Ghost detections in VL53L7CX ToF sensor zones after crosstalk calibration

[benpbabu]

    I have performed crosstalk calibration of VL53L7CX ToF sensor using a 0.2 mm thick polycarbonate cover glass. I had used a sheet of white paper (assuming 80% reflections) placed at 600m as reference surface & 4 samples were used for calibration.

    The calibrated sensor's accuracy was tested using a similar setup but with reference surface at 400 mm. The distance values observed in the 64 zones ranged from 396 to 426 mm. The standard deviation of the distance values were in the range of 2.2 to 4.5 mm. The heatmap of the zones also looked uniform.

    But when I mounted the sensor in an enclosed space (around 1m X 1m X 1.5m) upside down pointed towards my objects of interest, I'm observing some random ghost detection in some of the zones. There are no objects in these zones at distances given by the ghost detections. But there could be objects/surfaces in nearby zones. Also, these ghost detections were not present when I did the same experiment in the same enclosed space & target objects without the cover glass (before performing crosstalk calibration).

What could be the reason for these ghost detections? What can be done to avoid them?

Thanks in advance.

[AlexCloned]

The primary goal of crosstalk calibration is to subtract the predictable parasitic light signal that bounces internally off the cover glass and returns to the SPAD array. When ghost detections appear only after calibration and only when the sensor is mounted in the enclosed space, it suggests a new parasitic reflection source has been introduced.

To properly diagnose the issue and guide you toward a definitive solution, it would be extremely helpful to know the answers to the following three diagnostic questions:

1. Do these ghost detections occur prior to performing the crosstalk calibration?

2. Do the ghost detections occur if you completely remove the 0.2 mm polycarbonate cover glass?

3. Have you tried performing the crosstalk calibration in the actual enclosed environment instead of laboratory conditions?

   • When you call the crosstalk calibration function, you can pass in the real reference distance and experiment with different values for the estimated reflectivity of your reference target. Calibrating in the final setup will force the routine to account for the actual environmental noise and surrounding reflections.

4. You are checking status code returned by those ghost detections. True?

It's vital to remember that we are dealing with a massive amount of photons —often several orders of magnitude greater than the small number of useful photons that are actually detected in the SPAD and accumulated in the histogram bins of each zone.

Histograms, crosstalk calibration and an intelligent setup are often the most effective tools that let us obtain the best workaround method to avoid undesired reflections.

Please share your results or conclusions.

[John_Kvam]

I think you missed something. When trying to do calibration, as AlexCloned stated, you are looking for those photons reflected by the coverglass. 

To that end, you want to eliminate everything else. Ideally there would be no targets within a mile. And no sunlight.

But clearly, that's not practical. 

So the user manual has you placing a VERY BLACK target at 600mm. (not your 80% white.) 

Use a matte finish black spray paint - most of them are about 5% reflective. And place your target as far back as practical. 600mm is just a minimum. But remember, as the distance increases, the target size has to as well, so you need to cover the entire field of view.  When I do my calibration, I point the sensor at the office ceiling. It's white, but it's also 1.6 meters away. Seems to work just fine for me. 

By reducing the number of unwanted photons, you will get a better calibration. 

- john

[benpbabu]

1. No
2. No
3. No, that would be difficulst due to the angle of mounting the sensor & other objects in the region of interest
4. No, I wll do that

[benpbabu]

I shall try the black target. I'll also try a pitch black environment if possible

[Zhiyuan.Han]

Hi

Sounds like your issue is sharpener related, you can try to play with the sharpener to see if help. 

Br

Zhiyuan.Han