AWR1642: the questions about the function MmwDemo_rangeBiasRxChPhaseMeasure

Hi,

Can you please provide some more background about your question?

Thank you
Cesar

Hi Cesar,
If I want to conduct an experiment aboutcalibrating the range bias and Rx channel gain/phase offset ,how should I do it?In the index.html,the first step is "1.Set a strong target like corner reflector at the distance of X meter (X less than 50 cm is not recommended) at boresight."I don't unsderstand this step.
Firstly.I want to know the distance between the radar and the target, just more than 50cm and less than the max measure distance is OK, is that right?
Secondly, what should the angle be between the radar and target ?Should it be 0 degree,or any angle in the permission scope?
Regards,
Rata

Refer to the figure "coordinate geometry" in the "Data Path -X-Y estimation" of the doxygen documentation, where the X=0,Y=0 is flush with the sensor and X-Y is the azimuth plane. Because of antenna routing length differences and other factors that are specific to each EVM and some delays in the RF chain, if you place an object at a known position in the X-Y plane, what you may see as detection in the plots or looking at the x-y data in the point-cloud for that object may not match that position. E.g if you place a strong reflector at X = 2 m (this distance you know up-front using a scale or some other way or measuring, in other words the ground truth) and Y = 0 or (R,theta) = (2,0), then when you see the plot, it may show up at say (R,theta) = (2.3 m, 20 deg), here the 0.3 m is range bias and there is also phase error of 20 deg [these are not typical numbers, they are exaggerated for explanation purposes]. The purpose of the calibration procedure is to take out this error that is a property of the device and the particular EVM so you can see the object where it actually is. So this procedure requires you to specify the ground truth and this is currently restricted at boresight i.e Y = 0 so you need to place the object along the X axis, less than the max range but not too near ( we say > 50 cm because we don't support near field calibration presently which has more complexity) and then tell through the CLI this ground truth X position. Once the procedure is done, you get the calibration vector from the code on the CLI and then you embed this permanently in all your configurations so all objects for all configurations will use this basic calibration data to produce the expected point cloud data. Now during the measurement you can always place the object at some other orientation e.g at R = 3 m and theta = 25 deg, and then if you do the procedure, it will show the object once calibration is done at X = 3*cos(25), and Y = 0. i.e it kind of rotates to the boresight, but this is usually not what you want to do (you lie about the ground truth) but this is one way in which we test this feature particularly for the angle error because for most boards we already see angle is already good without calibration [range bias is always there in some delta over a mean] so we deliberately place object at a non-zero angle and when we do the calibration, the object on the plot reorients at theta = 0 (Y = 0). [Note at boresight (Y=0 or theta = 0), X = R hence we document as X meters, but the code is actually measuring the range R where object is located].