• State TI Thinks Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 2 answers
  • Subscribers 30 subscribers
  • Views 202 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

Original question:

AWR1642BOOST: Corner Reflector- calibration process and its dimensions

AWR1642BOOST: bias between detected-object in demo visualizer and actual position

Woody Boyd
Prodigy 70 points
Part Number: AWR1642BOOST

Hello ,

This is my setup :

- mmw demo firmware with sdk 3.6

- profile_2d .cfg

- mmwave demo visualizer .exe file

my application for distance detection very very short range (less than 1 meter) object. I hope to get milimeter accuracy.

Then, I experienced this situation that there's same bias value between detected-object in demo visualizer and actual position.

For example : distance/range from radar sensor to object from demo visualizer is 39 cm. But actual distance is 30 cm. All experiment show 9 cm gap repeatedly.

I went through doxigen docs from mmw demo, and found this : For most boards, a range bias of few centimeters has been observed. User can estimate the range bias on their board and correct using the calibration procedure described in Range Bias and Rx Channel Gain/Offset Measurement and Compensation.

So, Question is

1. Can I conclude this 7cm gap is calibration requirement ?

2. When I follow that calibration sequence it said it require corner-reflector. What kind of corner-reflector ? Can I use flat surface from aluminium ? or aluminium foil ?

3. I use aluminium foil temporarily, I got compensation values

compRangeBiasAndRxChanPhase 0.0773442 -0.72714 -0.68649 -0.41965 -0.27042 -0.32486 -0.35538 -0.21402 -0.47507 -0.61176 -0.63177 -0.53976 -0.14435 -0.48978 -0.24066 -0.37936 -0.39862
compRangeBiasAndRxChanPhase 0.0773278 -0.81265 -0.58273 -0.42734 -0.22263 -0.36185 -0.32385 -0.27078 -0.43539 -0.65771 -0.52158 -0.54474 -0.09161 -0.50769 -0.17157 -0.40378 -0.34702
compRangeBiasAndRxChanPhase 0.0773362 -0.67667 -0.73630 -0.39194 -0.29092 -0.29755 -0.37305 -0.18750 -0.48502 -0.56705 -0.64749 -0.52231 -0.18570 -0.47070 -0.26773 -0.35141 -0.41080

then I insert to cfg file.

As you can gues ,  bias 7.7 cm. If I use better corner-reflector, bias range should be around 9 cm, correct ?

Thank you for your reply

  • 0
    TI__Guru*** 143714 points

    Hi,

    Hi there are several things to consider

    • Range calibration must be performed to have best accuracy

    A corner reflector is a strong reflector. If a CR is not available one can use a metal plate. Basically we need to create a strong reflection peak at an accurate distance.

    • Range accuracy also depends on the Range FFT size

    Based on profile_2d.cfg, max distance is about 8.64m. The range FFT size is 256 point. This means that the accuracy is about 864/256 = 3cm

    % Max beat freq (80% of ADC sampling rate)  MHz       5.76
    % Max distance (80%)    m                             8.64
    % Range resolution  m                                 0.042
    % Range resolution (meter per 1D-FFT bin)   m/bin     0.042

    % Frame time (total)    ms                            2.944
    % Frame time (active)   ms                            2.496
    % Range FFT size    -                                 256
    % Doppler FFT size  -                                 32

  • 0 in reply to Cesar
    Prodigy 70 points

    Thank you for your reply.

    For profile_2d.cfg I forgot to mention that I use 100 cm. So accuracy is 100/256 is 0.4cm. Correct ?

  • 0 in reply to Cesar
    Prodigy 70 points

    Another related question, during calibration  with mmw demo firmware, I shouldn't have to use DCA1000 to get calibration params correct ?

  • 0 in reply to Woody Boyd
    TI__Guru*** 143714 points

    If you are using the mmWave SDK demo, there is no need to use the DCA1000

    If the profile_2d.cfg was modified for a max range of 100cm, you are correct the accuracy should be 0.4cm

    thank you

    Cesar