface I see is in English, which is a refreshing surprise.

- Thomas

TD-er · #8 Post by **TD-er** » 18 Jun 2025, 07:49

Maybe you can install an app like Duolingo to get your daily dose of foreign languages, to not get too used to this "English" you're talking about.

ThomasB · #9 Post by **ThomasB** » 18 Jun 2025, 23:37

Given what I see happening in my country, I probably should learn Chinese or Russian.

TD-er · #10 Post by **TD-er** » 19 Jun 2025, 00:02

Well there are more options.
Also with different read/write directions.

Anyway, it is absolutely horrific what is happening over there and it is truly unimaginable what they seem to get away with and still seem to get support.
Everyone with very basic history education should be able to see where this is heading to.

So let's focus here on getting people educated with basic fysics, electronics, programming etc. to allow them to break free from vendor lock-in for products which are way too easy to simply make it yourself and will often be even better.
If we can achieve that, I believe the rest of 'common sense' will become 'common' again some day.
I really miss that 'common sense' quite a lot lately when looking at the news. (not only in the US)

ThomasB · #11 Post by **ThomasB** » 27 Jun 2025, 00:10

I received the HLK-LD2451 Doppler radar detector and performed some testing. So far it appears to be a winner. Here's a summary of the test setup.

The radar board was installed in a 3D printed chassis that can be tripod mounted. Power is by a single 18650 cell with a 5V boost regulator. See photo.

TripodMount1.jpg

No need for a ESP32 at this point. The HLK-LD2451 has BlueTooth and there's a developer app that does all the heavy lifting. Google solved the language issues since I don't understand Chinese. The app allows changes to some useful device parameters.

App_Screenshot3.jpg

I setup the tripod on a wall in my front yard. The radar antenna is pointed towards incoming cars.

StreetView2.jpg

If you look closely at the photo, you can see a yellow X at an intersection up the street. This is 65 meters away and the radar detector reliably senses vehicles turning at that location. Impressive!

My testing was performed this morning and the road was very quiet. So only a handful of vehicles were monitored. It did a good job of reporting their presence but I have not had time to validate the reported KPH speed.

The app is very basic. It has a radar sector graphic and the vehicle's distance and relative position are shown on it. Text data is shown that has distance, angle, and speed.

I need to do more testing before I commit to integrating it with an ESP32. The module has a 3.3V logic serial output (115.2K baud) and all data is sent as binary. VSCode's terminal has a hex display mode and I tested it with a USB serial adapter. Got valid data, nice.

- Thomas

TD-er · #12 Post by **TD-er** » 27 Jun 2025, 00:35

Great!

Here some screenshots of the app...
https://www.laskakit.cz/user/related_fi ... l_v1-0.pdf

And some code links: https://forum.arduino.cc/t/using-hlk-ld ... nt/1328105

And very descriptive protocol specification: https://h.hlktech.com/download/HLK-LD24 ... 0V1.03.pdf

Translated:

LD2451 串口通信协议 V1.03 (1).zip

ThomasB · #13 Post by **ThomasB** » 27 Jun 2025, 03:41

Thanks for posting all the links. I already had the info but no doubt other folks might want to learn more about this interesting device.

Here some screenshots of the app...

The screenshots shown there are from an earlier release. The latest Android release (V1.6.0) looks much different. The radar display has been simplified.

- Thomas

ThomasB · #14 Post by **ThomasB** » 27 Jun 2025, 18:51

The technical working looks a lot like the LD2410 supported by ESPEasy, but most likely there are 'irreconcilable differences', not allowing it to work with that plugin without adjustments, just like other models from that LD24xx family

Thanks for the tip. I looked at the LD2410 to see if the command interface is similar to the LD2451. The good news is that they use the same command structure. So ESPEasy's LD2410 (P159) plugin looks like a good starting point.

- Thomas

ThomasB · #15 Post by **ThomasB** » 29 Jun 2025, 02:10

The roadside display will be a 45cm x 61cm (18 x 24 inch) aluminum sign. It will be produced by an online printing business that I have used in the past.

I have some 1cm wide white colored strip LED's that can be cut to form the two digit 7-segment MPH readout. They operate on 12V and have extreme brightness. Easily seen in direct sunlight.

To minimize the ESP32-C3 pin count usage, LED control will utilize WS2811 LED driver chips. One can control three 12VDC LEDs, so a total of 14 chips will be needed (or only 9 if I gang segments together). It's nice that only one output pin is needed and ESPEasy supports the WS2811.

