Startup Troubles on Dev Kit

[SteveHx]

I was able one time to power up the nRF9160 dev kit and see it connect to the cellular network and to nRF Cloud. Since then, it will only blink LED3, which indicates that it can't connect. I would like to re-flash the firmware, but I'm totally lost in all the circular links in the documentation. I have used the nRF52832 dev kit extensively so I'm quite familiar with the process, but it appears the nRF9160 dev kti adds several more layers of complexity. I would like to be able to load a simple "hello world" application, and then build on that with my application source code. Please guide me to some concise documentation.

[SteveHx]

I understand what it's asking for. What I need to know is where to get the GNU ARM toolchain - I don't think I have that installed. In any case, I always allow published installers to install to their default directories, because I've found way too many whose developers didn't bother to check what happens if the user uses a different directory - it often breaks the product. Where should I get the GNU ARM toolchain, in a version compatible with this toolchain?

[Didrik Rokhaug]

Sorry, I misunderstood what you were asking for. In the Getting Started Assistant there are two links at the bottom of step 1 "Install the toolchain":

Windows 32-bit Installer (Signed for Windows 7 and later)

Windows 32-bit Installer (Unsigned)

We recommend to install the toolchain into the c:\gnuarmemb folder

[SteveHx]

Well, I'm still slogging along here. I finally got to the step of opening an nRF Connect project, entered the starting paths etc. as shown in the Getting Started Assistant, and here's the result:

I don't understand why this process is so difficult, when I installed exactly according to the directions, and stayed with all the default directories, etc. Did anyone test this before release?

How should I best resolve the problem reported in the box above?

[Didrik Rokhaug]

What did you set the environment variable GNUARMEMB_TOOLCHAIN_PATH and the SES option "GNU ARM Embedded Toolchain Directory" to?

And where did you install the arm-none-eabi toolchain?

Regarding where to unzip SES to, the Getting Started Assistant says:

"Extract SEGGER Embedded Studio to your preferred folder."

I put it in C:\

[SteveHx]

Re: What did you set the environment variable GNUARMEMB_TOOLCHAIN_PATH and the SES option "GNU ARM Embedded Toolchain Directory" to?

There are no such environment variables on my system. Are you referring to something other than environment variables, and if so, where would I find them?

Re: And where did you install the arm-none-eabi toolchain?

I don't recall seeing that. Whatever it was, I simply took the defaults at every step. This process is so long and convoluted I'm losing track of anything I did. Certainly nothing like the comparatively clean and simple installation process for the 52832 dev kit.

Re: Regarding where to unzip SES to, the Getting Started Assistant says:
"Extract SEGGER Embedded Studio to your preferred folder."
I put it in C:\

Mine is in C:\Segger. I do not recall explicitly choosing that directory.

Another issue: Every time i start up SES, it claims there's no license installed. I request and receive an activation link for use with Nordic products, install that, and it's fine - until I close SES and later reopen it.

• Up 0

[Didrik Rokhaug]

You need to create the GNUARMEMB_TOOLCHAIN_PATH environment variable (along with the ZEPHYR_TOOLCHAIN_VARIANT, though I believe they are only needed when building from the command line).

I sent you links to the toolchain in this comment: https://devzone.nordicsemi.com/f/nordic-q-a/52266/startup-troubles-on-dev-kit/211001#211001

Make sure that you use the "Nordic Edition" of SES, not the regular version (links here: https://devzone.nordicsemi.com/f/nordic-q-a/52266/startup-troubles-on-dev-kit/211314#211314).

You can unzip the folder anywhere. There are no other install steps other than unzipping. When using SES with NCS, make sure that you run the correct executable, you can find it in <SES install/unzip folder>/arm_segger_embedded_studio_v418_win_x86_nordic\bin\emStudio.exe

When you first open SES (Nordic Edition), you need to set two options. You can find them by clicking Tools > Options and select the nRF Connect tab.

• Zephyr Base - the location of your cloned Zephyr repository (the full path to ncs/zephyr)

• GNU ARM Embedded Toolchain Directory - the location of your GNU ARM Embedded Toolchain installation (for example, c:\gnuarmemb)

I agree that there are a lot of dependencies to install, however, if you follow the Getting Started Assistant application in nRF Connect it should not be very hard.

[SteveHx]

Every time I open Embedded Studio I get this:

Whether i answer yes or no, I always get a notice that I have no license, and have to jump thru the hoops to get the free license for Nordic users. is there any way to make the license stick?

[Didrik Rokhaug]

Hi.

1. The Segger Embedded Studio Nordic Edition zip should contain all the necessary parts, and installing a package should not be necessary. I have never encountered that message before, and from the path mentioned in the pop-up, it does not look like you are using the Nordic Edition of SES. Can you confirm that you are using the correct SES version?

2. I just tried to set up NCS on two different Linux VMs and encountered your license issue on one of them until I was able to open one of our example projects. For now, you should be able to simply press "continue", while we sort out your other issues.

3. Can you send me a screenshot of the "nRF Connect" tab of SES's options window, as well as the content of the folder, pointed to by the "GNU ARM Embedded Toolchain Directory" setting?

Best regards,

Didrik

[SteveHx]

1. The Segger Embedded Studio Nordic Edition zip should contain all the necessary parts, and installing a package should not be necessary. I have never encountered that message before, and from the path mentioned in the pop-up, it does not look like you are using the Nordic Edition of SES. Can you confirm that you are using the correct SES version?

