• State
  • Locked Locked
  • Replies 7 replies
  • Subscribers 30 subscribers
  • Views 477 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: Distorted IF waveform on custom cascade radar

Li Niu
Prodigy 131 points
Part Number: AWR2243
Other Parts Discussed in Thread: MMWCAS-DSP-EVM, MMWCAS-RF-EVM

Hi Radar experts:

The ADC data we got (by reverse FFT the no-window-R-D map data) from the our AWR2243P's IF waveform output is distorted / with noises

Here is a comparison between TI-dev boards (MMWCAS-RF-EVM + MMWCAS-DSP-EVM) & Our custom designed 4-chip cascade radar board 

We captured the data in an empty parking lot with a Corner reflector (>100m^s RCS) a few meters away.(so there's an dominating sine wave)

We tried a lot of things but didn't help:

1. Tried using external power supply

2. We should have no-worse PI design for the AWR2243s

3. Used 30dB RX gain (was 50dB)

4. The 40MHz clock for all 4 2243 are good.

Our board is very similar to TI's reference design except some power supply ICs, Antenna designs, antenna feeding transmission line and LO signal distribution.

Can someone please shine some lights on what else we could try?

Some more IF waveform from our radar board:

Thanks!!

Li

  • 0
    TI__Genius 13805 points

    HI,

                 I would recommend to have a common profile being run on the cascade EVM and your board and compare the graphs.

    There could be several reasons for distorted waveforms. Especially with you changing some critical sections of design like supply, antenna, etc.

    Regards

  • 0 in reply to Abhed Misra
    Prodigy 131 points

    Hi Abhed:

    Thanks for your reply!

    Yes, for the comparison graphs on the top, we did have both TI's reference board and our board running the identical chirp configuration and frames.

    We thought about antenna and supplies.... For the our changed antenna design (still a board surface patch antenna), it seems unlikely to introduce higher harmonics since it's a linear system. For the power supply, we tried to cut out the DCDC from the board and feed the AWR2243s with external 1V 1.2V & 1.8V, the distortion persist. We also checked the local decoupling design carefully and was not convinced it's the issue....

    One interesting thing we found was that with TI's reference design, some virtual channel had similar "distortion" as well (not as server though)

    This is the 12 TXs & RX1 with a corner reflector at ~4.5m.

    We understand that the distortion / SNR issue could of many causes, but any suggestions or hints would be really appreciated! (we've been working hard on this for more than a month now...)

    Thanks!

    Li

  • 0 in reply to Li Niu
    TI__Genius 13805 points

    Hi Li,

           When you say you have used the reference design except the Antenna and supply, does that mean 20GHz LO topology is same as reference design?

    If yes did you check that section also?

    Regards

  • 0 in reply to Abhed Misra
    Prodigy 131 points

    Hi Abhed:

    Good question (I should have mentioned), yes the LO topology is the same, we used both FM_CW_CLKOUT & FM_CW_SYNCOUT on the master AWR and they further branched into 4 LO signals feeding all 4 AWR2243s. We used the same wilkinson divider as the reference design.

        

  • 0 in reply to Li Niu
    TI__Genius 13805 points

    Hi,

                  Did you simulate these splitters as from the look they appear to be improper and definitely different then reference design? BTW was the layout also reviewed by TI? 

    Regards

  • 0 in reply to Abhed Misra
    Prodigy 131 points

    Hi Abhed:

    We didn't have our designed reviewed by TI (We sent the files but didn't hear back...). 

    It is true that they are different but I felt the essence of the Wilkinson divider is honored: all the impedances, a lot symmetrical GND vias, 0.25-wavelength trace between 100Ohm and the split points etc. 

    Even our design is inferior, I don't feel like it will introduce harmonics and higher frequency components (due to the fact that it's a linear system) as we observed, how do you think?

    Li

  • 0 in reply to Li Niu
    TI__Genius 13805 points

    Hi,

            I didn't mean to say that your design is inferior if not reviewed and apologies if I sounded so.

    I believe if the sims have been done before fabrication to the expected performance, it is important that the fabrication has been done with due diligence also. Has your fab vendor been doing RADR boards for you earlier?

    Regards