Getting started with Ultra-Wide-Band 3D Positioning DWM1000 and DWM1001 Modules

The possibilities of accurate 3D tracking are endless: Drone navigation, logistics, human-machine-interfaces… but unfortunately developing Real Time Location Systems is expensive and takes a lot of time, or does it?

This is how you can start developing your custom applications with DWM1001 modules for under $20 in under an hour.

 

 

0. Preparing the environment

We will be using Segger Embedded Studio because it provides a free license for Nordic nRF family development.

 

Once installed, we will need the following packages:

  • CMSIS 5 CMSIS-CORE Support Package (version 5.02)
  • CMSIS-CORE Support Package (version 4.05)
  • Nordic Semiconductor nRF CPU Support Package (version 1.06)

 

You may encounter build errors using the latest SES version, it can be better to install an older one. Version 3.34a has been verified to work.

 

You can build the examples and start getting measurements in minutes!

 

 

 

1. Getting to know the API

DecaWave provides a very comprehensive API that allows you to integrate your design ideas in a minimal amount of time.

The ways you can access the API are:

  • User C code: Using the provided toolchain.
  • SPI or UART: Using Type-Length-Value format.
  • UART: Shell mode.
  • BLE: Via services and characteristics. (The documentation on the BLE API is incomplete, but you can check the Android app code for reference).

Even though lots of examples and a very decent amount of documentation is provided; unfortunately, the PANS API source is not provided, we would like to see it published to be able to customize or extend it.

 

The API structure builds up from the NRF SoftDevice, which should be familiar for those of us who have worked with Nordic devices in the past. Just mention that this is closed-source too. It includes BLE Central, Peripheral, Broadcaster and Observer roles, and supports OTA firmware upgrades. The included version is S132.

 

 

The eCos RTOS includes the Board Support Package (BSP), and all the drivers for BLE, Accelerometer, and most importantly, the DW1000 driver.

A quick way of knowing if the module is being moved physically is implemented through a component of I2C Accelerometer LIS2DH12TR slave at address 0x19. You can read it at 0x33 and write it at 0x32.

 

The Android App is really useful for a quick first-time configuration and for programming the modules over-the-air.

 

 

 

DRTLS Network

The stack allows discovery, joining and leaving the network. The UWB packets adhere to the standard 802.15.4 format:

 

 

Two-Way-Ranging Solver (Location Engine)

Likely the most important part for us. This part of the API calculates the X, Y, Z coordinates from the results of TWR. Although it’s very useful to get up to speed quickly, this engine can be disabled to use custom filtering and solving algorithms.

 

 

 

 

 

2. The development toolchain

It is really great that DecaWave provides such a comprehensive set of resources already configured, download the toolchain and with this guide, you will be developing in minutes. Kudos to DecaWave for this.

You can see that the PANS API, being closed-source, is provided as a compiled static library .a, and the nRF SoftDevice as .hex.

 

 

Restoring the factory image

If you ever need to go back to factory settings, download any binary or erase the chip, just use J-Flash Lite, which you can download from J-Link Software and Documentation Pack.

 

 

 

Setting up the development Virtual Machine

You can create a New virtual machine with these settings:

Assign a generous amount of RAM, and select the provided .vdi to be used as Hard disk.

Or just Add a new machine and use the default settings by selecting the .vbox file.

 

 

Once ready, you can run it and login: user dw password dw

 

Download the sources from DecaWave. The link may go obsolete, look for a package with a name similar to “DWM1001, DW10001-DEV and MDEK1001 Documents, Source Code, Android Application & Firmware Image”.

At this point, the documentation is a bit outdated. You need to find the dwm folder which is in DWM1001_DWM1001-DEV_MDEK1001_Sources_and_Docs_vX/DWM1001/Source_Code/dwm1001_on-board_package_v1p0 and place it inside /Documents in the Virtual Machine.

The easiest way to copy the folder into the Virtual Machine is creating a Shared Folder (Devices > Shared Folders). Create a new one, point it to a folder in your computer and select Auto-mount and Permanent. Reboot the VM and you should see the shared folder as a removable device or inside /media/

 

 

Now, you can connect the DWM1001-DEV via micro-USB. For the same purpose, you can connect any J-Link that is connected to a nRF52832.

Make sure the J-Link device is ticked inside the Virtual Machine. That will make the device available inside the VM.

 

 

Factory reset

To verify that the setup is working, you can reset the device to factory settings with:

cd ~/Documents/dwm/examples/dwm-simple
make recover

It should end after a few seconds and the onboard LEDs with be lit.

 

 

 

 

3. Using your own code

This is no doubt the main purpose of this guide. As introduced before, there are several ways to use your own code to control DWM1001 and DWM1001-DEV modules. Let’s try an example of each one of them.

 

3.1 User C code within the DWM1001 firmware

This is the way to go to integrate custom C code that runs on-board while using the high-level PANS API.

