• State TI Thinks Resolved
  • Locked Locked
  • Replies 14 replies
  • Answers 3 answers
  • Subscribers 31 subscribers
  • Views 2512 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.

IWR1642BOOST: Misaligned Antenna Beam

Boaz Menis
Intellectual 510 points
Part Number: IWR1642BOOST
Other Parts Discussed in Thread: IWR1642

Hi,

we have tested some new IWR1642 BOOST radars with the out of the box demo

using the mmWave Demo Visualizer, working with the demo bin file of SDK 1.1.0.2

We have seen a strange phenomena with some of the radars where the main lobe is directed

upwards at ~20 [degrees] (tilted upwards in elevation).

We suspect that there is some kind of antenna calibration procedure that should have been performed and 

haven't.

We work in a mode of 2 Tx and 4 Rx. The configuration is attached.

We did 6 experiments:

1. Mounting a corner reflector of RCS = 1.5 [m^2] in a distance of 1.3 [m] and angle 0 ("Nominal Mode") - the visualizer did not detect the corner reflector.

2. We then  tilted the radar at -20 [deg] in elevation  - the reflector was detected with very high peak.

3. We tilted the radar back to elevation angle = 0 but this time mounted the corner reflector at the height of 45 [cm] above the radar - the corner is detected with high peak.

4. Working in a mode of 1 Tx and 1 Rx in "Nominal Mode"- the corner is detected.

5. Working in a mode of 1 Tx and 2 Rx in "Nominal Mode"- the corner is detected.

6. Working in a mode of 1 Tx and 4 Rx in "Nominal Mode"- the corner is sometimes detected and sometimes not.

We want to emphasize that we have purchased a few IWR1642 in the past and haven't seen such a phenomena.

Do you know what is the reason of such phenomena?

We also tried to use the option of "compRangeBiasAndRxChanPhase" but it didn't help.

We will be glad for some help with this issue.

Thanks,

Boaz

The configuration:

sensorStop
flushCfg
dfeDataOutputMode 1
channelCfg 15 3 0
adcCfg 2 1
adcbufCfg -1 0 0 1 0
profileCfg 0 77 69 7 28.55 0 0 50 1 128 6250 0 0 30
chirpCfg 0 0 0 0 0 0 0 1
chirpCfg 1 1 0 0 0 0 0 2
frameCfg 0 1 128 0 100 1 0
guiMonitor -1 1 1 0 0 0 1
cfarCfg -1 0 0 8 4 4 0 3000
cfarCfg -1 1 0 8 4 4 0 3000
peakGrouping -1 1 0 0 1 127
multiObjBeamForming -1 1 0.5
clutterRemoval -1 0
calibDcRangeSig -1 1 -5 4 256
extendedMaxVelocity -1 0
compRangeBiasAndRxChanPhase 0.0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0
measureRangeBiasAndRxChanPhase 0 1.26 0.3
sensorStart

  • 0
    Intellectual 510 points
    Hi,
    Just saw in the forum that someone has the exact same problem,
    but the issue wasn't resolved.
    e2e.ti.com/.../661927
    Boaz
  • 0 in reply to Boaz Menis
    TI__Mastermind 37500 points

    Hello,

    We have asked the antenna designer.    The antenna azimuth boresight is based on the frequency of the chirp.  In the range of 78.5Ghz center frequency, the azimuth boresight is estimated at 5 degrees.   At 81Ghz the elevation boresight is ~17 degrees.

    It is also possible that from old to new IWr1642 EVMs, that there is some tolerance difference between the L1 antenna pad structure, and the L2 ground structure.    

    If you look at swru521a, the Antenna figure for elevation shows 0 degree.   The Antenna designer has output these two graphs at 78.5Ghz 5-7degree elevation, 81Ghz 15-17 degree elevation.

      a) suggest you rerun your test with mmWave SDK 1.0 - to check if its a software concern

      b) if you have an old EVM, run with mmWave SDK 1.1 - this would also check hardware/ software (I only have a 6mo old version)

      c) change the Chirp so that you can isolate the frequency band (see if starting at 77 vs 79 shows a different behavior

           profileCfg - startfreq 77Ghz

                                   Freqslope - 28.55

                                   ramp_end_time - 50

                                   endfreq = startfreq + (Freqslope * ramp_end_time)

                                    freqcenter = avg(startfreq, endfreq)  77.7 Ghz

      d) please send us the board serial numbers, we will see if we can find one in this block.  So we can test it.

      Note: in the newest antenna graph the antenna has a higher sensitivity at 15-17 degrees, but at higher frequency ~80Ghz.

    Regards,

    Joe Quintal

  • 0 in reply to Radio-Joe
    Intellectual 510 points

    Hi Joe,

    Thanks for your reply.

    I am not sure I quite understand.

    I'm already working in the 77-78 band, which according to your graphs is the "good band".

    I also don't understand why moving to an older SDK could resolve my problem, it looks like an HW issue.

    In addition, 17 [deg] is a very high tilt to work with and will not meet our specifications.

