IWR6843AOPEVM: Status of MMWAVE_DFP- 867

Duplicate question. Closing the thread.

Best,

Nate

This is for a different bug, so I disagree that it's duplicate.

I'm sorry Morten! Working too fast this morning :).

Morten Nissov said:

• Is this present in the out of box demo for IWR6843AOP?

Yes DFP should be the same for both

Morten Nissov said:

The release notes mention that "slope error of 1-bit can cause errors in range estimation and also affect RF bandwidth"

• What order of magnitude impact does this have on range estimation

Very small, 1 LSB is at most an error of 1/256 for an 8-byte int, so I would expect a similar order of magnitude in range estimation.

Morten Nissov said:

My understanding is this can be avoided by selecting parameters which are even. Can this be avoided only by changing parameters in the .cfg file, or do things need to be changed in the implementation itself.

I would confirm in the implementation to be sure. You can find the API calls in the demo and see if they correspond 1-to-1 with the config file parameters or if there are unit conversions that take place.

Best,
Nate

Similar order of magnitude meaning ~4mm (1/256)?

Hello, 

Nathan is currently out of office and will respond when he returns next week.

Best Regards,

Josh

Hi Morten,

If your maximum distance based off the frequency slope is 4mm * 256 = 1.024 m then yes, but I suspect your maximum slope is higher.

Best,

Nate

Hi,

Sorry but I don't really understand the example. Maybe we can try some more realistic numbers. Say the slope is 35 MHz/us for a chirp time of 55us. The value of the LSB is decimal 1.

This bug is then saying that the slope may be rounded even, therefore 36MH/us instead. This means my range resolution changes from c / (2 * 35e6 * 55)==0.0779 to c / (2 * 35e6 * 55)==0.0757

This configuration has a 2mm resolution error. For a range FFT w/ 256 bins, this is the difference between 19.38m and 19.93m max range, otherwise ~0.5m.

Is that correctly understood?

Morten