IWR1443BOOST: Odd Range Doppler Map

John Kablan

Part Number: IWR1443BOOST
Other Parts Discussed in Thread: IWR1443, IWR1642, UNIFLASH

Hi I'm trying to do some very basic tests with my IWR1443 / DCA1000 and the data I'm getting from the radar doesn’t seem to be very good quality.

I'm interested in a shorter range application so I've increased the chip frequency slope and decreased the ramp end time. The following are my programmed parameters.

The data was recorded indoors. I have a flat metal plate that I held in front of myself. And walked toward / away from the radar. The space is somewhat cluttered but there are no items that move in the office e.g. fans.

The following is the range doppler map of the first frame. Note there's nothing moving in front of the radar. The "shape" of the range doppler map looks odd to me. I would expect a clutter ridge to run from 0 range to positive range, instead it seems as if there is a clutter ridge that runs along . I've taken these screenshots from mmWave Studio so I don’t think this isn't an issue with the IQ matrix being transposed.

I also noticed that there are some large spikes in the adc samples (see approx 1.8e-5 s below).

If I replace these spikes with the last valid sample (as determined by a threshold) I get a cleaner looking range doppler map but its still odd looking since the clutter ridge extends across the doppler dimension. I've tried applying a Hanning window across the Doppler dimension but that doesn’t have an appreciable effect on "tightening" the "Doppler clutter ridge"

When I play the frames back I can see what appears to be my plate moving back and forth in the range dimension. The following map shows the plate(see the line horizontal line around range bin ~110) but the is spread across the Doppler dimension.

So my questions are as follows

Are the spikes in the ADC data expected or indicative of something wrong with my configuration of the radar?
Why do I have a clutter ridge that lies along the Doppler dimension of my range doppler map? And why is my target energy not better concentrated in the doppler dimension?
Why do I get so much energy in the range image bands (negative frequencies ). I receive the data in the complex2x format so I would assume that this wouldn’t be the case. Does this possibly have something to do with ambiguous returns (returns from previous chirps)

Thanks,

John

over 6 years ago

0 Radio-Joe over 6 years ago

TI__Mastermind 37500 points

Hello,

I have not done the 2D Doppler experiments, but I have forwarded your email to other engineers in the group.

a) Normally I use Complex 1x for measurements, and only use Complex 2x if I am performing interference mitigation. We do not expect negative

frequencies from properly designed chirps.

b) I would use the Sensing Estimator tool, to design a profile, chirp, and frame. The 10Mhz IF Rate, 12.5Msps DFE output rate, can be used on an IWR1443. The IWR1642, needs to be limited to 6.25Msps.

You could also use the USB/UART with the Visualizer, and Out of Box demo. The purpose is to experiment first with the Out of Box demo, and then refine your example using the Mmwave studio.

I reprogrammed the IWR1443 with BSS and MSS software using Uniflash.

This is used with TDM MIMO 2Tx one at a time, and 4Rx.

I have used the Mmwave SDK Out of Box demo, and the Visualizer 2.1. This is used to get the general parameters for your test, I used the Best Velocity Resolution example, under the Visualizer, see the attached .cfg (you can load this, need frame rate to be 4fps or less)

I used this to show velocity waving my arm in front of the sensor.

1443_velocity_visualizer.cfg

Once this works for you, it could be transfered (profile, chirp, frame), suggest using complex 1x data mode.