MMWCAS-RF-EVM: Radar object detect use case clipping maximum range at around 15 meters irrespective of settings.

Asher Farooqui

Part Number: MMWCAS-RF-EVM
Other Parts Discussed in Thread: MMWCAS-DSP-EVM,

Hi,
I'm evaluating Radar SDK use case on MMWCAS-RF-EVM mounted on MMWCAS-DSP-EVM, specifically the use case radar_object_detect.

cascade_radar_object_detect.zip

These are my CC, coefficient.bin, coefficient script, and the use case demo script files. I have changed almost every parameter inside the Matlab script but for some reason the RADAR clips the maximum range to 15 meters. I tried all of the profile numbers as well as the modes supported by the demo script but nothing changed the maximum range of 15m.

The other peculiar thing is each of the graphs has a different maximum range. For instance, during one of my evaluation runs, the range doppler profile shows a maximum distance of 2000 meters whereas the RDM and Range doppler points only show a maximum of 150 meters(as shown below). In the range doppler profile it shows that I covered 2000m but in reality, I only go up to 15. Changing the sampling frequency or the frequency slope didn't help either.

I suspect there's a very basic setting that I'm overlooking that unlocks the full range of MIMO but I cannot locate it at this moment. All help is appreciated.

over 4 years ago

0 Cesar over 4 years ago

TI__Guru*** 141504 points

Hi,

Did you try settingd oob. I think the default config should provide more than 15m.

Thank you

Cesar

0 Asher Farooqui over 4 years ago in reply to Cesar

Expert 1821 points

Hi Cesar,
we're using the OOB demo that came with it. Even with a default config, the range doesn't go beyond 15 m. And the peak in the range doppler profile doesn't die down. It disappears after 15m, which would suggest that it can see farther than that but somehow the output is clipped to 15m.

0 Asher Farooqui over 4 years ago in reply to Cesar

Expert 1821 points

Hi,

Is there a way to debug this issue?

0 Piyali Goswami over 4 years ago in reply to Asher Farooqui

TI__Mastermind 30235 points

Hi Asher

The first thing to do is to check the FFT output and see if the Range and doppler Heat map is able to detect the peaks for the objects at > 15 m distance. If you do not see bright spots at the correct range bins, then the debug is more around the front end configuration, if the spots are visible in the heat map, then the next step would be to see the output of the CFAR to see if the objects are getting detected correctly in that case.

You would need to use the CCS connectivity and step through the code in the AlgorithmFxn_RadarDspCascadeMimoProcess function in the file: vision_sdk\apps\src\rtos\radar\src\alg_plugins\alg_fxns\radardspcascademimo\radarDspCascadeMimo.c to be able to see each stage of the processed output.

Thanks and Regards

Piyali

0 Andreas Baumann over 4 years ago in reply to Asher Farooqui

Intellectual 470 points

Hello Asher Farooqui,

did you find a solution? We have actually the same problem with our system.

Best,

Andreas

0 Asher Farooqui over 4 years ago in reply to Andreas Baumann

Expert 1821 points

Hi Andreas,

I found out changing the chirp parameters in the source code solved the problems for me. I think Matlab was unable to overwrite chirp parameters from the console. I have yet to try Piyali's suggestion. Please let me know about your findings if you do plan on following her suggestion for debugging.

0 Andreas Baumann over 4 years ago in reply to Asher Farooqui

Intellectual 470 points

Hello Asher,

unfortunately we stumbled over an other error. It seems that device 2 is not starting. (https://e2e.ti.com/support/sensors/f/1023/t/881793)

I will come back to you when we can continue with this issue. :)

Cheers,

Andreas

0 Piyali Goswami over 4 years ago in reply to Andreas Baumann

TI__Mastermind 30235 points

Hi Asher,

I believe your original issue is resolved.

Hi Andreas

Once you are able to get past the SPI issue, please let us know if you are able to proceed with the 15 meters issue.

Thanks and Regards

Piyali

0 Asher Farooqui over 4 years ago in reply to Piyali Goswami

Expert 1821 points

Hi Piyali,

You are correct, we figured out that the range can be increased by modifying the chirp parameters in the source code of the firmware. The issue still remains is to figure out why Matlab is unable to overwrite the chirp parameters.

0 Piyali Goswami over 4 years ago in reply to Asher Farooqui

TI__Mastermind 30235 points

Hi Asher,

When you run the data processing chain on the TDA2x platform, the MATLAB input is not taken into account to configure the front end.

If you would like to change the paramters without having to recompile the code when you are running the data processing chain on the TDA2x you would need to store the configuration in an SD card which the TDA2x can read.

Refer: 3.9.1.6.5 File-based Use Case Configuration Support in vision_sdk\docs\Radar\ProcessorSDKRadar_UserGuide.pdf

The MATLAB script for this demo is only used to make sure that the MATLAB based visualization is able to interpret the output of the TDA2x correctly.

Thanks and Regards

Piyali

0 Andreas Baumann over 4 years ago in reply to Piyali Goswami

Intellectual 470 points

Hello Piyali,

I finally managed it (Adjusted the Chirp Parameters in the LUA-Script) and can now receive information beyond 15 meters.

Thanks and Regards,

Andreas

0 Piyali Goswami over 4 years ago in reply to Andreas Baumann

TI__Mastermind 30235 points

Andreas

Glad things are now working for you!

Regards

Piyali

Sensors

Sensors forum

MMWCAS-RF-EVM: Radar object detect use case clipping maximum range at around 15 meters irrespective of settings.