1"x1" Centimeter-Level Accuracy RTK Module

updated September 5, 2015

25mm x 25mm S2525F8-BD-RTK single-frequency cm-level accuracy RTK module is nearly ready.

Below is a test board that allows receiving base-station RTCM 3.0 data from Bluetooth 4.0 over smartphone's 3G/4G network running ntrip, the rover module receives global positioning satellite signal, performs on-board RTK processing to generated cm-level accuracy position data by itself, and output the result over Bluetooth 4.0 to other applications. 

Alternatively another carrier phase raw measurement module can be used in place of the base station for high-accuracy relatively positioning applications such as precision farming, grass cutting, or UAV multi-rotor precise landing.

A new evaluation board is being prepared with interface options of Bluetooth 4.0 or USB 2.0 or pin-header for both input and output. The UART pin-headers can be used to connect to other wireless interface modules when needed.

Dynamic Testing

Antenna: Harxon HX-CSX601A
Baseline: 1.7Km
Max Speed: 81.2Km/hr (due to speed limit of the road, not capability of the receiver)
Blue point: single solution
Yellow point: float solution
Green point: fixed solution

The dynamic test on the road, being able to maintain fixed solution at speed over 80Km/hr, shows promising result for precision navigation guidance beyond low speed agricultural and data collection applications.

We are setting up fixed IP for our base-station, to allow more robust RTCM 3.0 transmission over the Internet for longer baseline faster speed dynamic testing.

Aug 27, 2015

Below is test result we got earlier showing performance of our single frequency RTK receiver in terms of baseline distance. 

= = = = = = = = = = = = = = = = = = = =

The rain is pouring heavily today. Not knowing how single frequency RTK would perform under such condition, our engineer went out and did some testing.

Rover Antenna: HX-CSX601A
Test Environment: Road with open sky view

Static Testing with 4.5Km Baseline: It took slightly more than 4 minutes to get fixed solution; which is much longer time than earlier results on a sunny day.  

Dynamic Testing with 4.5Km Baseline: On section A the car drove on adjacent 3 lanes separated roughly 2 meters apart, on section B the car drove on the same lane on each pass. The result shows mostly float solution with small number of fixed solution. What's remarkable is that the 3 tracks on section A are distinctively on separate lanes roughly 2 meters apart running in parallel, and the 3 tracks on section B roughly overlap; a result similar to getting all fixed solution while actually it's float solution most of the time. It's operation under such non-ideal conditions that typical users will encounter and pushing to get good performance out of single-frequency RTK that we have been spending time optimizing.

Sep 5, 2015

We received a compact active helix antenna sample for high precision application. Incredibly when mounted on roof top for preliminary testing, it does receive signal very well and can get RTK fixed solution. This antenna + S2525F8-BD-RTK module + Bluetooth + battery configuration, receiving RTK correction data via NTRIP over smartphone's Internet connection and make the NMEA output position data available to other smartphone App, might be able to serve as a portable RTK receiver for anyone seeking 100X better position accuracy than what their smartphone offers (from meters to centimeters), or for data collection usage as mentioned in this interesting "democratization accuracy" article by Brent Jones.

More testing will be done later with this interesting antenna. Below is how it might fit with SMA connector device:

Email me when people comment –

You need to be a member of NavSpark Community to add comments!

Join NavSpark Community


  • The article mentioned a software receiver based method, more flexible to try different approaches to the problem. Currently our implementation is hardware based, less able to try unconventional methods. Thanks!

  • Another technique worth trying: "for low-quality GNSS antennas ... wavelength-scale random antenna motion substantially improves the time to [single-frequency] integer ambiguity resolution."

  • Thank you Felipe for the suggestion, we'll check it out. With fair performance RTK antenna around $100 ~ $150 apiece, I'm not sure if it's feasible for commercial product to implement this though.

  • You might wish to consider the antenna-swap technique to speed up ambiguity resolution; here's a quotation from chap. 6 in the recent book by Leick et al. (2015), see doi:10.1002/9781119018612.ch6 for details: Antenna Swap Technique 
    In view of modern processing of multifrequency observations, the antenna swapping technique may today be perceived as
    impractical, although when introduced by Remondi (1985) it was an innovative and major step forward in making kinematic surveying practical at the time. Although today multifrequency observations are processed recursively as explained in Chapter
    7 and the transition to the kinematic survey is automatic as soon as the ambiguities are fixed, let us step back for a moment to see how it used to be (at the time surveyors mostly operated single-frequency receivers).
    Basically, a kinematic survey requires an initialization. This means the double-difference ambiguities are resolved first and then held fixed while other points are being surveyed, assuming of course that no cycle slips occurred while the rover moves
    or that cycle slips are repaired appropriately. A simple way for initial determination of ambiguities is to occupy two known stations. The procedure works best for short baselines where the ionospheric and tropospheric disturbances are negligible. The
    double-difference equation (6.4.25) can be readily solved for the ambiguity when both receiver locations xk and xm are known. Usually, simple rounding of the computed values is sufficient to obtain the integers. Once the initial ambiguities are
    known, the kinematic survey can begin.
    Remondi (1985) introduced the antenna swap procedure in order to initialize the ambiguities for kinematic surveying, requiring only one known station. Assume that four or more satellites were observed at least for one epoch while receiver R1 and its
    antenna were located at station k and receiver R2 and its antenna were at station m. This is followed by the antenna swap, meaning that antenna R1 moves to station m and antenna R2 moves to station k, followed by at least one epoch of observations to the same satellites. The antennas remain connected to their respective receivers. During data processing, it is assumed that the antennas never moved. (...) Equation (6.4.57) can be solved for xm, given xk and observations to at least four satellites (three double differences). Once the position of m is known, the ambiguities can be computed from (6.4.53).
    Initialization by antenna swap on the ground is conveniently done for a very short baseline of a couple of meters. A typical [standard/navigation] point positioning solution for xk is sufficient for such short baselines.
  • Hi Alan,

    Here is a brief intro on how S2525F8-BD-RTK can be used:

    For the web store, we are working on a compact size carrier board with S2525F8-BD-RTK + Bluetooth 2.1 + ArduPilot & Pixhawk interface support, so that it can easily be hooked up to build an RTK UAV, and also be used as an RTK receiver for Android smartphone as mentioned in: and Hope to have it before end of September, if not early of October. Pricing will be known later before putting it on the store.


  • Hi Oliver, this looks great. Do you know when the modules will be available on your store? And do you have any idea of the price?




  • Hi Michele, Don't think they have distribution in Europe. Send you an email earlier. Let me know if you didn't receive it.

  • Mind blowing!

    On a sidenote, I noticed how you started using the XN116 in favour of the CAN5125. Is it possible to buy those CAN5125, XT116 or RX3007 front-ends from Europe?

    I would like to design a GPS sampler with a dual constellation chip but large "western" distribution does not carry anything like that!



This reply was deleted.