AWR2243: Self calibrations and monitoring timing/programming

Hello Federico,

As you are probably aware, the various monitors of the device are described in section 10 of the datasheet AWR2243 Single-Chip 76- to 81-GHz FMCW Transceiver datasheet (Rev. C) (ti.com) 

Additionally, as described in the following document it is recommended to set  CALIBRATION_PERIODCITY so that the calibration occurs at least once per second.  It is not recommended to set N so high that the calibration occurs less than once per second.  

mmWave-Radar-Interface-Control.pdf

I am following up with my team about which monitors may be always available.  However, in the mean time could you let me know what monitors you are hoping to use?

Regards,

James

Thanks for the quick response! Well I saw the document and yes, I saw these monitors, however I found also another table that I include here below, and it's not clear if these are exactly the same monitors included in the document you specified. 

To give just an indication, monitors we are interested to read in runtime, during the chip is chirping/transmitting are:
- temperature

- RX saturation

- PMCLKLO internal analog signals and 20G Sync

Instead of course it's clear that other monitoring will be performed every x minutes/hours if possible. This is just because from the table it's clear that the reading of some monitors takes long time, longer than the frame itself that we planned. In this conditions I'd have a long interruption of the uniformly programmed sampling of loops/frames and it's a problem for our application.

Concerning calibration, up to now we made calculations including only mandatory calibrations (APLL and synth VCO and some few others), so that after each frame basically we make perform self-calibration with a total time equal to a frame itself, so that we will have a quasi uniform sampled chirps (uniform chirps with uniform holes/missing data). You suggested to perform self-calibration once per second, avoiding the N to be too high. But, exactly I think there is a limit for these self-calibrations. Do you know which is this limit? I mean, the number N of frames between calibrations has a limit? 

Thank you,

best regards,

Federico

Hi Federico,

To go back to an earlier question - after checking with my team I have determined that you should ensure you have enough inter-frame time for your programmed monitors/calibrations.  i.e. these should occur between frames, and you should plan accordingly.  

Regarding your latest question, I will check on this, but could I ask if you are trying to greatly increase the time beyond one second?  This isn't generally recommended.  

Regards,

James

Hello James,

yes I know that I need to calibrate and eventually read monitors during inter-frame. To this aim, since I need in my application that the radar data is acquired uniformly and with fast chirps, we thought that basically we perform just APLL and Synth calibration in inter-frame and then perform other calibrations and monitor reading with a higher time scale (e.g. every 60 minutes). Can I know which calibrations, except those already mentioned, are very important to be performed more frequently of what I was thinking?

Best regards,

Federico

Hello Federico,

The answer here will likely largely depend on the intended operating conditions, especially how often you think temperature may change, and the likely temperature gradient you need to support when temperature does change.  

Generally, these are the most important to be done frequently:

1. APLL calibration, to ensure accurate frequency especially over temp changes

2. VCO calibration, to ensure VCO voltages (and also hence frequencies) are maintained.

3. LO distribution, to ensure that the distributed LO signal is maintained across temperature 

4. Peak detector calibration to ensure accurate BIST and safety

5. TX power calibration especially if you anticipate large temperature variation

6. RX gain calibration especially if you anticpate large temperature variation

More change in the environment during operation generally means you should plan for more frequent calibrations.

Hello James,

thanks for your reply. I have new questions now: is it possible to decide, from inter_frame to inter_frame which calibrations/monitors do? I mean for example:

- Frame 1 --> interframe --> (APLL + VCO) (always on)  + LO

- Frame 2 --> interframe --> (APLL + VCO) + Peak detector

- Frame 3 --> interframe --> (APLL + VCO) + TX power calibration

- Frame 4 --> interframe --> (APLL + VCO) + RX gain calibration

- Frame 5 --> interframe --> (APLL + VCO) + Temperature Monitor

and so on.

Consider that, for what we're going to program, each frame has a duration of about 49 ms, and interframes about 3 ms