AWRL1432: How to calculate the PeakVal?

Hi Jasper,

Yes, the ADC samples are after the LNA in the front-end chain of the 1432 device.

Could you share the source of the DPU image?

Thanks.

Hi Deexith

The DPU image is from TI document, MotionPresenceDetectionDemo_documentation which is in SDK.

Best Regards

Jasper

Hi Jasper,

I'd recommend following the answers in the below threads for more information: 

https://e2e.ti.com/support/sensors-group/sensors/f/sensors-forum/1042060/iwr6843aop-mmwave-radar-chip-and-board-selection-for-water-spray-quality-detection 

https://e2e.ti.com/support/sensors-group/sensors/f/sensors-forum/914685/awr1642-awr-series-received-power-estimation 

https://e2e.ti.com/support/sensors-group/sensors/f/sensors-forum/613223/awr1243-how-to-calculate-the-precise-power-on-the-tx-and-rx-pin 

https://e2e.ti.com/support/sensors-group/sensors/f/sensors-forum/595684/iwr1443-coherent-and-non-coherent-processing-gain-summary 

https://e2e.ti.com/support/rf-microwave-group/rf-microwave/f/rf-microwave-forum/1118993/total-measurement-time-in-radar-equation 

https://e2e.ti.com/support/sensors-group/sensors/f/sensors-forum/595225/range-fft-processing-details 

FFT processing gain arises from the nature of FFT and it increases as the number of bins inceases. or use equation FFT gain = 10*log(Nfft/2) where Nfft is FFT bin size.There is an additional processing gain from the combination of multiple RX antennas. This is an incoherent processing gain as we sum powers across the antennas before detection.

Regards,

Tim

Hi Tim

Thank you for your reply.

If I use the formula to calculate the Rx_power (dBm) equals -30dBm and convert to Voltage by using following equation, is it right?

V=sqrt((10(dBm/10))∗0.001∗R) where R is 50 ohm.

After converting to the Voltage, I use the ADC (12-bit) to calculate the value.

In this project(2T3R), I use the HWA to do the Range_FFT (FFT Size = 64) and Doppler_FFT (FFT Size = 128).

After these calculation, i can get the real peakVal which is shown in RD map, am I right?

Could you help me to calculate the real peakVal in this condition?

Best Regards.

Jasper

Hi Jasper,

Your calculations seem correct to me - the peakVals will then show in the RD heatmap as you explain.

Regards,

Tim

Hi Tim

Could you give me an example (Rx power = -30 dBm) to calculate the real PeakVal?

Best regards.

Jasper

Hi Jasper,

I'm unsure what you mean by example here - your calculations above could be used to devise a hypothetical example.

I recommend you collect some data with our radar modules and a corner reflector so you can experimentally see the peak values for a corner of a known RCS.

Regards,

Tim

Hi Tim

I think you don't solve my question, please don't close the issue by yourself.

Currently, we have measure the RCS by using the corner (RCS=10 dBsm) which is at 2m.

The blue line is range profile with the corner at 2m.

The orange line is range profile without objects.

The peakVal is in dB scale (20*log10()).

Could you answer the question: how do I get this value by using the formula?

Hi,

My apologies - I need to look at it more and talk internally and then I'll provide a response here.

Regards,

Tim

Hi Tim

Any updated?

Jasper

Hi Jasper,

No, sorry - we will take a look at it next week.
Regards,

Tim

Hi, Jasper:

Tim is on vacation, so I can be helping you with this e2e thread.   

Can you share the following excel sheet you have on your original post? Are you have the target at bore sight at 5.856m as shown in the figure below? 

If you have the dbFs, and if you have the ADC data, you should be able to see how well it matches.  

What is your goal of calculating the theoretical peakVal?  To be honest, it is usually hard to match the theoretical results with the measurement results.  It took a lot of effort and you may still find it not so close. 

Best,

Zigang

Hi Zigang

Thank you for your reply.

I put the corner(RCS=10 dBsm) at the 2m and measure the at the above figure. And formula as follows:

As a system engineer, I need to define the spec to make sure the PCBA performance which follow another E2E link:

https://e2e.ti.com/support/sensors-group/sensors/f/sensors-forum/1062043/awr1843-how-to-evaluate-pcb-board-performance-and-problems?tisearch=e2e-sitesearch&keymatch=AWR1843%252525252520PCB#

So we set up the experiment in our chamber to measure the SNR.

Or do you have the best way to evaluate the PCBA performance?

I just want to know what does TI calculate this value in SDK from analog signal to RD map, no matter it is not close to experiment.

Best Regards.

Jasper

HI, Jasper:

If your goal if to compare how your PCB board perform, is it possible to use the same setup, and try TI EVM and your board to see any difference?  Since you have the same chirp, same target (same RCS) at same distance, same process gain.  The only difference should be your antenna gain and trace loss difference. 

Best,

Zigang

Hi Zignag

It is not a good idea because I use the different TI EVM to measure SNR and also get the different results. You may say that is the tolerance of experiment.

Maybe I ask another question you will clear what I want: How does TI define the spec to check the SNR is ok or not? What value can you accept the SNR in TI EVM and why?

Best regards.

Jasper

Hi, Jasper:

That is a good question.  At device level, the team will measure the TX power and measure the noise floor.  At the EVM level, I will need to check whether we do any test to check TX antenna gain or RX antenna gain.   Will that information be helpful?

Best,

Zigang

I am forwarding this e2e thread to our PCB design team to help.

Best,

Zigang 

Hi Zigang

I prefer not only how to check T/Rx antenna gain but also to know how to get the peakVal in TI SDK.

Best regards.

Jasper

Hi Jasper,
First of all, SNR performance of the design and its measurement depends on lot of factors like TX/RX antenna gain, RCS of the object, path loss, environment/noise, trace loss, process gain etc. Therefore, if you want to evaluate your design performance, you should compare the SNR of your board with a known proven design for example TI EVM, and then performing the measurements using the same setup.
You may want to consider using a corner reflector with a known RCS at a fixed distance from the Antenna/Board. Please use same appimage/FW, same chirp/profile configuration on both the EVM as well as on your design for data capturing. Use same code for processing the data captured from EVM as well as from your PCB to eliminate the gain difference induced from processing. Non-zero doppler bins in 2D-FFT will give you noise floor and near zero-doppler will give you the signal power. You can calculate the SNR at the range bin where the corner reflector falls. Difference in the SNR between EVM and your PCB should account for antenna gain and trace loss difference. The XWRL1432BOOST has an antenna gain of approximately 5-6 dBi across different antenna pairs.

Thanks and Regards,
Sivaprasad

Hi Sivaprasad

Thanks for your reply. In TI site, how do you check the EVM performance if you have no competitor?

Or how do you evaluate the performance before you got the EVM?

Best regards.

Jasper