IWR6843AOP: Query on Sensing Performance and Simulation for IWR6843AOP in Custom PCB Design

Hi Dileep,

Bandwidth does not have direct impact over the max range, max velocity, velocity resolution, angular range or angular resolution. Field of view also depends on the antenna radiation pattern and is unaffected by chirp bandwidth.

However, it has impact over the range resolution. With 500MHz bandwidth, the range resolution will be 30cm. Hence, it will also have some impact on static presence detection and people counting/tracking. If multiple objects stay less than 30cm apart, they will be considered as a single object.

For a better understanding on the chirp parameters and their impact over radar performance, I would like to refer to the application note to you: 

https://www.ti.com/lit/an/swra553a/swra553a.pdf (link)

Please take a look on the appnote for other detailed considerations.

Thanks,

Swarnendu

Hi Dileep,

I am doing well. Hope you too are doing fine.

It is OKAY to put the AOP device close to the PCB edge on one side. In that case I would recommend having the trapezoidal cut out on the other side of the PCB. I assume the device will not have PCB edges in both the north and south sides. If cut out or PCB edge is not placed in both the sides, ripples could be observed in the antenna radiation pattern in the elevation side as depicted in the device errata.

For radome design, please refer to the below appnote:

https://www.ti.com/lit/an/swra705/swra705.pdf 

Longer components should not be placed in the device side of the PCB. 

Thanks,

Swarnendu.

Dileep,

I would recommend you stick with the provided dimensions in errata for the trapezoidal cut:

These dimensions are optimized through antenna simulations. Also, I would like to recommend the device to PCB edge distance be less than or equal to 0.3mm. This is to prevent higher order radiations as well as to prevent unwanted emissions from other PCB components affecting the antenna performance. The design you provided with one side of the device facing the PCB edge and the other having the trapezoidal cut is alright, given the cut dimensions and the device to PCB edge distances are same as what is provided in the errata.

As an alternate workaround, as you asked, I can suggest using board to board connection. In this, the AOP device can be placed on a different board maintaining the edge to device spacings and taking out important signals out of the board to connect to the main board with other functionalities. This will help addressing area constraints.

However, the proposed design would work fine (i.e. PCB edge at one side and trapezoidal cut out at another).

With the recommended design, we should be able to address the surface wave artifacts. Regarding the antenna field of view, the north and south sides are anyway having no components. Considering the east and west sides, I would like to recommend placing components at least one wavelength away from the device. It is best if we can manage all other components to the other side of the PCB, but a one wavelength space is also fine.

Thanks,

Swarnendu

Hi Dileep,

From the diagram you shared, I can see the radar device placed on a narrow PCB that is allowing the radar device having PCB edges in both North and South sides. I would like to ensure that device edge to PCB edge space is <= 0.3mm. Also, is this a different PCB, as based on the previous discussions we came to a point that one side the device will have PCB edge and the other side a trapezoidal cut?

As far as I can see, the heat sink is below the radar device and not intersecting the radar device plane. In that case it is okay. Even though I would like to ensure that no component should come inside the RADAR FoV.

For the second question, yes there could be non-conducting material. The idea of introducing a trapezoidal cut is to ensure that at least in the marked area, there should not be any PCB to provide conducting surface so that it would interact with the radar performance. Again, I would like to sensitize that any components placed in the vicinity of radar device should not affect its FoV, hence the height of these components need to be checked properly.

Thanks,

Swarnendu

Dileep,

For the enclosure/radome design you can take a look at the below application note:

https://www.ti.com/lit/an/swra705/swra705.pdf 

Yes, as the heat sink lies below the device plane, it should not impact radar performance.

Also, integrating Wi-Fi should not interfere with the radar performance.

Thanks,

Swarnendu. 

Hi Dileep,

I would like to refer the below documentations for the antenna radiation patterns:

IWR6843AOP 

https://www.ti.com/lit/ds/symlink/iwr6843aop.pdf (Device data sheet - Section: RF Specification)

Regarding the surface wave artifacts, the optimized workarounds are mentioned in the device errata. Those should be well enough to deal with the same.

FoV is determined based on the antenna radiation pattern. The considerations are keeping the radar device at center. In both azimuth and elevation sides, the field of view is -60 deg to +60 deg.

Thanks,

Swarnendu

Hi Dileep,

1. The operation of each individual subsystems such as DSP/Processing highly depends on device configurations especially sampling rate, #Chirps, duty cycle etc. Hence it is difficult to standardize power dissipation for each of these subsystems.

