• State
  • Locked Locked
  • Replies 7 replies
  • Subscribers 30 subscribers
  • Views 972 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

AWR2243: AWR2243 matlab post process results look the same for different inputs

Chen Zhang
Prodigy 30 points
Part Number: AWR2243
Other Parts Discussed in Thread: AWR1843

I had issue with AWR2243 data post process. I described the issue  wrongly previously in the forum and the problem did not get resolved. Here i rewrite my issue as follows:

---------

The outputs (processed by cascade_MIMO_signalProcessing.m) were almost the same for the following different inputs:

1. put the board toward to the sky (1s): Data here: www.dropbox.com/.../sky_04252021.zip

2. road side with pedestrian and SUV (5s)

3. corner reflector put at 20ft distance (1s)： Data here: www.dropbox.com/.../cr_20ft_04252021.zip

we used MIMO Configuration for data capture:

1. Cascade_Configuration_MIMO.lua for configuration of devices  (test_source_enable  =   0      -- 0: Disable, 1: Enable)
2. Cascade_Capture.lua for capturing data

Here is calibrationObj info read from matlab workspace

To remove the impact of calibration, we set calibrationObj.adcCalibrationOn=0. Basically, the process script generated the same plots with above mentioned three dataset.  (if calibrationObj.adcCalibrationOn=1, the conclusion is the same.)

Output with calibrationObj.adcCalibrationOn=0

We did compare the raw data and they are similar but apparently different. 

 Is it HW issue, data collection issue, configuration issue, or data process setting issue?  

  • 0
    TI__Mastermind 31680 points

    Hello Cheng,

    A couple of questions here: 

    Chen Zhang said:
    We did compare the raw data and they are similar but apparently different

    1. What do you exactly mean from the above statement?

    2. What is your profile, chirp/frame configuration here? Are you using the default script with any modifications here?

    3. Is the MATLAB script using the default calib matrix here or have you got new measurements for your board?

    I would want you to try a few experiments here :

    a) Try capturing data with the test source enabled and see the post processed results after that.

    b) Instead of using the MIMO script with multiple chirps/TX combination, try to do a simple experiment with target at a known distance, all TXs enabled at once and using only one chirp. You could use the testSource script for this. Just disable the Test source while capturing real time data.

    c) If you get similar observations with the above experiments, try the cascade device as a singlechip device by enabling only the Master device and analyze your results.

    Regards,

    Ishita

  • 0 in reply to Ishita Kochar
    Prodigy 30 points

    1. What do you exactly mean from the above statement? 

    We compared the binary data with different inputs and they were different. This is to ensure we actually loaded different dataset when we switched to a new dataset (for sanity check)

    2. What is your profile, chirp/frame configuration here? Are you using the default script with any modifications here?

    We used the default config file i.e. Cascade_Configuration_MIMO.lua  and Cascade_Capture.lua

    3. Is the MATLAB script using the default calib matrix here or have you got new measurements for your board?

    We tried both the dummy one and the xx_high.mat one.  We generated the .._high.mat calibration table  by ourselves. Previously, we thought the the issue is caused by the calibration.  Then we turned the calibration off. The post process results for different inputs are still the same.

    for a, b, c, we will try and see the result difference.

    At the same time, do you have any example data to share? We want to double check if the issue comes from the HW or the post process? If it is HW issue, we plan to purchase a second board. Or can you help process one of our data (posted as above) by turning the calibration off (no calibration data needed) and using default configuration and see if you get different results? If the result looks different, then it is my post process issue. 

  • 0 in reply to Chen Zhang
    TI__Mastermind 31680 points

    Hello Chen,

    Chen Zhang said:
    We want to double check if the issue comes from the HW or the post process? If it is HW issue, we plan to purchase a second board

    Yes, this the reason why I requested for experiments a,b,c. This is help us streamline the issue. 

    It really shouldn't be a post processing issue since you're using the default scripts that come with the package and it worked just fine when we tested. 

    I would still recommend going with the experiments first and then we can take a call based on that. I myself have tried running the default scripts several times and it shows object based results.

    Regards,

    Ishita

  • 0 in reply to Chen Zhang
    Prodigy 30 points

    We tried a):

     a) Try capturing data with the test source enabled and see the post processed results after that. 

    Here is how the new result looks: 

    This did LOOK DIFFERENT compared with previous one!

    What does the test source mean? Is it a virtual target? Does the above plot look reasonable?

    we also tried b, but did not find so-called "testSource" script to process it. Can you tell which script I should use? 

  • 0 in reply to Chen Zhang
    TI__Mastermind 31680 points

    Hello Chen,

    Yes, a test source is a virtual target and the output of the post processed script seems correct based on the test source configuration in the MIMO script (if you enable test source).

    For b, you can find the scripts at the below location : 

    ..\mmwave_studio_03_00_00_14\mmWaveStudio\Scripts\Cascade\

    Try to capture a real object (by either putting a corner reflector or a known strong reflector like roof or something) and try to use the inbuilt post processing of mmWave studio to analyze the results. 

    Regards,

    Ishita

  • 0 in reply to Ishita Kochar
    Prodigy 30 points

    The suspected cause is unsuccessful load of configuration. Basically, one needs operate twice otherwise the configuration change will not be loaded. We are now taking more data to ensure there is not other issues.  Does TI provide any demo firmware of cascade AWR2243  for FFT/detection kind of function similar as AWR1843? Currently we can read original ADC data, and the postprocess is through matlab. The data volume is pretty significant if time is long.  

  • 0 in reply to Chen Zhang
    TI__Mastermind 31680 points

    Hello Chen,

    Currently, TI only provides example MATLAB scripts as a part of the mmWave Studio package for MIMO and TXBF usecase processing capabilities. These scripts can act as a base for further development, but the processing of data and algorithm development is mostly upto the user implementation. 

    And yes, the data size can be huge since it corresponds to  four device data and that comes with the usage of cascaded devices. You can decrease the number of ADC samples as per your needs if it is much of a bottleneck.

    Regards, 

    Ishita