IWR6843AOPEVM: Range profile data from log2 to linear values

Vinay Kulkarni

Prodigy 140 points

Part Number: IWR6843AOPEVM
Other Parts Discussed in Thread: IWR6843

Hi,

From Documentation the range profile data is represented by log2 magnitude with Q7.9 format.

So i tried converting this to linear by : x=2^(y/512) , where y is range profile in log2 form.

Then to match the TI demo visualizer, i performed 20*log10(x) = 20*y log10(2)/512 .

But this seemed to mismatch with Demo visualizer.

is my understanding correct or am i missing something ?

over 1 year ago

0 Nathan Block over 1 year ago

TI__Mastermind 27680 points

Hi Vinay,

If your output matches with the demo visualizer then I would assume it to be correct. See similar posts for more information about how the visualizer decodes the range/magnitude plot.

Best,

Nate

0 Nathan Block over 1 year ago in reply to Nathan Block

TI__Mastermind 27680 points

e2e.ti.com/.../iwr6843aopevm-structure-of-range-profile-data-in-out-of-box-demo

0 Vinay Kulkarni over 1 year ago in reply to Nathan Block

Prodigy 140 points

Thanks Nathan.

Its mismatching, hence needed proper interpretation.

The link that you shared is not translating to linear form. Can you please check .

0 Nathan Block over 1 year ago in reply to Vinay Kulkarni

TI__Mastermind 27680 points

Hi Vinay,

Do you want to convert this to linear or match the demo visualizer? The demo visualizer displays these values in dB format.

Best,

Nate

0 Vinay Kulkarni over 1 year ago in reply to Nathan Block

Prodigy 140 points

Both please .

First want to convert it to linear for my internal processing and then just to compare , i need relative power in dB as equivalent of demo visualizer.

Can you please check my understanding of formula is right, or i have to tweak ?

I think if i am chosing 3 tx and 4Rx, i may have to incorporate 12 virtual antennas, which i am not clear how to do .

Would be great if you can share Linear conversion of Range profile and then dB conversion to match demo visualizer.

0 Nathan Block over 1 year ago in reply to Vinay Kulkarni

TI__Mastermind 27680 points

Hi Vinay,

I am researching this, and I will get back to you by the end of the day on 3/28 Dallas time.

Best,

Nate

0 Nathan Block over 1 year ago in reply to Nathan Block

TI__Mastermind 27680 points

Hi Vinay,

To convert from the output to linear, you should divide by 512 to get rid of the Q9 format, then you can convert from log2 format to dB by multiplying the data by20*log10(2), then convert from dB to linear by doing 10^(x/10) on the data.

Linear format = 10^((output /512 * 20log10(2))/10)

We do not show the output in linear format, rather, we show it in dB. This formula is used (which contains similar steps as the equation above).

https://e2e.ti.com/support/sensors-group/sensors/f/sensors-forum/821420/iwr6843isk-ods-how-to-calcurate-relative-power/3106330#3106330

0 Vinay Kulkarni over 1 year ago in reply to Nathan Block

Prodigy 140 points

Thanks Nathan.

But it does not seem to match.

The formula you showed and in other e2e forum too, we have this formula

So we are missing numRangebins and NumVirtAnt . How is this really calculated ? Looks unnecessarily complicated.

Can you please check if you can equivalently show log2-Q9 format of Range profile with demo visualizer Range profile figure.

I am unable to get this equivalence.

0 Nathan Block over 1 year ago in reply to Vinay Kulkarni

TI__Mastermind 27680 points

Hi Vinay,

20log10(numRangeBins) is to normalize the graph so when there are many range bins and few range bins the results are similar.

The 2^ceil(log2(numVirtAnt))/numVirtAnt resolves to 16/12 when you use 12 antennas, and 8/8 when you use 8 antennas. It is to account for the fact that although there are 12 virtual antennas on the IWR6843, when we sum the zero doppler bin together to get the range profile, we are taking a 16 point FFT. So, the value is scaled by a factor of 16/12 to account for this.

I gave you the above formula because you asked how we would convert this to linear units, which is not what the visualizer does. I would also like to ask here why it is important to you to get the linear magnitude of this value out. If you want to match the demo visualizer, you can use the above formula. The results won't have much significance in a linear format because they're power numbers, which are typically used in dB.

Best,

Nate

Sensors

Sensors forum

IWR6843AOPEVM: Range profile data from log2 to linear values