2. Please refer to the application note: Thermal Design Guide for Antenna on Package mmWave Sensor (https://www.ti.com/lit/an/swra672/swra672.pdf)

You can use below function or API – to read the junction temperature.

mmwaveLink_getTemperatureReport()  

rlRfGetTemperatureReport()

3. If junction temperature is maintained below 105 deg C, below is the power on hours for the mmWave device.

You can check using the other knobs to reduce the temperature (Such as duty-cycle, Heat-sink options etc).

There should not be performance impact for frequent power cycles. But in longer terms it might stress out the PDN.

Thanks,

Swarnendu

Dileep,

Please find below my responses:

1. We do not have any comprehensive power numbers for the device idle state (i.e. front end inactive) because this is highly user specific. It depends on how many communication interfaces are active, what are the supplies to the device etc. However, the digital leakage power can be expected to be ~20mW.

2. There are many configuration parameters that the average power dissipation depends on. Definition of max. duty cycle also varies based on different regulations. To give you an estimate I am providing below some numbers. However, these will not be exactly specific.

2.1 Only 1 TX, High duty cycle: 1.7W

2.2 Only 2 TX, High duty cycle: 1.9W

2.3 All 3 TX, High duty cycle: 2.2W

3. I am stating below the improvements you will observe if more #TXs are used:

a) The transmitted power will be more.

b) #Virtual antenna array will increase with each TX antenna by a multiplication of #RXs, which will increase the angular resolution.

c) The overall noise floor will reduce.

Similarly with lesser TXs you will observe some degradation on the said factors.

You can further explore IWR6843AOP low power modes in the below resources:

https://www.ti.com/lit/an/swra689/swra689.pdf?ts=1734701696344&ref_url=https%253A%252F%252Fwww.google.com%252F 

https://dev.ti.com/tirex/explore/node?node=A__ALuG45QtveCbcI.ZvVcCOw__com.ti.mmwave_industrial_toolbox__VLyFKFf__LATEST 

Thanks,

Swarnendu

Hi Dileep,

Please refer to the below table note (Table 8-3. Maximum Current Ratings at Power Terminals) for the simultaneous 3 transmitter use case effects on power consumption:

There is more than one factor that could be affected for any change in the configuration. To verify the changes to examine other places that it can impact please refer to mmWave sensing estimator (mmWaveSensingEstimator).

Thanks,

Swarnendu

Hi Dileep,

You can measure the device level power consumption replacing the L11, L12, L13 and L14 with current sense resister. 

TCP202 current probe can be used. For further details on measurements please refer to the application note: xWR6843 Power Optimization (https://www.ti.com/lit/an/swra689/swra689.pdf) Section 5.1 Power Measurement Method.

For the software controls of idle power cycle please refer to the web page in TI resource explorer (68xx Low Power Demo User Guide).

Thanks and regards,

Swarnendu

Hi Dileep,

We can calculate the RCS (Radar Cross Section) of the corner reflector and calculate the SNR using the below formula:

Please refer to the below appnote:

(https://www.ti.com/lit/an/swra553a/swra553a.pdf)

Receiver EINF (effective isotropic noise figure) is 9dB. 

dBFS is dB with respect to full scale. It is a comparative measurement unit, and it is not for absolute measurement purpose. Hence, it cannot be converted to other absolute units. dBFS is used to estimate the comparative response between different factors (for example: FFT response of radar ADC data. The purpose is to identify the peaks with respect to the noise floor. Here, absolute signal strength is not required.)

In order to test the angle estimation accuracy, raw ADC data can be captured and post processed to get the radiation pattern by taking data all across the FoV. The angle accuracy can be determined in the post processing. You can get the reference matlab codes for post processing in the MMWAVE STUDIO download.

Thanks,

Swarnendu.

Hi Dileep,

You would require capturing the raw ADC data from the device. For this you would also need the DCA1000 capture card. The captured data can be post processed to get the plots.

To capture raw ADC data from the device, I would like you to refer the IWR6843AOP EVM user guide (https://www.ti.com/lit/ug/swru546e/swru546e.pdf) section - 4.8.2 DCA1000EVM Mode. In this section the guidelines to connect the device to the DCA1000 is mentioned. You can connect the device as per MMWAVE-STUDIO user guide to make the set-up ready for data capture.

I would like you to download the MMWAVE-STUDIO, from the product page (IWR6843AOP data sheet, product information and support | TI.com) in the Software development foil.

In the MMWAVE-STUDIO download you will also find the Matlab scripts to generate these plots.