The SLOW DOWN message will be high brightness 5mm LEDs. One ESP32 pin will be used to flash this message when a speeder is detected. I need to check AliExpress to find a suitable LED for this feature. If anyone has a suggestion for a super bright red or white 5mm LED then please chime in.

I still need to do some more radar module testing. If all goes well I'll order the sign and start building. Not in a hurry, so the finish line is a long ways off.

- Thomas

Ath · #16 Post by **Ath** » 29 Jun 2025, 10:28

You have found P165 I assume?

ThomasB · #17 Post by **ThomasB** » 29 Jun 2025, 18:30

You have found P165 I assume?

Very nice. Did not know about it. Thanks!

Unfortunately, I wanted to use my cheat method of having each WS2811 control three different LED's to reduce chip count. I can do that using rules.

But after seeing your 7-Seg plugin, I may change my design to use NEOPixels instead of my crazy-bright strip LEDs.

- Thomas

TD-er · #18 Post by **TD-er** » 29 Jun 2025, 20:04

Maybe also stick those LEDs on a piece of aluminium as they might heat up quite easily and probably already start at a high temperature when mounted on a plate which is heating up by the sun.

ThomasB · #19 Post by **ThomasB** » 29 Jun 2025, 20:23

Maybe also stick those LEDs on a piece of aluminium as they might heat up quite easily

Good point. They will only run for a few seconds at a time. Just enough time for the passing car to see their speed.

Not many cars go by, so most of the time the pixels will be off. So even on hot sunny days, I expect the LED temperatures will be somewhat safe using only the aluminum sign as a heatsink.

- Thomas

ThomasB · #20 Post by **ThomasB** » 29 Jun 2025, 23:56

Will the P165 plugin allow me to use SK6812 WWA LEDs? How about RGBWW?

RGBWW has both white and warm white. WWA is Warm White, White, and Amber instead of the traditional RGB colors.

- Thomas

TD-er · #21 Post by **TD-er** » 30 Jun 2025, 00:20

As long as the number of bits is either 24 or 32 bit per LED and the timing is roughly the same as with other addressable LEDs, then it should work. You may need to tune the colors a bit yourself though.

No idea if RGBWW actually needs 5x 8 bit. I would expect both "W" to act on the same byte.

From the top of my head, the clock frequency is about 800 kHz and you need to have some 80 usec 'reset time' at least for actually 'latching' the current pixel data to the LED.
Each high/low bit can be encoded in 4 time slots of roughly 0.3 usec. ('1000' for 'low' and '1100' for 'high')

You should take a proper look at the required data level of those as it is not always guaranteed they might work well on 3V3 signal level.
Typical issue then is that the first LED isn't working and any next LED is, though there might be some pixel errors.

Ath · #22 Post by **Ath** » 30 Jun 2025, 23:07

ThomasB wrote: 29 Jun 2025, 23:56 Will the P165 plugin allow me to use SK6812 WWA LEDs? How about RGBWW?

RGBWW has both white and warm white. WWA is Warm White, White, and Amber instead of the traditional RGB colors.

The SK6812 Leds should be compatible with the more common WS2812 types (AFAICS), so that should work. The plugin has support for GRB and GRBW type Leds (other types are quite rare), where the last setting should also match GRBWW as WW just means Warm White, the white component of the color adds a 4th parameter to the color settings, where normally 3 are required. (Not incorporated in the color defaults!)

ThomasB · #23 Post by **ThomasB** » 01 Jul 2025, 00:19

Thanks for the info. The SK6812 looked good at first glance. But after reviewing the total current to drive all the needed 5V pixels I went back to the drawing board.

I decided to use the WS2814 RGBCW 12V strips in the 84 LED/meter configuration. They are grouped with 3 RGBW chips per pixel, which works well with my 7-segment numeric layout. I picked Cool White (7000K) for daytime visibility. Each digit will be 6 pixels (18 LED) high x 3 pixels (9 LED) wide.

This minor design change means I'm back to filling an AliExpress shopping cart. I sure miss their lower prices from a few months ago. I used to toss in a lot of extra stuff that looked interesting. Now I just get the bare essentials.

- Thomas

ThomasB · #24 Post by **ThomasB** » 18 Jul 2025, 00:14

The WS2814 RGBW strip lights arrived a couple days ago. I prototyped one digit and it is working well with Ton's 7-Segment NeoPixel plugin. I need to duplicate it to create the two digits for the MPH display. The digit is big, about 23cm high.

