Open Ground Penetrating Radar

Make the invisible visible for about $500.

Similar projects worth following
Commercial ground penetrating radar systems cost tens of thousands of dollars. This project aims to create an open hardware alternative for about $500.

This project has three components: the Open Ground Penetrating Radar (oGPR) system, the rover and software to process the data and plot in on OpenStreetMap. This Hackaday project focuses on the first two.

The oGPR is designed to locate underground "voids," such as abandoned mines, were hazards may exist.

Before these hazards can be mitigated, they must be located.

Project Home Page:

Rover Wiki:

A Baofeng VHF/UHF radio is used a signal generator, which transmits via a Transmit/Recieve switch through an antenna. A RF detector is also connected to the T/R switch, which is connected to the MSO-19.

The radio controller is a 4N25 opto-isolator.

The T/R switch controller is a SN74HC04N (or similar) hex inverter.

Any Pi-compatible GPS and WiFi dongle will do.

The Raspberry Pi triggers the radio, T/R switch and MSO-19 via a buffered GPIO port once per second using a very simple python script.

In the United States, this project falls under FCC Part 97 Sec. 1 (b):
"Continuation and extension of the amateur's proven ability to contribute to the advancement of the radio art."

  • 1 × Link Instruments MSO-19 ($249) 2 GSa/s Mixed Signal Oscilloscope
  • 1 × Baofeng VHF/UHF HT Radio (Amazon, $30) Signal Generator
  • 1 × HMC784MS8GE (DigiKey, $140 eval board) Anetenna T/R Switch
  • 1 × AD8314ARMZ-ND (Digikey, $10) RF Detector
  • 1 × Raspberry Pi (Amazon, $30) Controller, Data Collector, Web Interface

View all 11 components

  • Intro Video

    Glenn Powers08/25/2015 at 07:36 0 comments

    Made with Ardour and OpenShot.

  • $80 Lenovo Tab 2's are great!

    Glenn Powers08/17/2015 at 04:21 0 comments

    I have a Lenovo A1000 tablet that I got about a year ago for ~$80. It works fine for ebooks and 480p/720p videos, but is really too slow for web browsing. The Tab 2's are a great buy for the money. Given the recent coverage of "The realities of a $50 smartphone," the reality of a $80 tablet is that the display isn't nearly as good as a Samsung Tab 4, but it's about half the price and runs Tower and FireFox just fine.

  • My New Robotic Friends

    Glenn Powers08/13/2015 at 01:27 0 comments

    Rovers are a lot like PHP libraries. Developers start by scratching a personal itch. But, in the end, they are all broken in their own unique way. So, I've decided to use rover platforms from SuperDroid Robots, who employs a dozen professional engineers and has been building robots for 25 years. I'll focus my efforts on putting my new friends to work.

  • They're Here!

    Glenn Powers08/13/2015 at 00:30 0 comments

  • BUD Industries Makes Great Boxes

    Glenn Powers08/11/2015 at 16:14 0 comments

    Or, "enclosures." Good design, well made, inexpensive.

    Just search Amazon for "BUD Industries Aluminum Econobox"

  • SV1AFN's $30 RF Switch

    Glenn Powers08/11/2015 at 02:50 0 comments

    HMC784 - 10W RF SPDT Switch, LF to 4000 MHZ (4 GHZ), 30 dB Isolation, 15-40 nSec switching times, Very Low DC current - Good IMD3 (+60 dBm)

    This uses the same chip as the eval board from DigiKey for $138.75 and includes a hex inverter for single wire control!

    Plus, the DC blocking capacitors are unpopulated. So, you can select the best values for the frequency you are using. Here is a brief discussion on picking the best values for the caps.

  • Stopping Pollution Before It Starts

    Glenn Powers08/06/2015 at 04:41 0 comments

    An anonymous visitor to my table at the Detroit Maker Faire mentioned that he worked for an undisclosed company that was using proprietary methods for locating valves in underground oil pipelines. These valves were then removed before they sprung a leak.

    He said the company considered using ground penetrating radar, but decided against it because of the cost ("$50,000 to $100,000 per system").

    I wonder how many potential problems could be located and mitigated with an unmanned ground penetrating radar system that costs 20% of current manned systems?

    I'll probably need to increase the frequency to be able to see the target signature of valves in a 3" pipe at "backhoe depth." But, it's on my growing list of targets.

    First, I'm going after the elephant in the room.

  • NAVIO+ is neat, but expensive

    Glenn Powers08/05/2015 at 15:55 0 comments

    "Navio+ is an improved version of Navio autopilot shield for Raspberry Pi A+/B+. It was designed both for your own custom robotic projects and as a platform for Linux version of APM (ArduPilot). Navio+ eliminates the need in multiple controllers onboard making development easier and increasing robustness."