To list it down, below are the tools you would require:

1. DCA1000 capture card

2. MMWAVE-STUDIO

3. Post process scripts (Matlab)

Also, to capture data across the azimuth and elevation FoV, additionally you would require:

4. Corner reflector

5. Anechoic chamber

6. A gimbal for device rotation to cover the FoV sweep.

Thanks,

Swarnendu

Dileep,

I looked at your calculations. You do not have to consider the antenna gains as the EIRP and EINF numbers already include the respective antenna gains for the AOP device. 

For the temperature you can read the temp sensor values from the EVM. However, it is okay to use the ambient temperature for the calculations.

Thanks,

Swarnendu

Hi Dileep,

Just use the EIRP and EINF numbers as is of the device from the data sheet. That includes the transmitter, and receiver gains as well as losses. Hence, you do not require to use the antenna gains separately.

If you would like to calculate the dBFs values, please refer to the below documentation:

4555.RadarStudio_matlab_scaling.pdf

Thanks,

Swarnendu

Dileep,

You can explore the low power modes in thebelow resources:

https://www.ti.com/lit/an/swra689/swra689.pdf?ts=1734701696344&ref_url=https%253A%252F%252Fwww.google.com%252F 

https://dev.ti.com/tirex/explore/node?node=A__ALuG45QtveCbcI.ZvVcCOw__com.ti.mmwave_industrial_toolbox__VLyFKFf__LATEST 

Thanks,

Swranendu

Hi Dileep,

1. I would like you to refer to the document: Self-Calibration in TI’s mmWave Radar Devices (https://www.ti.com/lit/an/spracf4c/spracf4c.pdf) to understand the importance of calibration.

Calibration ensures that the performance of the radar front end is maintained across temperature and process variation. Self-calibrations or runtime or boot calibrations are performed once after the device boots up or periodically after the device boots up. IWR6843AOP includes an internal processor and hardware architecture to enable self-calibration.

2. Certain calibrations, including Rx gain boot time calibration, Rx IQ mismatch boot time calibration, and phase shifter calibration can potentially be impacted if there is an in-band interference during the period of measurement. These are executed only during the Rfinit (boot time). To avoid any interference caused corruption it is important to perform these calibrations only at the customer factory in an interference-free environment and use the device calibration data save and restore APIs to inject that information back to the device in the interference-prone in-field operation.

Factory calibrations are done once, and the calibration data is restored afterwards. Antenna Beam Tilt Measurement, Range Bias and Receiver Channel Gain/Offset Compensation are examples of factory calibration.

3. Calibration can be enabled by CLIs. Please refer to the SDK user guide to point to the exact CLIs.

4. As mentioned in mmWave Production Testing Overview (https://www.ti.com/lit/an/spracx7/spracx7.pdf), The user is expected to design the test software and determine the appropriate limits for any tests based on the application of the radar. Also,

The following software programs are required as part of the setup:

• Radar Test GUI: Application software to program the Radar device and analyze the data. This is a custom GUI specific to a sensor. To program and analyze data radar studio can be used.

• Gimbal app: Application software to control the turn-table.

5. Please refer to the Self-Calibration in TI’s mmWave Radar Devices (https://www.ti.com/lit/an/spracf4c/spracf4c.pdf) to understand the importance of calibration.

Thanks,

Swarnendu

Dileep,

Dileep Sharma said:

Is the self-calibration process implemented at the binary level?

Yes. If you want to control different calibrations, you can rebuild the binary after updating the source code.

Some of the runtime calibration happens by default for example temperature calibration or some of the front-end calibrations such as APLL and Synthesizer VCO calibrations are always enabled at boot time and at runtime and cannot be disabled. 

All calibrations can be performed at the RF initialization phase (when the RfInit() API is called during the power up), and some can also be carried out at runtime. The user can select the calibrations to be executed at RfInit and runtime using the AWR_RF_INIT_CALIBRATION_CONF_SB API (to be called before RfInit API) and the AWR_RUN_TIME_CALIBRATION_CONF_AND_TRIGGER_SB API. 

Factory calibrations can be controlled through CLI commands.

Thanks,

Swarnendu

Dileep, 

You can start with the scripts in the studio download (below location):

\..\ti\mmwave_studio_0x_0x_0x_00\mmWaveStudio\MatlabExamples\singlechip_raw_data_reader_example

With this script (rawDataReader) you will have the peaks collected. Based on the data collection parameters you can plot the peaks across angles to generate the radiation pattern.

Thanks,

Swarnendu