RTK GNSS module build


I just realized this is the very first blog I ever wrote :) 

RTK GNSS is an interesting topic for many quadrotor diyers like myself. It can provide stable sub meter coordinates which is very attractive for autonomous quadrotor flights. Initially, I tried RTKLIB on windows platform, and the result is pretty encouraging. However, I cannot imagine to have a windows running computer on a quad. Intel edison is another option for me: reasonably priced, small enough, and powerful enough to run real time RTK algos. After having this idea for a few month, I noticed REACH is commercialized RTK GPS using a very much close methodology. My intention is not to commercialize my design so REACH's results confirmed Intel Edison is suitable for RTK algos.

In the second version design, I would like to have Linux take over the control. which means linux can get the data for running RTK algo without the help of of a MCU. In my design, I put a four port USB hub on board along with a few USB-UART bridges. These USB-UART converters will get rover data from GPS modules, and reference data from either MCU or 433Mhz radio. On top of that, the USB hub also connected to a 3G modules. in this way, the RTK module could get online to take advantage from real time networks. 

3689706126?profile=original 3689706074?profile=original

E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones


  • Hi YC, 

    Great job you did there congratulations. 

    Im more interested to use PPK over RTK for my UAVs. How did you built the RTK module?

  • HI YC

    this is an interesting test. 

    However I don't think that you prove that you have the right absolute position. You are "only" better in relative positionning.

    I personnaly test the M8N/RTKLIB in RTK on geodesic points and I hardly find a fix position ( over 6min at least ) and the result was not so good (about 30cm of error, even sometime more. ).

    However if I stay for about 30min and use a post treatment solution via RTKLIB ( with a personal base nearby or Ntrip around) I achieve very often, less than 3cm of precision in XYZ.

    For me , M8N is then good enough for post treatment or for better relative positionning in Single GNSS.

  • Hi Sabido,

    Yes, my M8N is using old firmware. 

    Be honest, galileo does not have much coverage so not having it is acceptable to me, beidou is the same case. 

    RTKLIB and M8N combination works fine to me. I am not looking for centimeter or millimeter accuracy from this hobby like setup. Tens of centimeter is perfectly fine to me.  

    The following is the realtime RTK solution and post GPS only solution. I was driving circles with minimal turn radius of ~5m. 


  • Hi Aleksander,

    this is the information I used to patch the GPS module.


    My M8N is fairly old, I bought it last year and only GPS/Glonass. Galileo and Beidou are not available.

  • Hi YC.

    You can still use M8N for raw ouput and use it in RTKLIB 2.4.3 . However , you need to stay in FW2.01 (GPS+ Glonass)  . The new FW3.01 (with galileo) doesn't work with RTKLIB.

    M8T is much simpler to configure.

    Be aware that the RTK option with RTKLIB and M8N (M8T and all other L1 GNSS also,I presume)  is rather poor ( difficult to Fix ) and don't expect the perfect solution.

  • Hi!

    Does the "patch" work with the latest firmware? Can you provide some info on the procedure?

  • Correct, but M8T is way too expensive to me.

    There is a way to "patch" M8N to have raw output. Just additional a few line of code would do the trick.

  • Hi

    I see ublox-m8N. I thought that this version does not have raw gps data that one need for rtklib calculations.

    Reach, uses M8t. Am i right?

This reply was deleted.