MMWCAS-DSP-EVM: Device is not identified on startup

Hi,

Due to local holidays please give us a few days to get back to you

thank you

Cesar

Hello Harald,

I apologize for the delay.

Looks like your mmWave studio wasn't installed properly or the device is not connected to your pc correct.

Can you send a picture of your setup and the available COM ports in your device manager window?

In terms of mmWave Studio Installation, there is no absolutely no difference for a singlechip or a cascaded device, the steps are exactly the same. Just make sure you've all the prerequisites. What is the mmWave studio version you've installed here?

Also, in cascade if you want to load the firmware over SPI, SOP4 mode is used and no RS232 connection is required in that case. Once your mmWave studio is setup properly, follow the steps provided in the mmWave Studio cascade user guide. But let's look at this at a later stage and get your studio set up first.

Regards,

Ishita

Hi Ishita,

some days' delay don't really matter when the initial planning is good...

mmWStudio is Version 3.0.0.14 , the SD image is 3.7.1., the DPF is 2.2.2.1

Nevertheless, we have been working on the issues during the last few days. There are some good and some bad news.

1: The first error (index out of bounds) is still around. It affects obviously the Matlab distributable only.

2: All other errors are solved. They have been caused due to some documentation issues: We use a cascade radar and did not find out for a while, that there are some different user manuals for mmWave Studio for this case. (Would be nice to have this mentioned in the user's guide pretty much the same way as you already do in the DFP guides: A red warning text, maybe :-) )

3: Flashing the SD was no problem. The formatting of the SSD was not needed, since it was already pre-loaded. Funny detail is, that on the machine with mmWaveStudio on, I cannot see any of the 115200 baud data using Tera Term or puTTY. On another Win10 machine (which has no Ethernet, though), I can see the complete startup data coming out properly. 

4: Now we are able to connect the device using the SOP4 mode and SPI and all 4 AWRs are detected and can be powered up.

--> The BSS file used is from the DPF: \firmware\xwr22xx_metaImage.bin. I use this as Masterss and Radarss. Is this ok? 

--> In the "SensorConfig", ARM fails (see attached documents) and the lua script fails as well...

Yours

Harald

USB: 

mmWStudio raw

mmWStudio powered

Output during setup (Matlab problem):radar_Output_during_setup.rtf

Output when trying to Arm:way_to_hell_1.rtf

output running Capture.Lua.

Hi again, Ishita!

Over the weekend I made most of the stuff running. At least when using LUA scripts (Cascade_*.lua). Via the GUI I am still stuck at the above situation. But at least I know that the system is healthy and it is "just a question of settings". So now it will take only a countable infinity of seconds until everything is in place :-) ...

Yours

Harald

Hello Harald,

Glad to hear you were able to get some of your stuff working.

Please help me understand a few things :

1. Are you able to capture data with the default scripts i.e using LUA scripts that come with mmWave studio package?

2. The errors you're getting above is when you try to setup the GUI manually right?

As I can see from your screenshots, the TDA width and height is 0, which means there was some configuration that was missed or given wrong.

Your output logs seem to be incomplete.

Would it be possible for you to highlight the steps you're doing to capture data using a video or something like that? As you said, its just a matter of a couple of settings and we're just trying to figure out what is that you missed.

Regards,

Ishita

Hi Ishita,

we will start some captures later today. Simply because we need to understand what is in the captured data files....

Simultaneously, I will document all steps.

So please bear with me for some hours.

Just a fundamental one: Are the ADC outputs big-endian or small-endian 12 bit numbers? There is clearly one nibble set to 0000 in the data.

Where can I find a format description for the *.bin files?

Yours

Harald.

Hello Harald,

Harald Merkel said:

So please bear with me for some hours.

Sure, please take your time and get back when you're done. But just to let you know, the correct sequence is followed in the example lua scripts, so in case that works for you, it is a green signal. You can take help from the lua script to understand the steps we're following in mmWave studio.

