Observation and queries for VL53L0X sensor for 97000 sample captured at @ 1 sec periodicity over 27 hours period

[YGude.1]

We are working on syringe volume sensing application with VL53L0X 

Our expected accuracy is +/- 2mm

We configured sensor in high accuracy mode with 900msec budget time

To check stability performance of sensor over long period of time we captured 97000 sample at @ 1 sec periodicity over 27 hours period with object at fixed distance.

Overall we found the sensor output is stable varying between 3 count for fixed object distance.

Attached is the 97000 readings which we captured and our observation and queries.

Please help us to provide answer to our queries.

.

[John E KVAM · Accepted solution]

i) Our understanding there are chances, temperature drift during 2 am to 8 am, which cause readings to drift to 103. Is our understanding correct?

Well temperature does affect the sensor. One should stop the sensor every now and do a VHV calibration.

ii) Is it possible to overcome drift in reading by doing some run time temperature calibration?

Yes, although it will still vary a bit.

Ref calibration is the calibration of two parameters (VHV and phase cal) which are temperature dependent.

iii) If you have similar data captured for VL53L0X for long period for constant temperature condition, then please share with us.

That use-case is pretty rare. I don't know of any such data.

iv) As shown below calibrated sensor data used to observed as 103 for only 2 to 3 sample in between for 3 to 4 raw reading sample of 146. 

We applied Exponentially weighted moving average (EWMA). It absorb one overshoot reading of 146

Basis on your reply for above queries, we might need to work on other filter which will absorb 3 to 4 raw reading

 Please suggest any suitable filter basis on your experience, which will absorb 3 to 4 raw reading 

Oddly, folks who find a solution to stuff like this don't generally share it.

I expect you moving average idea is the best there is.

But every now and again, stop the sensor, so the REF calibration and then restart it.

(Or, you might consider NOT doing the ref cal, but add a temperature probe and do your own temperature compensation. You might do better than we do.)