    But, at $168 the NAVIO+, $12 for the GPS antenna and $24 for the power module, it's three times the cost of a APM 2.6.

  • mechanical speed reduction

    Glenn Powers07/31/2015 at 23:50 0 comments

    "choosing the best [mechanical] reduction ratio is the single most important factor in determining how well the motor will work in a particular application."

    "Do not rely on electronic speed reduction to do the job of a mechanical speed reducer."

    AmpFlow whitepaper: mechanical speed reduction.doc

  • Choosing a Vendor, Letting Go

    Glenn Powers07/30/2015 at 18:18 0 comments

    Jim Summers recommended two battery suppliers: EM3EV and ElectricRider.

    I discounted EM3EV because they don't have exactly what I need. They have a "Request a quote" option, but are located in Hong Kong and don't list a phone number on their web site. When developing a product, it's often important to be able to talk with someone on the phone.

    ElectricRider has what I need. But, when I called the good people down in Texas and asked about a solar charge controller, they said their controller works on 120 volts and if my "solar panel puts out 120 volts, it'll work fine." *facepalm*

    I consulted the great AI known as Google and found Bioenno Power whose "products are trusted and employed internationally and domestically by a wide customer base ranging from amateur model makers using our batteries for RC applications to professional aviators trusting our batteries to power critical avionics and Ham Radio operators." And, they sell solar charge controllers for their LiFePO4 batteries. Wow! Sounds like a perfect fit.

    So, I called them. They were very pleasant and helpful, but insisted that 24 volt batteries need 24 volts of solar power. Inexpensive, small form factor, 100 watt (C/5-ish for 24 volt, 20 amp-hour batteries) pretty much don't exist.

    Not to be deterred in my quest, I consulted Google a few more times and found a vendor in California that sells a device that does what I want. But, it's manual is in poorly translated Chinese, which doesn't inspire confidence.

    So, I've decided to let go of my desire for a solar panel on my rover. This will allow for a smaller, lighter rover that isn't top heavy. I can charge the batteries from a pair of ground mounted 12 volt panels. I could use a DC/DC converter to step up the 12 volt solar panel to 24 volts for the charge controller. But, this is going past the design goal of "keeping it simple."

    Moral of the story: follow your heart. It may not lead you to what you see in your dreams, but it may lead you to a better vendor of LiFePO4 batteries.

View all 32 project logs

  • 1
    Step 1

    Here is the schematic for the power and trigger control board. You can view it in Scheme-It. Q1 is a N-Channel MOSFET to protect against an inverted polarity power supply. I used a LM7806 and a LM7808 instead of a single LM7505 to give a faster switching time. U1..U4 is a 7 channel hex inverter. J6 is a BNC connector. J3 should be a female header, others are male.

View all instructions

Enjoy this project?



Zygfryd wrote 08/19/2023 at 15:40 point

Nice scam with "Project Home Page" which is fake Microsoft web site

  Are you sure? yes | no

eakar74 wrote 12/17/2022 at 12:43 point

hello, I need the files of this gpr project for my school graduation project. Could you please send me the files of this project. Thank you very much to those who are interested.

  Are you sure? yes | no

BlueFox wrote 09/18/2019 at 05:51 point


Our team trying to make Multi Array GPR, Any one can send the project documentation?

  Are you sure? yes | no

John wrote 02/06/2019 at 15:39 point

When I first saw this project I was really excited but unfortunately it looks like the guy gave up on it since his last post was in late 2015.  If anyone knows of a GPR project that’s in the works or has been finished, please post a link.

  Are you sure? yes | no

osmancakal wrote 01/30/2019 at 11:19 point

I can not open the site. If anyone can email the project documentation to me.

  Are you sure? yes | no

bizadi wrote 08/30/2017 at 13:18 point

Cannot connect to Project Home Page:

  Are you sure? yes | no

bizadi wrote 08/30/2017 at 13:10 point

I really need this project to research. Thank you

  Are you sure? yes | no

bizadi wrote 08/30/2017 at 13:08 point

I can not open the site. If anyone can email the project documentation to me.

  Are you sure? yes | no

marcpechaud wrote 07/26/2017 at 16:08 point

Don't see feature on efficiency of this robot ? How many meters is the ground penetrating of it?

  Are you sure? yes | no

noega33 wrote 03/28/2017 at 16:43 point

Hi Glenn,

Cannot connect to Project Home Page:



  Are you sure? yes | no

muhammaduzairjawaid wrote 10/02/2016 at 16:15 point

Really nice project. helped us a lot in designing something similar, but we are facing few problems, 

1) we are using 2 different antennas one for receiver and one for transmitter but problem that we are facing is our receiver antenna is receiving signal directly from transmitter. how should i stop that, i made a faraday cage with a metal sheets with grounding that cage but still our receiver is receiving directly from tx. 

