An RTL-SDR Based Ground Penetrating Radar & Metal Detector

Thanks to Dr. Celalettin Uçar from Turkey for submitting a video of the work done by a PhD student who as part of his research created an RTL-SDR based ground penetrating radar simulation and metal detector. He writes:

This apparatus (YAĞRIN) was created with rtlsdr in a phd work. We achieved detecting a metal gasoline tube from the depth of aproximately 1 meters. Furthermore, we created the time domain signal and ploted the reflaction from the metal with using the matlab (simulink) model.

A video on YouTube is linked which we display at the end of the post. They write that the system consists of a 12V DC supply, step down voltage regulator, ADF 4350 programmable signal generator, 25W power amplifier (470 MHz, 45 dBm signal power), Philips omnidirectional antennas (RX,TX), a 64 dB low noise amplifer and an RTL-SDR and computer to display the output. The software he uses is SDR# which appears to simply listen for a tone and detect any changes that occur when something metal moves near it. The PC also runs a MATLAB Simulink model which we believe helps detect metal signatures by plotting the reflection.

In the past we posted about a similar but simpler metal detector implementation by Ancient Discoveries.


Notify of

Inline Feedbacks
View all comments

detect from 1 meter depth?
but how? 2.8msps sampling get very low resolution for receiving reflect from 1meter depth..


If they are transmitting at CW at a fixed power level then the signal strength reflected would give an estimate of depth. Which would vary depending on the material in the ground and the moisture content, basically anything that else that would change the attenuation.

Celalettin UÇAR

Hi Alex, sampling rate is low for gpr but about metal detecting our signal is too strong. 44 dbm is enough to radiate EM to 35 kilometer on air. Sure that it can be penetrate a several meter in ground. And 48 db SNR is enough to distinguish the metalic or non metalic field for 1 meter depth.


In the end, this device can work as a simple metal detector, it is impossible to determine the exact distance of different metal objects, because depending on the size of the object, there will be a different level of the return signal regardless of the depth.

so this device can not be called a GPR; the principle of operation and capabilities are more like a metal detector.
or I misunderstand

Celalettin UÇAR

Alex if you read the passage carefully… we described it “RTL-SDR based ground penetrating radar simulation” . It is only a gpr simulation for deep learning for ground penetrating radar that how to work. Ofcourse rtl sdr is not enough to create a real GPR. But if you use a per vices cirmison model upper version sdr ! It has a 2 gsps sampling rate and 16 bit, you will create a real gpr… This device show that the sdr capability about radar. Finally, we plotted the reflection of metal gasoline tube on matlab. But of course this was not a real radargram. Thank you for your contribution


I understand you, thank you for clarifying

Belayet H.

Understood the fact over your comment-conversations. Now it is more clear about principle of operation and capabilities are more like a Metal Detector. Thanks so much for everything.

Celalettin Uçar

A deep learning for gpr but ofcourse it is not real. Rtl sdr sampling rate capability is not enough to create a real gpr. Its only 2.8 mbps 🙂 a real gpr least sampling rate requirement is approximately more than 300 times. so i call it as a gpr simulation… on the other hand, if you use an enhanced sdr, it is possible with same method. Thank you