IWR1443: output width and height of CSI frame

Chris Meng

Part Number: IWR1443

Hi,

I would like to understand the width and height of CSI frame when we enable the CSI output in capture demo in mmwave sdk 1.0 for IWR1443. Below is my understanding. Would you pls kindly help to check if it is correct or not?

CSI frame width : samples per chirp*number of RX*number of TX*2 (if complex) or 1 (if real))

CSI frame height: number of chirps per frame

over 7 years ago

0 Randy Rosales over 7 years ago

TI__Genius 17770 points

Hi Chris,

Have you checked out the IWR1443 Technical Reference Manual? I think Section 13.4.1.1.2 "Low Level Protocol" in the TRM is what you are after:
www.ti.com/.../swru522a.pdf

Please let us know if this section answers your specific question.

Thank you,
Randy

0 Vaibhav Mahimkar over 7 years ago

TI__Mastermind 23625 points

Hi Chris,

I am assuming you have already looked at the AWR1xxx Data Path Programmer’s Guide for data format details. It has a CSI specific section in it.

Please look at Figure 25 for placement on Horizontal Start and End. (HS and HE) this should indicate what each horizontal line constitutes of.

We will try to get more exact answer for your question soon.

Thank you,

Vaibhav

0 Naveen Narayanan over 7 years ago in reply to Vaibhav Mahimkar

TI__Expert 5025 points

Hi Chris,

The correct computation would be:

CSI frame width (in bits) : samples per chirp*number of RX*2 (if complex) or 1 (if real))*number of bits per sample

The number of Tx chosen doesn't affect the Rx path and amount of data sent out. Also on the CSI2 lanes, as per the MIPI spec, the data is sent as 8 bit (1 byte) data distributed on the lanes. Hence, the long packet header data will contain the CSI frame width in terms of bytes of transfer.

CSI frame height: number of chirps per frame (as you noted)

Best regards,

Naveen

0 Chris Meng over 7 years ago in reply to Naveen Narayanan

TI__Mastermind 36035 points

Naveen,

Below is copied from file:///C:/ti/mmwave_sdk_01_00_00_05/packages/ti/demo/xwr14xx/mmw/docs/doxygen/html/index.html. As you can see the TX will also increase the ADC data based on the chirp/TX configuration. So I think the width of CSI data is also related to the number of TX used. Would you pls help to check again?

IWR1443 CSI supports 8-bit/12-bit RAW output. No matter how CSI packaged the data, the external host will get CSI data with 8-bit/12-bit as one pixel. So for CSI width to external CSI host, I think it need not multiply by bit per sample.

0 Vivek Dham over 7 years ago in reply to Chris Meng

TI__Guru 58035 points

Hello Chris,
If you use the MIMO configuration (time division multiplexing with different TX) then the number of chirps will increase based on the number of TX used. So the important part is total number of chirps, you need to take into account the total number of chirps sent , including chirps with different profiles etc. If you use two TX in the same chirp then there is no additional data width due to multiple TX.

regards,
Vivek

0 Naveen Narayanan over 7 years ago in reply to Vivek Dham

TI__Expert 5025 points

Hello Chris,

Hope Vivek's explanation answers your first query.

Regarding your second comment, even in RAW12 mode of CSI2, data is sent as bytes (8 bit) over the lanes and hence I just multipled bits per sample only to bring out this point. It just makes it easier to compute the long packet header length (divide by 8).
And yes, your point is correct from the receiver (external host) application point of view. This will be abstracted out by CSI2 receiver and the multiplication by bit per sample is not needed.

Best regards,
Naveen

0 Chris Meng over 7 years ago in reply to Naveen Narayanan

TI__Mastermind 36035 points

Naveen and Vivek,

It is clear to me now. Thanks all for your help! I summarized as below.

The CSI frame for external CSI host:
Width: samples per chirp*number of RX*2 (if complex) or 1 (if real)
/if using the MIMO configuration (time division multiplexing with different TX: samples per chirp*number of RX*number of TX*2 (if complex) or 1 (if real)
Height: number of chirps per frame

0 Vivek Dham over 7 years ago in reply to Chris Meng

TI__Guru 58035 points

Hello Chris,
In case of MIMO the width remains the same as non MIMO case , the height increases by number of TX since the number of chirps increase. The size of each chirp still remains the same, so the width remains the same.

Regards,
Vivek

0 Chris Meng over 7 years ago in reply to Vivek Dham

TI__Mastermind 36035 points

Vivek,

Thanks for your input.
But if we set interleaved mode, I think the width should be increased by number of TX.
For non-interleave mode, I am not sure how our code will treat increased ADC data from different TX in CSI frame. Increase width or increase height should be both ok, but it should depend on the code. I tried to find useful info in the csi/cbuff_csi related code, but still can't find such info. Would you pls help me to find out how our code to set one frame of ADC data to a CSI frame ( with width and height)?

0 Vivek Dham over 7 years ago in reply to Chris Meng

TI__Guru 58035 points

Hello Chris,
All ADC data is (Interleaved or non interleaved) is with respect to RX chain ONLY , the TX used does not matter from the perspective of sample size. You should not look at it as number of TX but just number of chirps. One chirp data comes as width and multiple chirps for the height.

Regards,
Vivek

0 Chris Meng over 7 years ago in reply to Vivek Dham

TI__Mastermind 36035 points

Vivek,

Thanks for your patient help. Now I am clear now. Number of chirps per frame including the all chirps from different TXs if using MIMO config (time division multiplexing with different TX).

Sensors

Sensors forum

IWR1443: output width and height of CSI frame