IWR6843: Determining valid chirp parameters for SDK Demo

Hi,

With regards to the MMWAVE-STUDIO installation errors, have you tried installing MMWAVE-STUDIO on a different machine? Also, make sure that you're installing Matlab 8.5.1 runtime 32 bit version as mentioned in the user guide.

1. You can also use the mmWaveSensingEstimator which is a browser based tool that allows you to calculate the chirp parameters based on the application level requirements such as Max Range, Max Velocity, Velocity resolution, RCS etc.
2. The internal calculations of the MMWAVE-STUDIO Ramp Timing Calculator are not exposed. However, the FMCW equations and Radar front-end constraints used by the Ramp timing calculator as well as the mmWaveSensingEstimator are mentioned in this app note:  Programming Chirp Parameters in TI Radar Devices.
3. Please note that the mmwavelink and HWA are two separate entities. mmWaveLink is the interface with the Radar Front-End gets configured with the chirp configuration. A profile/chirp/frame configuration is valid from the front end perspective if it doesn't generate an mmWaveLink error. HWA or the Radar Hardware Accelerator on the other hand is the digital processing engine for processing the ADC data coming from the front-end. Please refer to the HWA user's guide for more details: http://www.ti.com/lit/pdf/swru526. You can also look at the Radar Accelerator Training Video for an overview.
4. Note that we have not yet RTM'ed the 6843 device and the current 6843 silicon is engineering sample (ES1.0). The mmWaveLink code in the SDK is the interface towards the peer mmWaveLink which is part of the radarss firmware. The production radarss firmware as well as the mmWaveLink SDK code will be available with the production version of the 6843 device.

Please mark this thread resolved if your question was answered.

Regards

-Nitin

Nitin --

Thanks for the reply and we will try installing mmWave Studio on another machine.  We are familiar with both the mmWave Sensing Estimator and the app note.  

Here's an example of our problem.  We want to develop a profile with (among other things) maximum radial velocity.

1.  In the mmWave Sensing Estimator, select IWR6843 and Short Range Default.
2.  Change Velocity Resolution to .5 km/h.  This gives an error message, "radar cube size is larger than available memory".
3.  Change Velocity Resolution to  1.0 km/h.  No error message.
4.  Enter the resulting parameters into the configuration file.  In this example, C:\ti\mmwave_sdk_03_00_00_08\packages\ti\demo\xwr68xx\mmw\profiles\profile_3d.cfg.  So:

profileCfg 0 60 7 5.9 49.15 0 0 71.26 1 223 5279 0 0 30
frameCfg 0 2 54 0 100 1 0
etc.

For startFreq, 60 GHz fails in the IWR6843 SDK Demo.  We find we must use 60.25.

For adcStartTime, we find that 5.9, 5.0, 4.0, 3.75, 3.6 all fail.  3.0 and 3.5 work, no difference in the memory stats that the SDK Demo displays in the CCS debugger.

For number of loops in the frameCfg, 32 gives a best radial velocity value (empirically, from reading COM prot data stream) of 0.8366 m/s.  Number of loops values of 40, 48, 56 and 60 all give radial velocity values of 0.4183 m/s.  A value of 64 loops fails with error -40112, DPM report 4.

The way to find the performance limits and characterize the device have been trial and error. Using mmWave Sensing Estimator gives values that appear to be illegal when used in the SDK Demo.

Next steps are to try getting the Radar API running -- maybe that will give different configuration values we can use -- and to start digging through the SDK code, to understand what is causing the errors we see with the values provided by the mmWave Sensing Estimator.  That's a lot of work while we are still just trying to evaluate the device suitability for our application.  Maybe you can bring some clarity to this and shorten our path.

Dan Lewis

Sensing estimator values:

Example error from CCS.  Fails right after sensorStart command.

We set up a VM over the weekend and tried again to install mmWave Studio.  We successfully installed the mmwave_studio_02_00_00_02_win32 executable, the FTDI drivers, the MS C++ runtime, and the MatLab runtime.  Running the mmWave Studio executable as administrator, we get a "Error registering Lua functions" error + exception, see below. 

Again, our reason for installing mmWave Studio is to gain access to the Radar API calculator, in the hope that it will provide useful configuration values for the mmWave SDK 3.00.00.08.  We don't need mmWave Studio for data acquisition, at least for now.

Our goal is to be able to determine valid parameters for the .CFG file, which we cannot do with the sensing estimator. 

Thanks
Dan Lewis

Image was not attached to the last post, here it is:

Hi Dan,

We have not tested MMWAVE-STUDIO on a VM. I would request you to install it on a native Windows machine along-with the required tools mentioned in the user guide for MMWAVE-STUDIO 2.0.0.2. You need to install only MMWAVE-STUDIO and the Matlab Runtime and additionally Visual C++ 2013 Redistributable if installing on Windows 10.

Regards
-Nitin

Thanks, Nitin. Doesn't seem like a VM issue but we will try installing mmWave Studio again on a Windows 10 machine. Meanwhile, I'll try restating our original question.

Why does the mmWave Sensing Estimator give chirp parameters that cause the IWR6843 SDK Demo to halt with an error?

Is there a tool that provides valid configuration parameters for chirp, frame, and profile for the IWR6843 SDK Demo?

Thanks
Dan Lewis