Multizone sensor calibration buffer

[_jh]

Dear forum,

I try to understand the bin shape of the VL53L8CX / CH sensor calibration buffer while evaluating different sensor cover and their influence on ranging. Using the example8 file from the ULD driver the shape looks more or less always the same (as eg. shown by @John E KVAM ) for most cover glasses I used. Nevertheless doing the math with a bin size of 3.75 cm the light path must be in range of ~40 cm, which seems to me highly unlikely with a 1 mm cover glass placed in 0.5 mm distance. Moreover I realized that the calibration buffer yields 144 bins whereas I read somewhere here that there are only 128 bins for ranging.

Perhaps someone can elaborate on that a little bit.

Thanks!

[John E KVAM · Accepted solution]

You are correct in a lot of this. 

1) there are 144 bins, but only 128 of them are very useful. The "Normalized" part of "Compressed, Normalized Histograms" means we dealt with the ones that are before the light turns on. 

2) each bin is 3.75cm. But the Flash of light is 2ns. And that's way more than a bin or two. It's a lot closer to 11 or 12. And the laser still emits after the power is shut off. Although at a decreasing amount. This can lead to a 'tail'.

During the Normalization process we cleaned up the signal somewhat, so you didn't have to deal with it.

We remove the crosstalk for instance. So what you get is all signal.

You can try this yourself. Slap a coverglass on the sensor without doing any crosstalk, and range into open air. 

All the bin data you see will be the crosstalk. (note that the bins in the middle of the first column are the worst.)

Then calibrate it and run again. Compare the data. 

Then try the same experiment with some sunlight. You should get about the same numbers. We removed the sunlight for you.

- john

[_jh]

Dear John,

thanks for the fast and insightful reply.

---

@John E KVAM wrote:

1) there are 144 bins, but only 128 of them are very useful. The "Normalized" part of "Compressed, Normalized Histograms" means we dealt with the ones that are before the light turns on. 

Putting it in simpler words you are starting the "internal" clock ealier than the laser pulse and correct the range calculation later by some offset? I expected the xtalk peak in the bins 0-12 when starting the clock with the laser pulse?

---

All the bin data you see will be the crosstalk. (note that the bins in the middle of the first column are the worst.)

---

I also observed this. Interestingly the backscattered xtalk light is focused to this zone when the cover glass thickness increases. Thus was hoping if one was able to calculate the dominating light path from xtalk histogram data to find an reasonable explantion for this effect. But due to the offset and the limited resolution this seems to be impossible. Is there an explanation for the "focussing" from ST?

[John E KVAM]

It's just physics. The light goes out in a pyramid of 45 degrees. With the receive opening on the right, you can see why the light going off on the left has no adverse effect. And the light going straight does not contribute anything either. But the light on the right side is going to hit the underside of the glass and bounce back. Some light will penetrate the glass and bounce off the top of the glass. (Some will also make multiple bounces, but thinking about that makes my head hurt.)

We put a 'light trap' between the holes to stop of lot of the bounces. But the direct path of the light will cause one side of the photon detector (SPAD) array to be more affected than the other side. 

But if you do a good job of calibration, your CNH data will not have any crosstalk in it. 

- john

[Christian N]

This post has been escalated to the ST Online Support Team for additional assistance. We'll contact you directly.

[_jh]

Hi John, thanks for the additional comment.

Best, _jh