If I "continue" as you suggest below, I do not get the Nordic version. If I do as I have been doing all along, and re-activate my license every time I open SES, I clearly get the Nordic version as this screenshot shows. Could there be a connection here? Also, note that my legal company name includes a comma, but the activation will not accept a comma in the company name so I must compromise and leave out the comma.

2. I just tried to set up NCS on two different Linux VMs and encountered your license issue on one of them until I was able to open one of our example projects. For now, you should be able to simply press "continue", while we sort out your other issues.

See above.

3. Can you send me a screenshot of the "nRF Connect" tab of SES's options window, as well as the content of the folder, pointed to by the "GNU ARM Embedded Toolchain Directory" setting?

Note that all locations etc. were guided by the nRF Connect Getting Started Assistant.

[Didrik Rokhaug]

1. On my Linux VM, I got the Nordic Edition also when pressing continue. I do not know of any other solution, but what I did was simply to type a single letter into every box (except for email, which must look like an email, e.g., [email protected]), and use an old license. It is obviously not a very good workaround, but that is the best I have at the moment.

3. The "GNU ARM Embedded Toolchain Directory" is wrong. It should point to the folder where you installed the toolchain, i.e., compiler. The default path I got from the installer was: C:\Program Files (x86)\GNU Tools Arm Embedded\7 2018-q2-update. If you followed the recommendation from the Getting Started Assistant, it would be in C:\gnuarmemb. Regardless, the content of the folder should be:

[SteveHx]

David Martin stepped in to help, but he's traveling today so asked that I come back to this ticket. So here's the update. We went back and started over with the Getting Started Assistant. Step 1 of the 4 steps is complete and appears successful. In step 2 "Clone the nRF Connect SDK", the first 3 subparts are complete. In the 4th part, installing Python modules, the very first install command is

pip3 install -r zephyr\scripts\requirements.txt

That results in an error message that David was unable to handle immediately. Here's the screen dump from the command box:

11:39:42.08 d:\296\FW\ncs> pip3 install -r zephyr\scripts\requirements.txt
Collecting wheel==0.30.0 (from -r zephyr\scripts\requirements.txt (line 1))
  Using cached https://files.pythonhosted.org/packages/0c/80/16a85b47702a1f47a63c104c91abdd0a6704ee8ae3b4ce4afc49bc39f9d9/wheel-0.30.0-py2.py3-none-any.whl
Collecting breathe==4.9.1 (from -r zephyr\scripts\requirements.txt (line 2))
  Using cached https://files.pythonhosted.org/packages/2c/a5/040234179dc9140be4a1cb4350af362327d001d0ca7e5bba4d580916ceb4/breathe-4.9.1-py2.py3-none-any.whl
Collecting sphinx==1.7.5 (from -r zephyr\scripts\requirements.txt (line 3))
  Using cached https://files.pythonhosted.org/packages/1b/9f/98d67201c5a6e1aececed03a44a819d0e32adba81414081e303cfaf8c54c/Sphinx-1.7.5-py2.py3-none-any.whl
Collecting docutils==0.14 (from -r zephyr\scripts\requirements.txt (line 4))
  Using cached https://files.pythonhosted.org/packages/36/fa/08e9e6e0e3cbd1d362c3bbee8d01d0aedb2155c4ac112b19ef3cae8eed8d/docutils-0.14-py3-none-any.whl
Collecting sphinx_rtd_theme (from -r zephyr\scripts\requirements.txt (line 5))
  Using cached https://files.pythonhosted.org/packages/60/b4/4df37087a1d36755e3a3bfd2a30263f358d2dea21938240fa02313d45f51/sphinx_rtd_theme-0.4.3-py2.py3-none-an
y.whl
Collecting sphinxcontrib-svg2pdfconverter (from -r zephyr\scripts\requirements.txt (line 6))
  Using cached https://files.pythonhosted.org/packages/79/83/d712a25303429570c641caa113d1282c27012bbd4157417d07829947a549/sphinxcontrib_svg2pdfconverter-1.0.1-p
y3-none-any.whl
Collecting junit2html (from -r zephyr\scripts\requirements.txt (line 7))
  Using cached https://files.pythonhosted.org/packages/69/93/f2ccef1136dba078dada0513c324bad9ffd2be9236229d09bce71e1338a8/junit2html-022.tar.gz
Requirement already satisfied: PyYAML>=3.13 in c:\python37\lib\site-packages (from -r zephyr\scripts\requirements.txt (line 8)) (4.2b4)
Collecting ply==3.10 (from -r zephyr\scripts\requirements.txt (line 9))
  Using cached https://files.pythonhosted.org/packages/ce/3d/1f9ca69192025046f02a02ffc61bfbac2731aab06325a218370fd93e18df/ply-3.10.tar.gz
Collecting hub==2.0 (from -r zephyr\scripts\requirements.txt (line 10))
  ERROR: Could not find a version that satisfies the requirement hub==2.0 (from -r zephyr\scripts\requirements.txt (line 10)) (from versions: 0.0.1, 0.1.0, 0.1.
1, 0.1.2, 0.1.3, 0.1.4, 0.1.5, 0.1.6, 0.2, 0.2.0.1, 0.2.0.3, 0.2.0.4)
ERROR: No matching distribution found for hub==2.0 (from -r zephyr\scripts\requirements.txt (line 10))

11:40:06.99 d:\296\FW\ncs>

Your serve...... ;-)

[Didrik Rokhaug]

That is a known problem and has been fixed on the master branch.

What has happened is this:

The hub python module that is used by Zephyr has changed the name to git-spindle. Then, a new module was created with the name hub. If you edit the name of the package in the requirements file to "git-spindle", it should work.

Sadly, bugfixes like this have not been backported to the v1.0.0 release.