7-seg_disp1.png

Here's a couple comments about some initial obstacles:

1. TD-er has previously warned about using low cost ESP32-C3 boards. I bought a $5 USD AliExpress device that had an external antenna. Was happy with the WiFi range and compact size. But it unexpectedly died. Could have been me that killed it, or maybe a component failure.

The replacement ESP32-C3 board has an integrated antenna and WiFi operation is horrible. WiFi recalibration does not help. As TD-ER mentioned, board layout near the antenna input is likely an issue.

So long story short, my experience seems to confirm that cheap ESP32-C3 boards are troublemakers. I will validate this further by ordering a different ESP32-C3 with external antenna and see if it works reliably.

2. The WS2814 RGBW strip light I got from Aliexpress was also a troublemaker. Unlike most other WS28xx based LEDs I've used in the past, this RGBW absolutely needed a signal buffer (I used a 74AHCT125N). Also, it did not tolerate the use of a signal line resistor on the output of the buffer; Anything beyond the built-in 100 ohms that is already installed on strip's flex PCB caused serious problems (missing pixels, wrong colors, etc).

These issues have been resolved. But thought I would share some recent information about "SPEEDY" the Radar Speed Meter project.

- Thomas

TD-er · #25 Post by **TD-er** » 18 Jul 2025, 09:45

Please have a look here for an antenna mod for those C3 boards.
https://peterneufeld.wordpress.com/2025 ... ification/

TD-er · #26 Post by **TD-er** » 18 Jul 2025, 09:50

Oh and if you have any ESP32 with dual core at hand, maybe also try using one of those to see if it will give less issues with the addressable LEDs.
With dual-core ESP32's the WiFi code is ran on the other core.

N.B. There still is an issue with ESP32-S3 with OPI PSRAM and addressable LEDs.
So when running on an ESP32-S3 with 16M flash/8M PSRAM, you may want to use the build without "OPI PSRAM" in the name.

This is being worked on, but it is still a bit unclear what may cause these issues. It seems like it may have to do with RMT-code trying to access a cache line which is temporary locked by PSRAM code.
Adding the flag to consider RMT code cache/lock free isn't the whole fix.

ThomasB · #27 Post by **ThomasB** » 18 Jul 2025, 17:39

Thanks for the antenna mod info. WiFi unexpectedly started working fine. This might be related to when I switched to Ton's 7-segment plugin (from the regular Neopixel plugin). Or maybe not.

I didn't realized the C3 was single core. That seems like a bad idea to me. And my project board is already wired for the C3 super mini.

- Thomas

TD-er · #28 Post by **TD-er** » 18 Jul 2025, 23:12

Don't buy the S3 versions of those extremely small boards.
The "cheap ones" are absolutely horrible.
I've got 2 here and I can't even get them to properly enter flash mode and if they do behave for about 1 minute to complete the flash, (at least it looks like that one does finish the flash procedure) it is impossible to get them to start doing WiFi stuff.

The S3 Seeed Studio units in that small form factor also seem to be a lottery... at least mine seem to work fine, but I've also heard from others they weren't that happy with those.

ThomasB · #29 Post by **ThomasB** » 19 Jul 2025, 01:02

Maybe we should create an ESP32 Hall of Shame to showcase all the garbage boards.

- Thomas

TD-er · #30 Post by **TD-er** » 19 Jul 2025, 13:20

Problem is, it seems to be batch related.

For example those purple Wemos-D1-mini sized ESP32-S2 boards, those with hardly any parts on it except USB connector and ESP.
Those are available in "Just perfect" and "Utter garbage" and you can't see the difference.
Probably some batch with wrong parts on the antenna matching circuit.
Or maybe something as basic as "rotated on the PCB substrate during production" or "different dielectric on PCB as the direction of the fibers in the PCB and/or layer stack of the PCB matter for the impedance matching of the antenna.
This means that if in the same batch some were rotated to fit more on the same PCB, you can end up with different results in the same batch.

Anyway, a good rule of thumb:
Chip antennas need space around them, if there isn't any, it will work extremely bad.

ThomasB · #31 Post by **ThomasB** » 20 Jul 2025, 04:48

Seems that getting a "good" ESP32 is a bit of a lottery.

I am going to try making a plugin for the HLK-LD2451 sensor. I plan on using the P159 as a template for it. What Pxxx number should I use for the new sensor?

