IWR1443BOOST: slope noise floor

Hello,
Using the continuous Tx mode, I suggest you setup 4 or 6 single Tx examples, and look for spurs in the Tx. There is an example of the Continuous Tx mode discussed in the mmwave SDK User Guide. If you substituted a 40Mhz, capacitor coupled 1v sine input for CLKP no load cap, and placed a zero ohm resistor in place of the CLKM load cap, the mmwave sensor will use your external clock source.

I would also check the 1v3 and 1v8 LDO power supply outputs with an active probe, to make sure the LDO noise output is within the Datasheet table limits.

Please send a picture.

Do you see the same problem on 2 boards?

Regards,
Joe

Hello Feng,
Please make you picture a PDF or JPG, I can't see it. You could also send the IQ data from the Radar cube memory.

Trying to simplify the test, take 2 boards, and Continuous Tx, transmit on board a at 78Ghz. Check each board's output with a reference
horn antenna, RF down converter, and spectrum analyzer. We are looking for spurs in the Tx output.

If both boards have better than 40db SNR on Tx, set one board to Continuous Tx with an output of 78Ghz. Make sure that the first board
is transmitting (you can measure with horn antenna above). On the second board, in the chirp configuration set the Tx to 0 (not enabled) with a profile of 77.999Ghz, Frequency slope of .001Mhz/us. Check the 2nd board Rx, you should see a chirp result that varies from 1Mhz to 1Mhz - (slope * numADCsamples_chirp / DFE output rate). You can adjust the RF coupling between the 2 boards, and check each Rx that is connected.

Note: for the second test, you could substitute a Frequency source, Upconverter, and Horn antenna as a Tx input.

Please make sure that unused Rx are connected to GND.

Please check the 1.8v, 1.4v, 1.3v, and Vout_PA 1.0v, you need to meet the ripple and noise of the supplies to the mmwave sensor .

Regards,
Joe Quintal

Hello Feng,
It is important to use an active probe, or use a 40Msps or higher AC coupled oscilloscope measurement, we need to make sure the power supply noise is not coupling into the Radar Rx.

You can tell me what you setup is, so this can be measured, or you can measure this comparing to the EVM.
The data capture demo, in later mmwave SDK under the drivers, you can capture the radar cube memory.
Regards,
Joe Quintal

Hello
The oscilloscope measurements are for both the 40Mhz crystal, which you have done, and to measure the ripple and noise 1.3v, 1.8v in the datasheet.
You should also measure the noise and ripple of the 1.4v outputs, and the RF_PA 1.0v output.
You can compare your measurements to the EVM board you have.

Tx1 is 0x1, Rx4 is 0x8 - these are bit fields. Please see section 3.4 in the mmwave SDK User Guide.

Unused Rx inputs should be tied to GND.

Regards,
Joe Quintal

Hello Feng,
The slope is indicating you get an Rx reflection close to the Tx and Rx , the frame-configuration-chirp is not correct, the radar power supply is noisy (AC ripple and noise comparison measurements to EVM), or possibly that the BSS and MSS software needed should use the latest version of the mmwave SDK compatible with your device. Current released part for IWr1443 ES2, mmwave_sdk_01_02_00_05.
Note: lower power - level sensing TIDEP-0091 will be re-released this quarter.
Regards,
Joe Quintal