2) how to calculate travel time? we are transmitting frequency of 433 MHz and 40 Mhz.

3) our output power currently is 25 dBm for both frequencies, how much more amplification we need for penetration of 10m .


  Are you sure? yes | no

friedtj wrote 11/27/2016 at 09:57 point

10 km with GPR is just not possible at 434 MHz: you might consider seismic for that. Quick maths: free space propagation loss is 2*(20log(d)+20log(f)-147.55) in SI units, ie f=434e6 Hz and d in m: 10 km would be 210 dB loss. Best GPR will exhibit 100-110 dB dynamic range, and here I am not considering conductivity losses. Radar range is not limited by Tx power but  by isolation between Tx and Rx (the radar equation gives Rx/Tx power). 10 km has been demonstrated in antartica with HF GPR (5-10 MHz) in dry ice, ie no conductivity losses. Faraday cage will create ringing, which you do not want in a broadband system => replace the reflective faraday cage with absorbing foam.

  Are you sure? yes | no

sebiblaurq wrote 02/08/2018 at 16:08 point

hi Muhammad could you please send me your schematics?  Please. Im am doing a GPR project. 


  Are you sure? yes | no

friedtj wrote 06/19/2016 at 14:16 point

I was searching on the web if there is any GPR-related project and found this page. There are a few surprising things though

1/ why are you using an air coupled yagi antenna rather than the usual ground coupled dipole (or bowtie) antenna ?

2/ how can you switch the RF source quickly enough to provide a single period (your switch will not be phase synchronized with the radiofrequency oscillator) ? GPR requires a single (+/- ringing) period, not a long pulse train. An avalanche transistor unloading with a time constant inverse of the frequency at which the reflections coefficient of the antenna is minimum is the usual approach (allows to power a couple hundred volts, or peak power of a few kW). Or are you frequency-sweeping the source for an FSCW approach ?

3/ related to the other requests about depth, there is no single answer: the link budget is dependent on soil conductivity and operating frequency, from a few hundred meters in arctic regions (dry ice is transparent to electromagnetic waves) at 100 MHz to a few cm in concrete when operating in the GHz range. Typical penetration depths in temperature regions and 100 MHz is in the meter range.

  Are you sure? yes | no

Ronald wrote 06/10/2016 at 22:25 point

If you have a working GPR, it would be great if you could test for ability to detect/locate sea turtle egg clutches.

  Are you sure? yes | no

Delfim Machado wrote 06/07/2016 at 13:14 point

Hi guys, if i put this in a car, what is the max speed for good data?

  Are you sure? yes | no

Paul wrote 06/12/2016 at 07:07 point
In the post it's aid they were doing 1pulse/reading per second, which is very slow

  Are you sure? yes | no

kayaa wrote 06/06/2016 at 23:11 point

rassbery pi program ?

  Are you sure? yes | no

Gage Kitchen wrote 05/27/2016 at 19:16 point

Sorry if I'm putting too many questions into my post here but I'm working on this myself at home based on your components list. I was wondering if you can post any images of the internals of the GPR and how you attached them to your power source. Additionally did you end up developing your own software to get each device to communicate properly with each other and your display or did you use an existing software out there like MALA. And if so where did you download this software. Overall though I am a huge fan of this design and you've really got something great here.



  Are you sure? yes | no

Jim Patek wrote 02/25/2016 at 19:15 point

Did you use the C# DLLs available for the MSO-19 to develop your programs?

  Are you sure? yes | no

bobregalia wrote 12/12/2015 at 20:19 point

as an avid metal detector, I was thinking about something like this or  an electromagnet scanner. I have a few 7.9Gz transceivers and was researching how to use them to scan into the ground as a GPR. I'm going to keep an eye on this one...

  Are you sure? yes | no

geoperdito wrote 11/16/2015 at 12:03 point

Will the Baofeng really work as a signal generator?

I've read that in order to achieve a decent penetration depth you will need a very high power pulse(about 100v) which the Baofeng cannot generate.Correct me if i'm wrong.

Also,would the picoscope 2200a work instead of the mso-19?It's about 100$ cheaper.

  Are you sure? yes | no

Paul wrote 06/12/2016 at 07:05 point
I think they are also using the MSO-19 as the pulse generator, but there would be quite a lag between the the PTT and the actual transmit, that is is course in comparison to the speed of everything else.

  Are you sure? yes | no

abdulhadi_ics wrote 11/10/2015 at 08:00 point

Nice project kindly provide circuit diagram.

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates