IWR1642BOOST: Observed Max range of 25m - how to improve?

Hello Anton,

Please refer to these other threads related to maximum range:
e2e.ti.com/.../625168
e2e.ti.com/.../2261030

Amanda

Hello, Amanda.

Thanks for the two links, but please can you explain it a little further (for someone new in this).

In the first thread, is it confirmed that the max range limit for pedestrians of iwr16xxx is 30m - "Using the AWR1642 EVM , we've been able to detect moving people upto 30 m."?

I tried to make a quick estimation with the given in the second link - e2e.ti.com/.../2261030 equation after removing B. In my case NChirps is maybe different, but with the given there radar params, the result I got is:

σ  = 0.5 sq,m,   dmax = 28.80 m

σ  = 1.0 sq,m,   dmax = 34.26 m

σ  = 5.0 sq.m.   dmax = 51.23 m

I am not sure how correct are these results, here are my detailed calculations:

Pt = 12 dBm =  0.0158W
Gtx = Grx = 9 db = 7.94
wavelength = 3.95mm
SNRmin = 12 dB = 15.85
Temp = 294K
k = 1.38064852e-23
F = 16 dB = 39.81

Tmeas = number of chirps * Chirp duration = 65 * 55us = 0.003575

dmax =    sqrt(4) (σ  * 0.0158 * 7.94 * 7.94 * 0,00395 * 0.00395 * 0.003575) / ( (4 * 3.14159)^3 * 15.85 * 294 * 1.38064852e-23 * 39.81) ))

I noticed that "Maximum Range for desired RCS (m)"  field in mmWave Demo Visualizer shows higher dmax values:

σ  = 1.0 sq,m,   dmax = 42.60 m

σ  = 0.5 sq,m,   dmax = 35.82 m

σ  = 5.0 sq.m.   dmax = 63.70 m

But what I experimentally see matches rather to the lower values (calculated by Eq. 1).

If I understood right these dmax results are hardware limits of the iwr1642 ENV? So for higher ranges it would be necessary changes in the board design?

But if so I do not understand how IWR1642BOOST ENV have been used in TIDEP-0094 for much larger ranges (50-80 m).

I am estimating the possible usage of IWR1642BOOST (out of the box) for both pedestrians and cars detection applications, in the the range (20-80m). I wanted test with Visualizer Demo  because it uses the demo app from the SDK.

Thanks.

Hello, Amanda.

I am reposting my latest question - from the Equation-1 of the second link you provided I got:

σ  = 0.5 sq,m,   dmax = 28.80 m
σ  = 1.0 sq,m,   dmax = 34.26 m
σ  = 5.0 sq.m.   dmax = 51.23 m

Detailed calculation:

Pt = 12 dBm =  0.0158W
Gtx = Grx = 9 db = 7.94
wavelength = 3.95mm
SNRmin = 12 dB = 15.85
Temp = 294K
k = 1.38064852e-23
F = 16 dB = 39.81
Tmeas = number of chirps * Chirp duration = 65 * 55us = 0.003575
dmax =    sqrt(4) (σ  * 0.0158 * 7.94 * 7.94 * 0,00395 * 0.00395 * 0.003575) / ( (4 * 3.14159)^3 * 15.85 * 294 * 1.38064852e-23 * 39.81) ))

Calculated with the mmWave Demo Visualizer "Maximum Range for desired RCS (m)"  field shows however higher dmax values:

σ  = 0.5 sq,m,   dmax = 35.82 m

σ  = 1.0 sq,m,   dmax = 42.60 m

σ  = 5.0 sq.m.   dmax = 63.70 m

Which of these estimations are more correct for the IWR1642BOOST ENV (out of the box)? Can these max. ranges  be improved with this device so to reach 50m for pedestrians and 80 m. for cars? 

I am able to detect with the Demo-Visualizer application (X-Y Scatter Plot) the 0.5 sq. m. sized objects for 25m Max. Range.  Am I doing something wrong?