Regarding the format of .bin file, please refer to the below links :

https://www.ti.com/lit/an/swra555a/swra555a.pdf?ts=1618833266998

https://e2e.ti.com/support/sensors/f/sensors-forum/952458/tidep-01012-raw-data-format-of-tidep-01012/3519917#3519917

Regards,

Ishita

Hi Ishita,

thanks for the link.

The first one is a piece of solid gold. We are going through it. This is much better than reverse-engineering the Matlab code.

Referring to this code, it would be rawDataReader.m to start with the *.bin files, do I guess right?

And this code would be applicable to Cascade radars as well?

It is interesting to note that the pair { Cascade_Configuration_ContStream.lua , Cascade_ContStream.lua } scripts work fine but not the { Cascade_Configuration_Basic.lua , Cascade_Capture.lua } pair. Do you know why?

There it still indicates "ARM failure". 

Yours

Harald

Hello Harald,

Harald Merkel said:

It is interesting to note that the pair { Cascade_Configuration_ContStream.lua , Cascade_ContStream.lua } scripts work fine but not the { Cascade_Configuration_Basic.lua , Cascade_Capture.lua } pair. Do you know why?

Cascade_Configuration_Basic.lua script configures the radar device only till "DataConfig" Tab. Before pressing the TDA ARM button, the device needs the profile, chirp and frame configuration as well (a part of "SensorConfig" Tab) which is not a part of that script.

You can either do that manually in mmWave Studio, or use Cascade_Configuration_TestSource.lua with Cascade_Capture.lua. (In case you don't want to use test source there, just disable the test source in line 50 of that script).

Regards,

Ishita

Hi Ishita, this makes sense. I'll try asap.

Thanks a lot. 

Now there are three small problems left:

1: Apparently the radar collects data and there are files transferred to my computer when I run the ContStream scripts. In the JSON files though, it is indicated that there are no captured files are available. It seems that the files that I get from the DSP are always the same and their date is set back to 2019.

2: There are files copied into a folder with a name that is not equal to the folder name specified in the JSON file.

{"numFilesCollected" -> 0,
"fileBasePath" ->
"C:\\ti\\mmwave_studio_03_00_00_14\\mmWaveStudio\\PostProc\\adc_\
data", "files" -> {}}

3: Having gone through the swra555a.pdf it tells what the chips would do but there is not a lot of information on how this output is converted to a bin file. On the other side, Kawaguchi-san's link is excellent.

--> Would it apply to the Cascade radar as well? 

Kawaguchi-san has almost exactly the same date settings and date discrepancies on the files, that I have.

The conversion routine using Mathematica is growing and at a certain time, it will work (knock on wood!). Since you do help me a lot, I could mail you the Mathematica script when it is working and you can post it on this forum and distribute it, if you like. Clearly, not a lot of people in this community use Mathematica but there may be some, especially with theoretical physics as background....

Yours

Harald

Hello Harald,

Please find my comments below :

1. The date of capture in SSD is set somewhere between May/June 2019 and hence you're seeing that, but don't worry about that, the captured files are the latest ones. For confirmation, you can try changing the capture file folder name and after every capture session, you can look for the corresponding files in /mnt/ssd directory.

Also, regarding the parameter "numFilesCollected" in the setup.json file, it is a don't care parameter and doesn't reflect the capture file status, hence you can ignore that.

2. "fileBasePath" variable reflects the capture folder name and you should see the name you set. Are you looking at the correct setup.json file here?

3. The following link that I attached in my previous reply applies to cascade radar as well : https://e2e.ti.com/support/sensors/f/sensors-forum/952458/tidep-01012-raw-data-format-of-tidep-01012/3519917#3519917

Also, we're very glad to hear that you want to help the entire community with your findings. In case you want to share that, please feel free to post information about that in a new thread and I will mark that as FAQ for our wider audience.

Thank You again for you help and cooperation.

Let me know if you need clarification on any of the above points.

Regards,

Ishita