IWR1443BOOST: flat surface object turned gets ignored

Hello,
You can consider the radar Tx beam like a flashlight, that has a focusing ring. The IWR1443 BOOST EVM has a field of view in azimuth of +/-70 degrees, elevation +/- 25 degrees. The IWR 1443 BOOST EVM User Guide Figure 12 shows the antenna gain profile.

The RCS Radar Cross Section of the object, is reflecting the signal power back to the Rx. Depending on the radar configuration, you can detect the signal level. As you turn the object, if you note the magnitude of the signal peak vs the noise floor , the magnitude drops , the Constant False Alarm Rate CFAR, is the SNR above the background noise. if the RCS return is lower in magnitude, you would need to:
change the Rx LNA gain - more gain provides a stronger return signal

change the number of chirps per frame, or the chirp design for more samples - more integration time, we can detect a smaller RCS target

change the CFAR setpoint so less magnitude DB difference is needed to detect an object.

Note: These can be changed by saving the base Visualizer configuration, and modify the parameters following the mmwave User Guide, the swra553 chirp application note, or other changes you can make.

The mmwave response to a turned object is discussed in this TIREX
"dev.ti.com/.../

I will send a lab experiment tomorrow, I will use a metal plate to show how this works.

Regards
Joe Quintal

Hello,

I have used an IWR1443, mmwave SDK 1.1, used the Out of Box demo, Best Range resolution.  T

he metal plate in the picture IMG_1101.jpg, is the boresight center access of the IWR1443 BOOST EVM antenna at about 2.4meters distance.    

In picture IMG_1102.jpg is the Visualizer tool GUI that is used with the Out of Box Demo, note the top right graph has a magnitude peak @ about 2.5meters.  The top left shows the XY scatter plot of the detected object.

If we rotate the plate to the left or right, the RCS return  energy is  not focused to the mmwave receive antenna.   If you look at picture IMG_1103.jpg you can see the metal plate is rotated  to the right of the mmwave antenna.   Much less energy is now returned.   Given the initial setup of the example, the return peak magnitude had to be 15 db above the surrounding clutter.  The peak at 2.5meters is no longer 15db above the surrounding clutter, so the orange dot is not present (top right), and the detected green dot is off in the top left of IMG_1104.jpg.

So we either need to change the CFAR db setting, or redesign the software-chirp, or have a different antenna.   In the next example, we changed the Visualizer CFAR setting to 12db.   (this is on the Configure portion of the Visualizer), then resend the configuration.   In IMG_1106,jpg, there are more objects detected, one of them is the metal plate.

You can use the save configuration (Configure page), and load (Plots page) of the visualizer to save your custom settings.

I have attached the plots in the zip file.

If we have a certain reflection from an object called RCS, if the RCS is smaller, you need:

   a longer integration time, more chirps per frame, more samples per chirp.

   a higher gain antenna (physical antenna design or antenna combinations, or higher LNA gain (profile configuration setting)

   a lower CFAR setting

swra553 discusses the radar equation and profile, chirp, frame configuration

Regards,

Joe Quintal

OOB_Metalplate_example.zip