    I have seen this issue with 2 EVMs, but I haven't tested all the other units yet.

    The serial numbers of these 2 EVMs I have tested are:

    4765500192

    4765500035

    Thanks,

    Boaz

  • 0 in reply to Boaz Menis
    TI__Mastermind 37500 points

    Hello,

    I was trying to isolate the problem to EVM hardware, if your old batch of EVMs was tested with mmWave SDK 1.1 that is the information needed.   There are new software commands that affect the gain and phase of the receivers in mmWave SDK 1.1.  

    If this occurs on 2 EVMs out of a batch of 'n' EVMs,  Please inspect the EVMs for no solder mask or foreign material covering the RF traces from the device to the antenna.   In some older EVMs, we had seen the RF trace extensions near the patch antenna to delaminate, and have more trace bending, this affects the antenna performance.

    You can isolate the boards that are not working and arrange to send them to Dallas for rework / replacement.   If you can tell us from your batch the number of boards with different antenna profiles than your experience we will use that in evaluating the boards.

    Regards,

    Joe Quintal

  • 0 in reply to Radio-Joe
    Intellectual 510 points
    Hi Joe,
    I will test the other units on Sunday-Monday
    and will update you.
    Will also test them under SDK 1.1.0.2, 1.1.0.0 and 1.0.0.5.
    Thanks,
    Boaz
  • 0 in reply to Boaz Menis
    Intellectual 510 points

    Hi Joe,

    I have placed a corner reflector in a distance of 1 [m] from the radar at

    the same height and then rotated the radar in the elevation angle phi, and 

    documented the received power in dB as shown in the "mmWave Demo Visualizer".

    This is the results I get:

    As you can see, we lose around 20 [dB] when mounting the radar at phi=0 because of this radiation pattern.

    Can you explain this?

    Is it normal?

    Thanks,

    Boaz

  • 0 in reply to Boaz Menis
    TI__Mastermind 37500 points
    Hello,
    We don't currently have antenna measurements of the EVM. We have simulations. The simulations at 78.5Ghz do not show this offset. (see previous graph - shows 5 deg), The simulations at 81Ghz show the two peaks, but the null is shifted to -8 degrees. In the simulation graphs these are shown as azimuth changes in Rx.

    Please tell us how many EVMs are measuring this way in your batch, and if the older software has the same effect on 1 bad EVM.
    I will ask, if we have different batch EVMs that could replace the mis-performing ones.

    I will try to get an EVM in your reported SN range, to test. I don't see an elevation affect on my IWR1642 EVM which is about 6-8mo old.
    Regards,
    Joe Quintal
  • 0 in reply to Radio-Joe
    TI__Mastermind 37500 points
    Hello,
    We have found 2 EVMs within the serial number range. They have been tested with a mmwave output ~77-78.5Ghz.
    As you send back your 2 EVMs, these are being sent as replacements.

    Please test the replacement EVMs when you receive them, and let us know we will retest the returned EVMs and update you.
    REgards,
    Joe Quintal
  • 0 in reply to Radio-Joe
    TI__Mastermind 37500 points

    Hello,

    I am including the antenna tests done on the 2 EVMs within the serial number of those reported.

    What is recorded, is the antenna rotator angle, there is a sheet for the azimuth, and elevation.   An RCS target is initially located on the bore sight, the amplitude is recorded for each angle.  

    REgards,

    Joe Quintal5756.AntennaMeasurementCompare.xlsx

  • 0 in reply to Radio-Joe
    TI__Mastermind 37500 points

    Hello,

    If you use Radar Studio which can be downloaded from the Ti website, and you have the Devpack board,   You can use the sample configuration attached, and add one set of manual operations, (RegOps tab) to write a debug value, @0x4320140C, read, modify bits 16.19 with 0x7.  This will activate the 40Mhz crystal clock to the J2 connector, should be 1v amplitude, 40Mhz with 50 ppm.   This can be used to validate the 40Mhz clock.

    we have evaluated a returned EVM with the test procedure described in AntennaMeasurementCompare.xlsx, we are working on the cause of the problem.  

  • 0 in reply to Radio-Joe
    Intellectual 510 points

    Hi Joe,

    Thanks for your answer.

    I will need some description of the setup in order to test the PLL with the radar studio.

    Except the IWR1642 EVM, Do I need?

    1. The DFP board?

    2. The TSW1400 board?

    3. A scope?

    Thanks,

    Boaz

  • 0 in reply to Radio-Joe
    Intellectual 510 points

    Hi Joe,

    After following the instructions above I don't see any signal coming out to the J2 connector.

    I do see the 40 MHz sine wave when I connect directly to the oscillator.

    Boaz

  • 0 in reply to Radio-Joe
    Intellectual 510 points

    Hi Joe.

    I got it.

    The signal looks quite different from yours:

  • 0 in reply to Boaz Menis
    Intellectual 510 points
    Hi Joe,
    From my measurements I get:
    good board - 40 MHz with 12 PPM
    bad board - 40 MHz with 17 PPM
    Boaz