- Thomas

Ath · #32 Post by **Ath** » 20 Jul 2025, 10:30

I have reserved P182 for this Presence - HLK-LD2451 Radar speed sensor

ThomasB · #33 Post by **ThomasB** » 20 Jul 2025, 18:14

Thank you!

File name will be _P182_LD2451.ino

ThomasB · #34 Post by **ThomasB** » 30 Jul 2025, 00:07

I finalized the 7-Segment design's plastic case. The strip LEDs are now assembled and running on the Neopixel 7-Segment plugin. Very cool, glad Ton created it.

I would prefer that numbers below 10 appear on the right side digit. Left digit can remain blank. Is there a way to do that?

Example, here is the number 3. It is left justified. I would like it on the right side.

Digit_03.png

And to show both digits are working correctly, here is the number 14.

Digit_14.png

The VEML7700 light sensor plugin varies the display's brightness, depending on sunlight. The contrast looks much better in person. Big, Bright, and easy to read. The camera does not do it justice.

Next I need to tackle creating the Radar Speed plugin (P182). That will no doubt involve another rabbit hole to get lost in. Maybe ChatGPT will help me out.

- Thomas

Ath · #35 Post by **Ath** » 07 Sep 2025, 21:51

ThomasB wrote: 30 Jul 2025, 00:07 I would prefer that numbers below 10 appear on the right side digit. Left digit can remain blank. Is there a way to do that?

Example, here is the number 3. It is left justified. I would like it on the right side.

This thread got a bit out of sight, but it has a question I didn't follow up yet, sorry for that.

I assume you write the numeric value on the display using the '7dtext' command, then using the regular variable formatting should suffice to have that aligned correctly:

Code: Select all

  let,number,3 // Or some calculation
  7dtext,[var#number#d2] // Format number in 2 positions, no decimals, leading spaces if needed

ThomasB · #36 Post by **ThomasB** » 09 Sep 2025, 23:46

Hi Ton, I tried it and single digits (less than 10) are still left justified.

The MPH value is held by %v1%, so here is the rule command I used.

Code: Select all

7dtext,[var#1#d2]

EDIT: I found the solution. Just a minor tweak was needed. This works:

Code: Select all

7dtext,"[var#1#d2]"

I've finished building the speed sign. A lot of 3D printed parts and a sheet of hardboard. The end result looks convincing.

The blinking "SLOW DOWN" message LEDs have been failing. My fault, I'm driving them at 25-30mA. And they are probably Chinese factory seconds. So I ordered some new ones that will be driven at 20mA. Hand-wiring 150 of them took me all day, now I get to do it again.

MPH_Sign1.jpg

I haven't started the LD2451 Radar plugin. I need to get motivated. But it looks like it will be a challenge for me and I hate pulling out my thinning hair.

- Thomas

Ath · #37 Post by **Ath** » 10 Sep 2025, 08:20

Trying to post a reply but the board won't let me quote stuff

Ah, I should have added those quotes, of course, my bad

We'll see, if lacking the proper motivation, I can pick this up

My LD2451 arrived a couple of weeks ago, and plenty of speeding cars in my street, so the test-bed is available

TD-er · #38 Post by **TD-er** » 10 Sep 2025, 09:12

There are also those elongated larger LEDs which might even cover a full segment of the 7-segments display.
That could save quite a lot of soldering and hair pulling.
Only problem now is to find ones that are a perfect match for what already has been printed.

Those also have an aluminum backplate which might make cooling a breeze

ThomasB · #39 Post by **ThomasB** » 10 Sep 2025, 18:42

My LD2451 arrived a couple of weeks ago, and plenty of speeding cars in my street, so the test-bed is available.

That is fantastic to hear.

There are also those elongated larger LEDs which might even cover a full segment of the 7-segments display.

The "SLOW DOWN" message is just 5mm red LED's controlled by PWM. Unfortunately red tends to be lower intensity than other LED colors. I over-compensated by driving them with a lot of current for viewing in bright sunlight. Bad idea.

The 7-Seg display uses RGBW WS2814 strips. They produce a very bright white (all four LED chips on) for high visibility daytime viewing.

At night the white LED chip is turned off and brightness is reduced. Instead of a RGB mixed white color, a comfortable low intensity yellow color is used (less glare on the eyes).

Ambient light is sensed by a VEML7700; The 7-Seg MPH and SLOW DOWN message intensity is variable (extra bright during the daytime).

- Thomas

Who is online