Thanks!

Hello, Amanda.

Thanks about detailed explanation.

I understand that the chirp slope S limits dmax according to the first equation in e2e.ti.com/.../2261030. Then the smaller of the two limits of dmax (slope or RCS) is taken. In my particular experiment I have chosen in Visualizer the "Best Range" Configuration, I can see that "Maximum Unambiguous Range" is set to 50 m. Visualizer generates CFG files that contain the following profileCfg like :

profileCfg 0 77 7 7 108 0 0 15 1 624 6250 0 0 30

I can see that S is about 15 MHz/usec and ADC sampling frequency big enough to allows the 50m range. I played with the other parameters - Velocity and Range resolutions, Vmax and FPS but these seems not influencing dmax much.

I am trying to measure the limits and the performance of the device for both pedestrians and cars. For cars I also want to similar tests but there dmax is not so easy to measure.

In Range Profile Plot, at 10-15m distance to target, I can see well formed and visible peaks sometimes 20-30 db above the noise floor. But at 30m and beyond results are unstable, peaks are detected only at certain angles, and sometimes are hardly discernible, just 3-4 db above the noise floor.

I lowered Range and Doppler CFAR to 10 db.

I uploaded the following video with one of my tests: https://youtu.be/BYttCo4Ub3A. The web camera is placed about 10cm left from the radar. Radar is positioned 1m above ground.

Should I expect big difference if the radar is placed higher above the target, lets say 3 m above the ground? What possible ways there is to "increase the SNR before the object could be detected"?

Thanks.

Also your instinct is correct for increasing SNR. If we reference the radar range equation in the chirp parameters document you can see the relevant parameters for affecting change in SNR. Aside from the parameters governed by the chirp design note that antenna gain is also relevant. One way for increasing the TX gain is to design a different antenna pattern (generally you trade off field of view for increasing gain). For RX antenna gain you can increase this from default by modifying the last parameter of the profileCfg (default is 30 in chirp cfgs generated by the online visualizer - up to 48 is supported. Please see the SDK user's guide for more details).

With regards to sensor orientation and max range the primary thing to consider would be how the reflective surface area of the desired target and therefore RCS changes with respect to sensor orientation changes. For example if you are mounted at 1m straight ahead then you're looking at the front hood/bumper of a car but if you're mounted above tilted looking down you will reflect off the top of the vehicle for it's entire length - giving a higher RCS.

Amanda

Hello, Amanda.

Thanks a lot for the suggestions. I need to repeat the tests and to try the People Counting demo and the Chirp Configurations from the link you provided.

Only one quick question meanwhile - from the document http://www.ti.com/lit/an/swra593/swra593.pdf:

Object Versus Range

Table 1 lists objects versus range in meters - reported in the table ranges are much more than what I observed with IWR1642BOOST. For example human is detected at 40m, motorcycle at 60m and cars at 120m. At the next page of the same document - Measurement Protocol, I can see that the test methodology is using again the Demo Visualizer, however static objects are detected while I need to detect static and slowly moving objects:

"For data collection, an IWR1443 EVM running the mmWave SDK out-of-box demo was used with a Windows laptop running the mmWave Demo Visualizer. Further details related to using the mmWave Demo Visualizer can be found in the user’s guide for mmWave Demo Visualizer. Objects were considered to be detected at the specified distance if the visualizer returned a detected object marker at the range of interest."

Anyway these range results are pretty much what is required, but the tests are done with IWR1443BOOST. Comparing to my device IWR1443 operates again in the 76-81 GHz band but has 3 TX antennas, better gain and better antennas design. Should I expect big difference in ranges and performance between IWR1443 and IWR1642?

Thanks.

Hello Anton,

There should not be big differences between the 1443 vs 1642. The difference in the devices come to having a HWA in the 1443 while 1642 has a DSP and more memory.

Amanda