Help with insight on VL53L4CD drift

[AGreen]

I'm using the VL53LX API 1.2.8 to interface with the VL53L4CD in firmware and I implemented a tuner page in my android app to visualize it.

So far I have this thing ranging pretty well with +/- 1mm accuracy. Not bad!

I'm measuring a permanent grey 3D printed PLA object with a 10x5mm flat face that can move 10-45mm in front of the sensor.

My only issue is that when I start/stop ranging, it starts off on the correct distance (object not moving) and then after ~1min it starts drifting up +2-3mm. However, it tends to cap out there. So, if I'm reading 45 it will climb up to ~48 and settle. If I start/stop again it will repeat.

Is this due to temperature drift or something else? Is there anything to lessen this? I would think that if it was temperature then start/stopping would not cause it to jump back to reading 45mm right off the bat but instead leave off at the last drift value.

Project details:

Electrical:

-I2C at 100kHz

-VCC 2.8V

Typical Operation:

-Follows the UM2923 - VL53L4CX API ranging flow (field)

-Host polling, checking for Data Ready (~10ms intervals)

Boot up:

-Toggle XSHUT with delays between

-Wait Device Boot

-DataInit

-Set power level to 30 from 60

-Set Phase Cal Patch set to 2

-Set Offset Mode to PERVCSEL

-Set ROI to 4x4 TopLeftX:6, TopLeftY:9, BottomRightX:9, BottomRrightY:6

-Set Calibration Data

-Set distance mode to Medium (all though this is default)

Calibration:

-Ref SPAD

-XTALK (not really required since no glass but seem to help)

-VL53LX_PerformOffsetPerVcselCalibration(45mm)

-Store Cal data

Thanks,

Andrew

[John E KVAM]

Andrew -

The VCSEL does generate heat - and it's effency is affected by that heat. As the temperature rises,there is a VHV (Very High Voltage) calibration that goes on. It will adjust. But it does not have a very fine resolution. So if you warmed the sensor you might see a drift of a few mm, then a snap back as the temp compensation kicks in, then slow drift leaving you with a sort of saw-tooth pattern.

I suppose if you wanted to get the best accuracy, you might find some way to hold the chip at a known temperature.

In fact, if you had a handy way to do this is might be an interesting experiment.

Nicely done on your app. Looks pretty cool.

[AGreen]

Thanks, I've been working on this thing for a year now.

Unfortunately, I have no control over the sensor temperature in this project.

Is the VHV doing the same thing as the VL53L4CD_StartTemperatureUpdate() function in the low power driver or how do those differ if not?

Also what triggers the VHV to start a calibration and what's the threshold?