VL6180X Detecting Glass and Low Light Range Difference

[Michael Steege]

Posted on April 06, 2017 at 03:27

Two questions.  I'm implementing the VL6180X sensor in a system that needs to detect presence of objects placed in a tray.   The objects are generally small glass vials with different colored caps.  

Question 1:  In ambient office lighting, the objects are detected at expected distances if the colored caps face the detector.  If the glass bottom of the vial faces the detector, it appears as if the detected distance is the cap on the other side of the vial.  It does not seem to detect the glass bottom of the vial.   What would cause that and what is the solution to detect the glass bottom (note: the vial may be as close as 15mm and as much as 70mm from the

sensor- there is no cover glass on the sensor).

Question 2: After the object is placed and the distance detected, a cover door is closed which puts the object in the dark.   But in this low light condition, the detected range increases by as much as 30%.   (i.e. in low light at a distance of 45mm, the sensor returns a distance of about 60mm)  What causes the difference in range?   It seems that it would be highly accurate in low light conditions.  (note:  difference in ambient light vs low light was tested with a solid white object placed at a precise distance from the sensor - but it does the same with the glass vials).   (also note: I can detect the light levels and apply different range compensations if necessary)

The application needs to detect the object in both ambient and in low light conditions at a distance between 10mm and 90mm.

Any help, advice, or direction to application notes explaining this

phenomenon

 is appreciated.   Also, it appears I cannot use the VL6180X SW API as I am am using a soc with limited ROM and including the api as is

exceeds my ROM space when built. 

Thanks in advance,

Michael

[John E KVAM]

Posted on April 07, 2017 at 18:18

Question 1

The VL6180 works by emitting light at 850nm and detecting the returning photons. If you say that the sensor is only seeing the cap, then one can assume the glass you are using is transparent at 850nm. One solution would be to find some glass that is not transparent at that wavelength, or to range on the contents and not the glass. One might also try the VL53L0X sensor, which works at 940nm and might detect your glass. I would try a P-NUCLEO-53LV for a few dollars, and give it a quick try before investing more effort. The VL53 API requires a more memory than does the VL6180 however.

Question 2

If the ambient light contains some 850nm it will cause the SNR of the system to go up. Placing the sensor in the dark creates a noise-free environment. This is a best case situation for the sensor. Yet you seem to suggest that you get the wrong answer in the low-light case. 

Could it be that there is some other surface in the field of view? The senor 'sees' objects in a 25 degree cone, and averages all the returning photon times to get an answer. If the interior of your cabinet is within the range of the sensor it will effect the result. One might consider painting the interior of the cabinet a dark black to reduce this effect.

As a side note, LED lighting contains almost no component at 850 nm, whereas sunlight, florescent, and incandescent do. If you can control your lighting situation, you might want to try your experiment under LED lighting.

[Michael Steege]

Posted on April 07, 2017 at 19:02

thank you -   we have come to the same conclusion about question 1 regarding transference through glass and are investigating more of the operating space of the sensor within our closed environment.  It may be that the sensor is receiving a number of reflected photons from the cover creating a perceived time delay resulting in an increase in distance calculation.  We are aware of the VL53L0X and may look into that option.

[Michael Steege]

Posted on April 26, 2017 at 19:47

Here is what we found out.  As mentioned above, every surface within the field of view (FOV) returns photons to the detector and are averaged into the final range calculation.  So the sides of a tray holding the glass vials, the bottom of an access door cover, the glass of the vial, the liquid of the vial and the cap are all effecting the final range value.  Painting the bottom of the access door black effectively eliminated the ranging difference between ambient light and low light conditions. Painting the container holding the vials black resulted in more accurate detection of the glass vials.  It isn't that the glass can't be detected, it is that the objects around the glass were providing the bulk of the return resulting in the inaccurate ranging. So the best application for this sensor is when the reflectivity of the object to be ranged is much greater than that of the surrounding environment.