The 512KB flash (0 to 0x80000) is partitioned like this:

  • SoftDevice: provides BLE capabilities, you can find more information and examples in the Nordic Semiconductor website.
  • Bootloader: Allows choosing between FW1 and FW2 during boot.
  • The Environment area is used for configuration and is therefore preserved across power cycles, and even firmware re-flash. To clear it you must issue a full erase.
  • FW1 is used for the OverTheAir firmware update.
  • FW2 can be up to 240KB in size. This image includes the PANS lib and the user application. Here comes one limitation for the user code:

When bundled inside FW2, the user code runs as a thread and can only take up to 60KB in flash, and use 3KB of RAM.

 

 

To get started, right-click inside the Project Explorer and import the projects in /home/dw/Documents/dwm

Since we want to develop user code, and it has to be inside FW2, we have to select the second option in the build drop-down menu.

 

To create user code:

  • Add functionalities inside dwm/examples/dwm-simple/dwm-simple.c
  • All the C API funtions that you can use are listed in dwm/include/dwm.h

To quickly see how user code is executed, in dwm-simple.c add:

...

dwm_pos_t pos;

...

while (1) {
  /* Thread loop */
  dwm_pos_get(&pos);
  printf("x=%ld, y=%ld, z=%ld, qf=%u \n", pos.x, pos.y, pos.z, pos.qf);
  printf("\t\t Time=%lu \n", dwm_systime_us_get());
}

 

To download and debug the firmware right-click the project and select Debug Configurations.

You will find dwm_openocd, which by the way is a great on-chip debugger that works with GDB, and even has some interesting Boundary Scan capabilities that can be useful in FPGA design process.

 

Once you click Debug, the firmware will be downloaded and the Debug perspective will be opened. Here you can use all the Eclipse tools to debug your code.

To see the modifications running, disconnect the device from the Virtual Machine, and open a serial terminal to the J-Link COM Port. The default baud-rate is 115200 bps. The added console output should appear. If nothing shows up, start the Shell Mode (more on this in a minute) by pressing Enter two times.

 

Now you can start tweaking the code and adding more complex functionalities!

One important thing to mention is that this code is executed along with all the functionalities already available in the factory firmware, so you will still be able to use those components too.

 

It’s precisely one of this components, the BLE included in the SoftDevice, that can generate debugging problems. Since its interruptions have the highest priority, they will conflict with the user interrupt. Disable dwm_ble_compile() while debugging or mask the BLE interrupts to avoid this.

 

 

 

 

3.2 Interface the API from an external processor via UART

You can connect to the UART either through the USB-Serial converter, which shows up as a COM Port under the J-Link hardware, or through the pins:

If you are using a Raspberry Pi, it will show up as /dev/serial0

 

The UART works in two modes that can be switched:

  • GENERIC (default): Use Type Length Value encoded commands. To switch to Shell Mode, send Enter twice [0x0D, 0x0D].
  • Shell Mode: In this mode, you can interact with the module via cli. To go back to Generic Mode, send quit.

 

 

Shell mode example

With the J-Link device under the Linux VM, open a serial port with:

minicom -D /dev/ttyACM0

Enter Shell Mode pressing Enter twice. Now, the same function that we implemented before using User C code, can be run with the apg command:

 

 

Generic mode example

If you have a Raspberry Pi you can quickly test the Generic Mode with the host API, as the HAL is already provided.

Download the dwm1001_host_api package. (It is included in the complete docs&sources package), and browse the examples\ex1_TWR_2Hosts\tag folder from inside the VM. You can copy it to the shared folder we created previously. (Inside the downloaded package, the path will be similar to DWM1001_DWM1001-DEV_MDEK1001_Sources_and_Docs_v8\ DWM1001\Source_Code\DWM1001_host_api\dwm1001_host_api\ examples\ex1_TWR_2Hosts\tag).

Edit the Makefile, for example with

nano Makefile

Change

TARGET = 0
...
INTERFACE_NUMBER = 0

Make and run, and you should see the output as follows:

make
./tag_cfg

 

To use a custom platform you need to:

  • Provide your own platform HAL.
  • Assign it to HAL_DIR in dwm1001.mak (inside DWM1001_host_api\dwm1001_host_api\include\)

 

3.3 Interface via SPI

The encoding used in SPI is the same as in UART Generic Mode so you just need to connect the appropriate pins

And change the Makefile to

INTERFACE_NUMBER = 1

 

 

 

 

Conclusion

We’ve seen that the amount of available resources is really amazing, you can get started in minutes. The amount of ways in which you can interface the device allows for this modules to be integrated in almost any kind of system.

Once you have tested the different ways to control them, and you have decided which interface you will use, it’s time to dive into the API and the libraries. That’s the best way to create a reasonably highly custom-tailored solution in the shortest amount of time.

 

 

References

This guide is based on the official documentation from DecaWave, which is subject to changes. Not many links are provided since they break with each update. It’s better to browse https://www.decawave.com/ and download the latest versions directly. Always refer to the up-to-date official documentation when working on designs.